An Investigation into Goal Scoring Patterns in Association Football
Ronald Arthur Smith
A thesis submitted to the University of Canberra for the degree of
Doctor of Philosophy
July 2016
Faculty of Health
University of Canberra GOAL SCORING PATTERNS i
Abstract
This thesis investigates goal scoring in professional association football. There has been a vibrant debate in the research literature about how goals are scored. Researchers have discussed the location of the scorer on the field of play, the number of touches of the ball taken, the type of pass, the number of passes in the sequence preceding each goal, and when in a game goals are scored. There has been a growing interest in identifying the most successful area of the field where the final pass leading to a goal was made and has led to debate about one area in particular, Zone 14. The quantification of the number of passes preceding goals has fuelled debate about the tactical success of ‘possession based’ football and ‘Direct Play’. Approximately 90% of goals in association football are scored within 23 yards of the goal and the majority of these with less than five passes.
This research presented here analyzed goal scoring in Open Play to determine if the most successful method of gaining entry into the scoring area was from ‘Passing the ball behind opponents or to a player level with the last defender’, compared with ‘Crossing’ the ball and any ‘Other Methods’ that were not included in the other two categories. This new approach maps 7 areas of the field, rather than the 18 used in the extant literature, to record where the final pass was made in each category. It is argued that the use of 7 areas sensitive to the offside law yields a much better analysis of performance. Data were recorded about in which ‘third’ of the field possession was regained and the number of passes in each sequence.
The thesis presents new operational definitions for the quantification of lost possession. It is argued that these definitions provide a more accurate account of events preceding goals specifically in relation to what the literature has regarded as ‘zero’ pass goals. Data for this study were gathered from three seasons of the English Premier League and the Australian ‘A’
League and three tournaments of FIFA World Cups and UEFA European Championships. A total of 3,175 goals in Open Play were analyzed. These data enabled comparisons to be made GOAL SCORING PATTERNS ii
within and between league football and international tournaments. Goals were captured and coded with Sportscode Elite software V.9. Data were analyzed with SPSS software V.19.
The results presented here report that the most successful method of scoring in all international tournaments and in 4 of the 6 league competitions was from ‘Passing the ball behind opponents’; the vast majority coming from an area identified as Zone 14+, the area between the half way line and the penalty area. The majority of goals were scored with 5 passes or less and from regained possessions in the middle third of the field in every competition. The least successful category for scoring in 11 of the 12 competitions was from
‘Crosses’. The evidence from this research provides coaches with the most effective of three strategies to score goals in professional association football while leaving them to decide how best to implement these strategies with the players at their disposal. GOAL SCORING PATTERNS v
Acknowledgements
Throughout the past five years I have had assistance, encouragement and advice from a number of the University staff members and I would like to take this opportunity to acknowledge and thank them.
The thought of writing a thesis was a daunting task and I can confirm is a massive challenge. I attended Dr. Linda Li’s writing classes and learnt how to structure my thesis and use the chapters to explain everything about my research. It was an invaluable part of my learning process.
Maryanne Simpson, the Research Liaison Officer in the Faculty of Health was always available to solve any administrative problems I had and gave me advice when I needed it.
Dr Joëlle Vandermensbrugghe, Researcher Development Program Manager, for her assistance and weekly emails to advise me of workshops to help me through the process of researching and writing.
Julio Romero for his patience, tuition and expert advice to help through the statistical analysis of my data, I really appreciate the time you gave me.
Professor Roland Goecke, my second supervisor, for his support and advice throughout the years of my candidature. To my primary supervisor, Professor Keith Lyons, I am extremely grateful for your encouragement and guidance, and for sharing your experience and expertise in performance analysis; I have been extremely fortunate.
Murray Shaw, Executive Producer of Football at Fox Sports, for giving me access to the match files of the EPL that were missing from my own collection and to Andy Hoad,
Match Day Saturday producer at Fox Sports for his time and tuition to edit footage in the studio. GOAL SCORING PATTERNS vi
Ray Junna, former Football Coach at the Australian Institute of Sport for his time and willingness to assist me with reliability tests.
There are several people who influenced my thinking about the ‘game’, my coaching philosophy and my attitude towards research. Tommy Tranter, my coach and lecturer at
Borough Road College, who made me realize there was a good reason why everything happened in a match; he opened my eyes to ‘coaching’. Eric Worthington, former Football
Federation of Australia Director of Coaching, who helped me develop as a coach educator and Ron Tindall, former professional player, coach and manager who inspired me to think deeply about tactics and player behavior. Sadly, they are no longer with us but I regarded each of them as a friend and mentor.
Finally, I wish to thank my immediate family; Luke and Barney for their continued support, interest and encouragement to complete what I started years ago. To my darling wife
Alison, who has sacrificed so much in her life to support me as a professional coach, I could not have done this without you I am eternally grateful and love you.
GOAL SCORING PATTERNS vii
Table of Contents
Abstract ...... i Certificate of Authorship of Thesis ...... iii Acknowledgements ...... v List of Tables ...... xiii List of Figures ...... xvii Chapter 1 Introduction ...... 1 Overview ...... 1 1.2 Background of the study ...... 1 1.3 Technology and topics associated with scoring goals in football ...... 4 1.4 Justification for a different approach to analyze goal scoring ...... 6 1.5 A new approach to analyze goal scoring patterns in football ...... 10 1.6 The aim and objectives of the thesis ...... 12 1.7 Research Questions ...... 13 1.8 The scope of the research ...... 15 1.9 Synopsis of the Thesis ...... 15 Chapter 2 Literature Review ...... 19 2.1 Introduction ...... 19 2.2 When goals are scored ...... 21 2.3 How goals are scored ...... 24 2.4 The pitch location of the goal scorer ...... 26 2.5 The origin of ‘the final pass’ ...... 27 2.5.1 Area C5 or Zone 14 ...... 30 2.5.2 The ‘Crossing’ areas ...... 31 2.6 The number of passes made prior to a goal scored ...... 34 2.7 Regained possessions – the areas of the field ...... 45 2.8 Time in possession ...... 50 2.8.1 Possession in 5 minute intervals ...... 53 2.9 Counter attacks and transitions ...... 56 2.10 Characteristics of successful and unsuccessful teams ...... 59 2.11 Tactical and strategic analysis ...... 68 2.12 Football Analytics ...... 75 2.13 Summary ...... 78 Chapter 3 Methodology ...... 85 3.1 Introduction ...... 85 3.2 The software for coding, capturing and analyzing ...... 85 3.3 The three new categories of goals ...... 86 3.4 Data collection ...... 89 3.4.1 Video resources ...... 89 GOAL SCORING PATTERNS viii
3.4.2 Essential data ...... 90 3.4.3 Additional data ...... 91 3.5 Operational definitions ...... 92 3.5.1 Definition of ‘Possession’ ...... 92 3.5.2 Definition of a goal from a ‘Cross’ ...... 94 3.5.3 Definition of a goal from passing the ball behind the last line of defence or to a player level with the last defender who is in a position to take the ball forwards and shoot (Ball behind and strike – BB&S) or pass to a team mate. (Ball behind, pass and strike - BBP&S) ...... 94 3.5.4 Definition of a goal from ‘Other Methods’ (OM) ...... 95 3.6 Creating the code window ...... 95 3.6.1 Code buttons, Label buttons and the Code Matrix ...... 98 3.7 Capturing the data ...... 104 3.7.1 Equipment ...... 105 3.7.2 Method ...... 106 3.7.3 Accuracy of coding data ...... 107 3.8 Creating a database ...... 113 3.9 Data analysis ...... 114 3.9.1 Analysis using Sportscode Elite ...... 115 3.9.2 Statistical Analysis ...... 118 3.10 Justification of methodology ...... 134 3.10.1 Inclusion of ‘Own Goals’ ...... 135 3.10.2 Amendment to the ‘Definition of Possession’ ...... 135 3.11 Summary ...... 137 Chapter 4 Results and Discussion ...... 139 4.1 Introduction ...... 139 4.2 Data preparation ...... 139 4.3 Reliability test results ...... 140 4.3.1 Intra-Observer Reliability Test ...... 140 4.3.2 Inter-Observer Reliability Test ...... 145 4.4 Research question 1 ...... 146 4.4.1 Test results for categories of goals ...... 149 4.4.2 Discussion – Research question 1 ...... 156 4.4.3 The top three teams compared with the bottom three teams and the average score for BBSPS in the league competitions...... 158 4.4.4 The top four teams in international tournaments compared with the average score for the BBSPS category ...... 163 4.5 Research question 2 ...... 167 4.5.1 Test results for goals from five zones in BBSPS category ...... 169 4.5.2 Discussion – Research question 2 ...... 174 4.5.3 The top and bottom three teams for passes in Zone 14+ in league competitions ...... 176 4.5.4 The top 4 teams for passes in Zone 14+ in International competitions ...... 177 4.5.5 Passes along the ground or in the air from Zone 14+ ...... 177 GOAL SCORING PATTERNS ix
4.6 Confirmed and Unconfirmed Coding ...... 178 4.7 Goals from regained possession in each third and each half of the field ...... 180 4.7.1 Regained possessions by the top three and bottom three teams of the EPL and ‘A’ League ...... 189 4.7.2 Regained possessions by the top four teams in the World Cups and European Championships ...... 191 4.8 Regained possession in the Defending or Back Third ...... 192 4.8.1 The number of passes for goals from regained possessions in the Back Third of the field ...... 195 4.9 Goals from regained possession in the Final Third ...... 198 4.10 The number of passes preceding goals in each competition ...... 203 4.10.1 The number of goals and passes from regained possessions in four areas of the field ...... 205 4.10.2 Goals from 1–5 and 6+ passes from regained possessions in the Back Third of the field ...... 208 4.10.3 The percentage of goals in each category from regained possessions in the Back Third, Own Half of Midfield and Own Half of the field ...... 210 4.10.4 Goals from 1–5 and 6+ passes from regained possessions in the opponents’ half of midfield, ‘Their Half’ of midfield ...... 215 4.10.5 The percentage of goals in each category from regained possession in the opponents’ half of the field and the Final Third ...... 222 4.10.6 Zero pass goals in Other Methods ...... 225 4.10.7 Goals from Crosses and where possession was regained ...... 226 4.10.8 Goals from ‘zero’ passes, 1–5 and 6 or more passes from regained possessions in the Final Third of the field ...... 228 4.10.9 Goals from ‘zero’, 1–5 and 6+ passes for top and bottom three teams in leagues and top four in international competitions ...... 230 4.11 Goals scored inside and outside the penalty area ...... 233 4.12 Goals from Set Plays and Open Play ...... 234 4.13 Summary of Chapter ...... 235 Chapter 5 Conclusions ...... 237 5.1 Introduction ...... 237 5.2 Question 1 ...... 237 5.2.1 Implications for coaches ...... 239 5.3 Question 2 ...... 242 5.3.1 Implications for coaches ...... 242 5.4 Regained possessions and the number of passes preceding goals ...... 244 5.4.1 Goals from regained possession in the Back Third ...... 244 5.4.2 Goals from regained possession in the Middle Third ...... 245 5.4.3 Goals from regained possessions in the Final Third ...... 246 5.4.4 Regained possessions and styles of play ...... 247 5.4.5 Implications for coaches ...... 251 5.5 Future research ...... 254 5.6 Summary ...... 255 GOAL SCORING PATTERNS x
References ...... 259 Appendices ...... 267 Appendix 1 (1) Statistical Tests for Categories of Goals: European Championships and World Cups ...... 267 Appendix 1 (2) Statistical Tests for Categories of Goals: EPL and ‘A’ League ...... 268 Appendix 1 (3) Statistical Tests for Regained Possessions: European Championships and World Cups ...... 269 Appendix 1 (4) Statistical Tests for Regained Possessions: EPL and ‘A’ League ...... 270 Appendix 2 (1) Statistical Tests for Category (a) Goals: European Championships ...... 271 Appendix 2 (2) Statistical Tests for Category (a) Goals: World Cups ...... 272 Appendix 2 (3) Statistical Tests for Category (a) Goals: EPL ...... 273 Appendix 2 (4) Statistical Tests for Category (a) Goals: ‘A’ League ...... 274 Appendix 3 (1) Top three and bottom three EPL teams – Categories of Goals and Areas for Category (a) Goals ...... 275 Appendix 3 (2) Top three and bottom three ‘A’ League teams – Categories of Goals and Areas for Category (a) Goals ...... 276 Appendix 4 (1) Top four World Cup teams – Categories of Goals and Areas for Category (a) Goals ...... 277 Appendix 4 (2) Top four European Championship teams – Categories of Goals and Areas for Category (a) Goals ...... 278 Appendix 5 (1) Regained possessions in the EPL in each third of the field and each half of the field ...... 279 Appendix 5 (2) Regained possessions for the top and bottom three teams in the ‘A’ League in each third and each half of the field ...... 280 Appendix 6 (1) Regained possessions overall and for the top four teams in the World Cup in each third and each half of the field ...... 281 Appendix 6 (2) Regained possessions overall and for the top four teams in the UEFA European Championships in each third and each half of the field ...... 282 Appendix 7 (1) Goals in World Cups from zero pass moves, 1–5 and 6+ passes; goals after a save was made by the goalkeeper and from long forward passes from the Back Third ...... 283 Appendix 7 (2) Goals in UEFA European Championships from zero pass moves, 1–5 and 6+ passes; goals after a save was made by the goalkeeper and from long forward passes from the Back Third ...... 284 Appendix 8 (1) Goals in the EPL from zero pass moves, 1–5 and 6+ passes; goals after a save was made by the goalkeeper and from long forward passes from the Back Third ...... 285 Appendix 8 (2) Goals in the ‘A’ League from zero pass moves, 1–5 and 6+ passes; goals after a save was made by the goalkeeper and from long forward passes from the Back Third ...... 286 Appendix 9 (1) Goals in each category in World Cup 2002; Goals from zero, 1–5 and 6+ passes and; Goals in category (a), BBSPS from passes in designated areas ...... 287
GOAL SCORING PATTERNS xi
Appendix 9 (2) Goals in each category in World Cup 2006; Goals from zero, 1–5 and 6+ passes and; Goals in category (a), BBSPS from passes in designated areas ...... 288 Appendix 9 (3) Goals in each category in World Cup 2010; Goals from zero, 1–5 and 6+ passes and; Goals in category (a), BBSPS from passes in designated areas ...... 289 Appendix 10 (1) Goals in each category in UEFA European Championships 2004; Goals from zero, 1–5 and 6+ passes and; Goals in category (a), BBSPS from passes in designated areas ...... 291 Appendix 10 (2) Goals in each category in UEFA European Championships 2008; Goals from zero, 1–5 and 6+ passes and; Goals in category (a), BBSPS from passes in designated areas ...... 292 Appendix 10 (3) Goals in each category in UEFA European Championships 2012; Goals from zero, 1–5 and 6+ passes and; Goals in category (a), BBSPS from passes in designated areas ...... 293 Appendix 11 (1) Goals in each category in the EPL 2001–02; Goals from zero, 1–5 and 6+ passes and; Goals in category (a), BBSPS from passes in designated areas . 295 Appendix 11 (2) Goals in each category in the EPL 2005–06; Goals from zero, 1–5 and 6+ passes and; Goals in category (a), BBSPS from passes in designated areas .. 296 Appendix 11 (3) Goals in each category in the EPL 2011–12; Goals from zero, 1–5 and 6+ passes and; Goals in category (a), BBSPS from passes in designated areas .. 297 Appendix 12 (1) Goals in each category in the ‘A’ League 2008–09; Goals from zero, 1–5 and 6+ passes and; Goals in category (a), BBSPS from passes in designated areas ...... 299 Appendix 12 (2) Goals in each category in the ‘A’ League 2009–10; Goals from zero, 1–5 and 6+ passes and; Goals in category (a), BBSPS from passes in designated areas ...... 300 Appendix 12 (3) Goals in each category in the ‘A’ League 2011–12; Goals from zero, 1–5 and 6+ passes and; Goals in category (a), BBSPS from passes in designated areas ...... 301 Appendix 13 Type of Cross, the side of the field and the number of goals scored with a header...... 303 Appendix 14 Goals scored from inside and outside the penalty area, the number of touches for each goal and the number of goals scored inside the penalty area with one touch and from headers ...... 305 Appendix 15 (1) Goals and passes from regains in each third of the field and for the top four teams in Euro 2004 ...... 307 Appendix 15 (2) Goals and passes from regains in each third of the field and for the top four teams in Euro 2008 ...... 308 Appendix 15 (3) Goals and passes from regains in each third of the field and the top four teams in Euro 2012 ...... 309 Appendix 16 (1) Goals and passes from regains in each third of the field and the top four teams in World Cup 2002 ...... 311 Appendix 16 (2) Goals and passes from regains in each third of the field and the top four teams in World Cup 2006 ...... 312 GOAL SCORING PATTERNS xii
Appendix 16 (3) Goals and passes from regains in each third of the field and the top four teams in World Cup 2012 ...... 313 Appendix 17 (1) Goals and passes from regains in each third of the field and the top three teams in the EPL 2001–02 ...... 315 Appendix 17 (2) Goals and passes from regains in each third of the field and the top three teams in the EPL 2005–06 ...... 316 Appendix 17 (3) Goals and passes from regains in each third of the field and the top three teams in the EPL 2011–12 ...... 317 Appendix 18 (1) Goals and passes from regains in each third of the field and the top three teams in the A League 2008–09 ...... 319 Appendix 18 (2) Goals and passes from regains in each third of the field and the top three teams in the A League 2009–10 ...... 320 Appendix 18 (3) Goals and passes from regains in each third of the field and the top three teams in the A League 2011–12 ...... 321 Appendix 19 (1) Intra-reliability Test Data and Yule’s Q Test Results for Operator 1 .. 323 Appendix 19 (2) Intra-reliability Test Data and Yule’s Q Test Results for Operator 1 .. 324 Appendix 20 (1) Inter-reliability Test Data and Yule’s Q Test Results for Operator 1, and Operator 2...... 325 Appendix 20 (2) Inter-reliability Test Data and Yule’s Q Test Results for Operator 1, and Operator 2...... 326 Appendix 21 Goals scored from inside the penalty area, 18–23 yards and from 23+ yards ...... 327 Appendix 22 Goals scored from Set Plays ...... 329 Appendix 23 Category (a) goals from passes along the ground and in the air ...... 331 Appendix 24 Success rate of playing from the Back Third through Own Half of Midfield and into the Final Third, Attempts at Goal and Goals after regaining possession in the Back Third ...... 333 Appendix 25 (1) Code Matrix by Operator 1 in July 2012 for UEFA European Championships 2012 ...... 335 Appendix 25 (2) Code Matrix by Operator 1 in December 2012 for UEFA European Championships 2012 ...... 336 Appendix 25 (3) First Code Matrix after initial coding by Operator 2 in December 2012 for UEFA European Championships 2012 ...... 337 Appendix 25 (4) Final Code Matrix by Operator 2 in December 2012 for UEFA European Championships 2012 ...... 338 Appendix 26 (1) Chi Square Test results for categories of goals ...... 339 Appendix 26 (2) Chi Square Test results for goals from regained possessions in each Third of the field ...... 340 Appendix 26 (3) Chi Square Test results for goals from OH, WR, Z14+, WL and IPA 341
GOAL SCORING PATTERNS xiii
List of Tables
Table 2.1 Techniques used to score goals in World Cup 2006 ...... 25 Table 2.2 Key events preceding goals in World Cup 2006 ...... 30 Table 2.3 Goals from sequences between 0 to12 passes ...... 35 Table 2.4 Goals from regained possessions in each third of the field in World Cup 2006 ...... 50 Table 2.5 Time in possession preceding goals in World Cup 2006 ...... 52 Table 3.1 Total games and goals in each competition ...... 90 Table 3.2 Number of matches and goals analyzed in each competition ...... 105 Table 3.3 The combined factors for categorizing data in Yule’s Q Test ...... 121 Table 3.4 Calculations of Kappa and Cronbach’s Alpha ...... 122 Table 3.5 Calculation of Percentage Error ...... 123 Table 3.6 Alternative methods of calculating Percentage Error ...... 124 Table 3.7 Yule’s Q Test for coding two categories of events ...... 125 Table 3.8 Yule’s Q Test for BBSPS and Other Methods ...... 126 Table 3.9 Comparing data of equal numbers ...... 127 Table 4.1 Intra-Reliability Test data for Operator 1 ...... 141 Table 4.2 Inter-Reliability Test data for Operator 1 and Operator 2 ...... 145 Table 4.3 Goals Scored in FIFA World Cups ...... 147 Table 4.4 Goals Scored in the European Championships ...... 148 Table 4.5 Goals scored in the English Premier League (EPL) ...... 148 Table 4.6 Goals Scored in the Australian ‘A’ League ...... 149 Table 4.7 Statistical Test results for categories of goals in the European Championships .... 150 Table 4.8 Chi Square test results for categories of goals in European Championships ...... 152 Table 4.9 Statistical Test Results for categories of goals in World Cups ...... 153 Table 4.10 Chi Square Test Results for categories of goals in World Cups ...... 155 Table 4.11 Percentage difference between goals from BBSPS and Other Methods in all competitions ...... 157 Table 4.12 Comparison of top and bottom three teams in the ‘A’ League in 2011–12 ...... 159 Table 4.13 Comparison of top and bottom three teams in the EPL in 2011–12 ...... 160 Table 4.14 Goals in categories for Manchester United and Arsenal for three seasons over ten years ...... 162 GOAL SCORING PATTERNS xiv
Table 4.15 Comparison of top 4 teams with the average in the 2010 World Cup ...... 163 Table 4.16 Comparison of top 4 teams with the average in the 2008 European Championships ...... 164 Table 4.17 Goals and categories for Germany and Spain ...... 165 Table 4.18 Goals in the World Cup from a pass in each of the five Zones ...... 167 Table 4.19 Goals in European Championships from a pass in each of the five Zones ...... 168 Table 4.20 Goals in the English Premier League from a pass in each of the five Zones ...... 168 Table 4.21 Goals in the Australian ‘A’ League from a pass in each of the five Zones ...... 169 Table 4.22 Test results for goals in World Cups from the five zones in the BBSPS category ...... 170 Table 4.23 Test results for goals from five zones in the EPL ...... 172 Table 4.24 Mean scores for goals from (BBSPS) from each area of the field ...... 175 Table 4.25 Goals from passes along the ground in category (a) BBSPS ...... 178 Table 4.26 Comparison of ‘Confirmed’ passes with totals in each category ...... 179 Table 4.27 Comparison of ‘Confirmed regained possessions with total in each area of the field ...... 180 Table 4.28 Regained possession in each third of the field in the EPL and ‘A’ League ...... 181 Table 4.29 Test results for EPL goals from regained possession in each ‘Third’ ...... 182 Table 4.30 Chi Square test results (1) for EPL goals from regained possessions in each ‘Third’ ...... 184 Table 4.31 Pearson’s Chi Square Test Results (2) for regained possessions in the EPL ...... 185 Table 4.32 Regained possession in Own Half (OH) and Their Half (TH) in the EPL and ‘A’ League ...... 186 Table 4.33 Regained possession in each third of the field in the World Cups and European Championships ...... 187 Table 4.34 Regained possession in Own Half (OH) and Their Half (TH) in the World Cups and European Championships ...... 187 Table 4.35 The top and bottom three teams in the EPL in 2011–12 ...... 189 Table 4.36 Regained possessions overall and by the top four teams in Euro 2012 ...... 191 Table 4.37 Regained possessions in Middle Third (M3) overall and by the top four teams in the World Cups and European Championships ...... 192 Table 4.38 Regained possessions in the Back Third and Final Third ...... 193 Table 4.39 Goals from regained possession in the Back Third and the number of passes .... 195
GOAL SCORING PATTERNS xv
Table 4.40 Goals and categories from regained possession in the Back Third in the ‘A’ League 2011–12 ...... 198 Table 4.41 The number of goals and passes in each competition ...... 204 Table 4.42 Regained possessions and passes in European Championships ...... 206 Table 4.43 Regained possessions and passes in the World Cup ...... 206 Table 4.44 Regained possessions and passes in the EPL ...... 207 Table 4.45 Regained possessions and passes in the A League ...... 207 Table 4.46 EPL and A League goals from regained possession in the Back Third ...... 208 Table 4.47 Euro and World Cup goals from regained possession in the Back Third and the number of passes ...... 209 Table 4.48 Goals in each category from regained possessions in the Back Third, OH of MF and Own Half and number of passes ...... 210 Table 4.49 Percentage of goals from regained possession in the teams own half of the field in each category ...... 212 Table 4.50 Percentage of goals from regained possession in the teams own half of the field in each category ...... 214 Table 4.51 Regained possession in the opponents half of midfield Euro 2012 ...... 219 Table 4.52 Regained possession in the opponents half of midfield WC 2010 ...... 219 Table 4.53 Regained possession in the opponents half of midfield in the ‘A’ League 2011–12 and goals from 5 passes or less and 6+ passes ...... 220 Table 4.54 The percentage of goals from regained possession in the opponents Half, ‘their half’ (TH) and Final Third (F3) in World Cups and European Championships in each category ...... 223 Table 4.55 The percentage of goals in each category from regained possession in the opponent’s half (TH) and Final Third (F3) in EPL and A League ...... 224 Table 4.56 ‘Zero’ pass goals from 4 areas of the field and percentage of total in Open Play ...... 224 Table 4.57 Crosses and the areas where possession was regained in World Cups and European Championships ...... 226 Table 4.58 Crosses and the areas where possession was regained in the EPL and ‘A’ League ...... 227 Table 4.59 Goals and number of passes from regained possession in the Final Third ...... 229 Table 4.60 The top three teams in the EPL and the overall scores for goals from ‘Zero’, 1-5 and 6+ passes ...... 230
GOAL SCORING PATTERNS xvi
Table 4.61 The top three teams in the ‘A’ League and the overall scores for goals from ‘Zero’, 1–5 and 6+ passes ...... 231 Table 4.62 The top 4 teams in the European Championships and the overall percentages for goals from ‘Zero’, 1–5 and 6+ passes ...... 232 Table 4.63 The top 4 teams in the World Cups and the overall percentages for goals from ‘Zero’, 1–5 and 6+ passes ...... 232 Table 4.64 The distance from the goal line for goals in the World Cups 2002–2010 ...... 233 Table 4.65 Average percentage of goals from Set Plays and Open Play ...... 234 Table 4.66 Percentage of goals in each category of Set Plays ...... 234
GOAL SCORING PATTERNS xvii
List of Figures
Figure 1.1 The 18 zones of the field ...... 8 Figure 1.2 The 2 crossing areas and 5 areas for passes in category (a) goals ...... 11 Figure 2.1 Percentage of goals scored in intervals of 15 minutes ...... 22 Figure 2.2 Areas on the field from where goals were scored ...... 26 Figure 2.3 The 18 equal areas of the field ...... 28 Figure 2.4 The 18 unequal zones on the field of play ...... 28 Figure 2.5 Areas on the field from where the final pass was played prior to a goal in Open Play ...... 29 Figure 2.6 Origins of crosses leading to goals in Open Play in 2006 World Cup ...... 33 Figure 2.7 Analysis of the mean number of goals scored per 1000 possessions for the 1990 and 1994 World Cups ...... 41 Figure 2.8 Number of passes preceding goals in 2002 and 2006 World Cups ...... 43 Figure 2.9 Ratio of goals from regained possessions in each third of the field ...... 46 Figure 2.10 Accuracy and number of passes in UEFA Champions League 2012–13 ...... 72 Figure 2.11 Average total time in possession in UEFA Champions League 2012–13 ...... 73 Figure 3.1 Areas of the field for passes behind the defence and crosses ...... 87 Figure 3.2 Example of a code window using two generic buttons ...... 96 Figure 3.3 Code and Label Buttons and their identifiers ...... 97 Figure 3.4 The Timeline with highlighted incident #5 ...... 99 Figure 3.5 The Code Matrix – Code buttons Horizontally and Label buttons vertically ...... 99 Figure 3.6 Code Matrix – Combining label buttons ...... 101 Figure 3.7 The Code Window – Grey Code Buttons and Coloured Label Buttons ...... 102 Figure 3.8 The Code Window for the European Championships 2012 ...... 104 Figure 3.9 The Code Matrix – the number of goals from different passing sequences ...... 109 Figure 3.10 The Timeline with the highlighted incidents from the Code Matrix ...... 109 Figure 3.11 Highlighted incident #26 in the Timeline ...... 110 Figure 3.12 The Label Mode Function button, 3rd from left in the Code Window ...... 110 Figure 3.13 Updated Code Matrix...... 111 Figure 3.14 A Code Matrix with all Label Buttons included ...... 112 Figure 3.15 Creating a database ...... 113 GOAL SCORING PATTERNS xviii
Figure 3.16 Highlighted incidents to be transferred to the database ...... 113 Figure 3.17 Transfer the selected instances to the database ...... 114 Figure 3.18 Exporting the standalone movie file of incidents to the database ...... 114 Figure 3.19 Sample of a Timeline showing Code Buttons and Incidents in each row ...... 115 Figure 3.20 The Code Matrix showing Categories on the left, labels at the top and the frequency of recordings in the corresponding boxes ...... 116 Figure 3.21 Two Label buttons combined in the Code Matrix ...... 117 Figure 3.22 Three Label buttons combined in the Code Matrix ...... 117 Figure 3.23 Four Label buttons combined in the Code Matrix ...... 118 Figure 3.24 The Code Matrix with all Label buttons at the top ...... 131 Figure 3.25 Extracts of the full Code Matrix shown in Figure 3.24 ...... 131 Figure 3.26 A colour coded Code Window ...... 133 Figure 3.27 Extracts of the 2nd Code Matrix by Operator 2 ...... 134 Figure 4.1 The Code Matrix for the European Championships in June 2012 ...... 141 Figure 4.2 The Code Matrix for the European Championships in December 2012 ...... 142 Figure 4.3 The Code Matrix for the European Championships in December 2012 for Operator 2 ...... 146 Figure 5.1 Description of how goals in Open Play were scored in 2014 World Cup Finals . 238 Figure 5.2 A practice for two teams to replicate goal scoring opportunities in each category ...... 241 Figure 5.3 Net successful passes versus opponents in the 2014 World Cup ...... 251
GOAL SCORING PATTERNS 1
Chapter 1 Introduction
Overview
This chapter introduces the framework for the thesis. It begins with an overview of significant contributions in the fields of notational analysis and coaching theory. It summarizes some of the early studies on the topic of scoring goals in Association Football
(football) and provides a snapshot of the current state of research. There follows an explanation how the use of technology has changed in the analysis of performance and its impact in providing coaches information to broaden their knowledge of events preceding goals being scored. The chapter explains the purpose of the research and identifies the research questions that form the basis of the investigation reported here. The objectives and scope of the research are discussed. The chapter concludes with a synopsis of the remaining chapters in the thesis.
1.2 Background of the study
Chapter 2 provides a detailed literature review. Here, I outline some key issues that have framed my research. Charles Reep pioneered the notational analysis of football. He developed a short hand system to record match events and outcomes in real time and used his system over many years to produce evidence to support his findings that chance played a significant part in football (see, for example, Reep, & Benjamin, 1968). His research did note that some aspects of play did have a degree of predictability: for example, 80% of goals scored came from sequences of three passes or less; and that it takes an average of ten shots to produce one goal. Charles Reep worked with several professional football clubs between 1950 and 1967 and analyzed over 2000 matches during his career (Lyons,
2011). He continued to analyze matches for decades until he passed away aged 92. The principles of Charles Reep’s analysis resurfaced in Charles Hughes’ (1990) second book GOAL SCORING PATTERNS 2
and video series titled ‘The Winning Formula’. In 1973, Hughes produced his first book and film series titled ‘Tactics and Teamwork’, which received worldwide recognition for its detailed contribution to coaching. He wrote the book based on his experience of coaching the England Amateur football team and it was essentially about the benefits of good teamwork and how success could be achieved. The examples of good play in the video clips showed the England players in the 1966 World Cup. The foreword described the book as being ‘in a league of its own’. There were very few references to statistical analysis in the book but one statement was that 40% of goals were scored from Set Plays.
Another heading in the chapter titled, ‘The Do’s and Don’ts of Passing, read, “Don’t play long optimistic inaccurate passes’. Hughes went on to write, “There is nothing wrong with long passes as long as they are accurate. But so often the intention is incorrect” (Hughes,
1973:78). This is ironic because in 1990 Hughes produced his second series titled, ‘The
Winning Formula’ and for years afterwards he was criticized for encouraging the use of long forward passes to the back of the opposition’s defence to exploit their vulnerability.
Unfortunately, long forward passes became synonymous with ‘Direct Play’, which ignored other aspects of play that Hughes described as being integral in the ‘Winning Formula’.
Hughes’ second book and video series contained a lot of statistical analysis to support his theory that ‘Direct Play’ was the way to win matches and that ‘possession based’ football was not the style of play to adopt because of the very low percentage of goals from six passes or more. Hughes wrote, “patient possession football does not produce the goals that win matches” (1990:172) and “87% of goals scored, almost seven out of every eight, came from five consecutive passes or less” (1990:173). Hughes’ philosophy of ‘Direct Play’, involved more than long passes as he explained. “Our analysis has shown that most goals are scored from positions where possession has been regained in the attacking third of the field” (1990:174), revealing that 106 of the 202 goals in the 109 matches came that way. GOAL SCORING PATTERNS 3
Another attribute that Hughes advocated was attacking as quickly as possible to pass the ball behind the opposing defenders, when they were most vulnerable, which is immediately after possession is lost. Unfortunately, Hughes used video clips of Brazilian players when he said, possession based football was not the way to win matches, despite producing the evidence that Brazil, the most successful national team in international football scored two out of every three goals with five passes or less. Hughes’ reputation was tarnished and he lost credibility because possession football was appreciated for its style, grace and the majority of people around the world believe it is the way the game should be played and most teams try to play that way.
Mike Hughes and Ian Franks have written numerous papers about performance in football and have proposed that data on the number of passes preceding goals should be
‘normalized’ by dividing the number of outcomes by the frequency of their occurrences
(Hughes & Franks, 2005). They normalized data on passing frequencies and showed teams that adopted a possession-based style of play achieved more shots at goal from longer passing sequences. Quantitative analysis of football related activities, including events preceding goals, continued in earnest at Liverpool John Moore’s University (LJMU) under the direction of Professor Tom Reilly and subsequently Professor Mark Williams. The
Football Association engaged LJMU from the early 1990s until 2007 to conduct research on their behalf. At that time, LJMU was acknowledged as the centre for football research in the northern hemisphere.
In the past fifteen years there has been an enormous increase in data about football performance due to the emergence of companies such as Amisco, Infostrada, Opta and
Prozone. The companies provide quantitative data to football clubs and national football associations. Today, most professional football clubs employ full time performance GOAL SCORING PATTERNS 4
analysts. There is a proliferation of organizations and individuals who use the large amounts of data being generated to provide detailed commentary on football performance.
Anderson and Sally (2013) consider how the use of objective information has challenged tradition wisdom in the game. The data generated by the companies are used primarily to analyze performance for the benefit of the clubs but they are used for predictive and comparative purposes by different agencies too. For example, data on individuals are used by clubs for talent identification, recruitment and retention of players, team data are used by betting agencies to identify the teams most likely to win or be relegated at the end of the season; and by newspapers and online media to create content for their readers.
1.3 Technology and topics associated with scoring goals in football
Technology has had a direct impact on the quality of research in football performance. In the 1950s, the available equipment was limited to pencil, pen, paper and a stopwatch, which explains why notational analysis focused on topics that could be counted or timed live at games. Charles Reep focused on: the number and direction of passes, the areas of the field where possession was regained and the probability of scoring goals.
Eventually, he collaborated with Bernard Benjamin, a statistician, to produce the paper
‘Skill and Chance in Association Football’ (Reep & Benjamin, 1968). In those days a journal for football research did not exist. The paper was published in the Journal of the
Royal Statistical Society (Lyons, 2015). Charles Hughes produced his first book in the
1970s but the films for each chapter were in 16mm, which required a projector to play them. In the 1980s, videocassette recorders (VCRs) became affordable and this advance in technology allowed researchers the opportunity to rewind and replay events and notate events with greater accuracy and reliability in lapsed time. Analysis was still focused on frequency data such as the number and type of passes prior to a goal, the number of GOAL SCORING PATTERNS 5
touches prior to scoring or the number of goals from inside and outside the penalty area or the shots to goal ratios but the opportunity to rewind match play allowed additional information to be noted. For example, analysts were able to record how long the team had possession of the ball in each passing sequence, the total time each team had possession of the ball, the position on the field where possession was regained and where the final pass was made prior to a goal, the type or style of play, whether teams used counter attacks or had an ‘organized’ approach in attack, which led to more analysis of the most productive areas of the field.
The availability of VCRs allowed people with the time and inclination to analyze whatever they wanted as long as the quality of the footage allowed it because most of it was recorded from television broadcasts. VCRs created the opportunity for researchers to record and archive matches for future reference, which has allowed comparative analysis of events or teams within a competition and between different competitions. Comparative analysis has been used to identify differences or consistencies in football generally or between successful teams and unsuccessful teams over time. For example, quantitative data has shown: the ball was in play for between 7–8 minutes less in the four FIFA World Cups since 1998 (FIFA, 1998; FIFA, 2002; FIFA, 2006; FIFA, 2010; FIFA, 2014); the average number of goals scored since the 1958 FIFA World Cup is between 2.21 and 2.8 per match
(FIFA, 2014).
A consistent aspect of scoring goals is that the majority is still scored with five passes or less (Breen, Iga, Ford & Williams, 2006; Grant, Reilly, Williams & Borrie, 1998;
Reep & Benjamin, 1968).
Since the mid to late 1990s, the affordability and availability of user-friendly software, designed to analyze match events and provide quantitative data, has allowed performance analysis to expand rapidly to the point where ‘live’ coding and ‘live’ review GOAL SCORING PATTERNS 6
of match events is standard, better practice. Also, the availability of affordable storage devices and reduction of file sizes has facilitated the creation of huge databases of football games.
The more sophisticated the software, the greater the temptation there is to analyze more actions by individuals and teams. This diversity of analysis has produced data that could be described as ‘Useful’ and ‘Useable’. For example, analysis shows that more goals are scored with five passes or less but would that discourage players from trying to score with six passes or more? The information that many goals are scored in the fifteen minutes prior to half time is useful and might encourage a higher level of concentration during this period in a match, but it is not practical to implement at training. The information that approximately 80% of goals in Open Play are scored from inside the penalty area and with one or two touches is useable information that can be implemented at training by setting the same game conditions. Information on the success rate of maintaining possession from the Back Third of the field to the Final Third will drive the coaching process and can be implemented at training. With the ‘useable’ concept in mind a new approach to analyzing goals has been conceived in the research reported here.
1.4 Justification for a different approach to analyze goal scoring
There is a limit to what can be counted or timed in relation to events preceding goals and for comparative reasons it is helpful to analyze the same topics over time.
Comparative studies have confirmed similarities or highlighted slight changes in what happens in the game. Examples of this research include: the number of goals scored with five passes or less; the number of touches before scoring, the percentage of goals scored inside the penalty area, the percentage of goals scored in Open Play and from Set Plays, the most successful position of the goal-scorer inside the penalty area, the most common part of the body used to score goals, the shots to goal ratio of successful teams, that GOAL SCORING PATTERNS 7
successful teams make more completed passes in the final third than other teams, that time in possession is not an absolute indicator for success; and so on. Unfortunately, research on aspects of goal scoring mentioned above has remained constant for many years. A new approach to analyzing goal scoring in football is required and preferably, one that provides useable information that can be implemented in a practical setting.
It is a fact that the vast majority of goals are scored from inside the penalty area, which makes football predictable in terms of where every team has to get the ball and have at least one player who can propel it towards the goal. Therefore, it is logical to analyze how the ball and the player(s) arrive inside the penalty area or close to it, because very few goals are scored from outside a distance of twenty-three yards. The offside law allows the last line of defenders to determine where attackers can position themselves by moving forward to place the attacker(s) in a position closer to the goal than the defenders. The defenders can do this anywhere inside their own half of the field and if a forward pass is made to an attacker in this situation he/she will be offside and a free kick will be awarded to the defending team.
The challenge for attackers is to get into a scoring position inside the penalty area, or close to it, without being offside. There are three possible solutions. The first is to run with the ball, or dribble, past the last line of defenders to score. The second is to run towards the last line of defence, when a teammate can pass the ball forwards, with the intention of receiving the ball behind the defenders, without being off-side. The third solution is to receive the ball within scoring range when defenders are technically closer to the goal but unable to block the shot or header at the goal.
To date, there is no research evidence to determine which of the three solutions is the most effective for scoring goals but this thesis will do so by placing goals in one of the three categories linked to the solutions. Research has shown that the majority of goals are GOAL SCORING PATTERNS 8
scored with five passes or less (Hughes, 1990; Hughes & Franks, 2005). Grant et al. (1998) and Rees, James, Hughes, Taylor & Vuckovic, (2010) refer to a central area of the field, which is just outside the penalty area, as Zone 14, while Breen et al. (2006) refer to it as area C5 (see, Figure 1.1).
Figure 1.1 The 18 zones of the field Source: Rees, G., James, N., Hughes, M., Taylor, J. & Vuckovic, G. (2010:206)
Zone 14 was identified as the most productive area for passes leading to goals; but specific research has not been undertaken to analyze goals as a result of getting behind opponents or which areas of the field are the most effective to pass the ball behind the opposition. Reference has been made to goals from passes ‘over the top’, ‘through balls’ or from ‘diagonal balls’ (Hughes & Churchill, 2005; Hughes & Snook, 2006) but these have been to describe different types of passes rather than a method of scoring goals, which is the primary focus of this study. Another method of scoring goals in Open Play is from
‘Crosses’, which is recognized universally but without an accepted definition of what constitutes a cross. A distinction is made in the literature between goals in Open Play and GOAL SCORING PATTERNS 9
from Set Plays and the terms ‘counter attack’, ‘possession based’ or ‘direct’ are used more to describe a style of play than a method of scoring.
The recording of the number of goals from where possession is regained has added another dimension in the analysis of events preceding goals but previous research has not reported the number of goals, for example, from five passes or less from regained possessions in each third of the field. The definition of possession determines where on the field possession is regained. In most studies possession would be regained if a player on the opposite team had control over the ball following a ricochet off a defender or after a save by the goalkeeper. If a goal was scored in either of these circumstances it would be recorded as a zero pass goal from regained possession in the Final Third of the field. This interpretation explains why ‘zero’ pass goals are reported but usually excluded from analysis of events preceding goals. In other words, the goals are counted but treated as if nothing happened before they were scored and therefore excluded from further consideration, thereby reducing the number of goals included in the analysis. Goals classified as ‘Own Goals’ are excluded from further analysis too thereby reducing further the total number of goals. Analysis of goals in the literature is reported using whole tournament with limited comparisons of top teams’ performance. Most studies on successful and unsuccessful teams compare the teams at the top and bottom of the competition. This thesis compares the goal scoring patterns of top and bottom teams with all the teams in the league competitions but not with the bottom teams in international tournaments due to the small number of goals scored. This study also compares the goal scoring patterns in four different competitions over periods of four and ten years and between and within two international tournaments and league competitions. GOAL SCORING PATTERNS 10
1.5 A new approach to analyze goal scoring patterns in football
The first task in developing a new approach to analyzing goal-scoring patterns in football is to distinguish between goals scored in Open Play and from Set Plays. Goals from Set Plays are important and are accounted for but not included in the analysis in this thesis. The second task is to create a taxonomy of goal-scoring categories, which represent three strategic ways to solve the challenge of getting a player into a scoring position without being offside. These concepts are addressed within in this chapter but clarification of operational definitions is provided in Chapter 3. The three categories of goals in Open
Play are, (a) from passes behind the last line of defense or to a player level with the last defender, (b) from Crosses and (c) from Other Methods, the latter of which includes any goals not in the first two categories.
Goals in categories (a) and (b) relate to areas of the field where the final pass is made to determine the type of goal, for example, a pass into the penalty area from the side determines the goal to be from a ‘Cross’. In this research, I propose new locations for the origins of category (a) goals. These are described as Wide Left (WL), Wide Right (WR),
Zone 14 + (Z14+), Own Half (OH) and Inside the Penalty Area (IPA). To create the areas of the field where the pass is made for category (a) goals, the customary eighteen areas used in the extant literature (see, for example, Hughes & Snook, 2006) were reduced to seven. The rationale for reducing the number of areas is because this study, unlike other studies, acknowledges the impact of the offside rule and that it permits the last line of defence to be anywhere between the halfway line and the goal line. Therefore it is logical to divide the field vertically into three areas by extending the sides of the penalty area to the halfway line and horizontally by using the halfway line and an imaginary line, parallel with the penalty area but outside the width of the penalty area to the touchlines on both sides of the field. The final areas would be determined by the perimeter of the penalty area. GOAL SCORING PATTERNS 11
By using the existing line markings on the field and only two short imaginary lines it created: the crossing areas on either side of the penalty area; the area referred to as Zone
14+, between the penalty area and the halfway line; and the Wide Right and Wide Left areas, from the crossing areas to the halfway line outside the width of the penalty area. The penalty area has a marked perimeter and the halfway line creates the division between a team’s own half and the opposition’s half of the field. Improved accuracy is an additional benefit from using existing line markings when coding where passes are made on the field.
A diagram of the designated areas on the field is Figure 1.2.
Figure 1.2 The 2 crossing areas and 5 areas for passes in category (a) goals
It was mentioned earlier that scoring from Crosses is accepted universally but there is not an agreed operational definition to determine exactly where a cross might originate. GOAL SCORING PATTERNS 12
In this study, the ‘Crossing’ area is outside the penalty area and within twenty yards of the goal line. This study is different to other studies in the way regained possessions inside the penalty area are defined. This study includes all goals because once the ball has reached the penalty area any rebounds or ricochets are ignored and the goal is classified according to where possession was regained and how the ball reached the goal scoring position, which will be through one of the three categories. There are fewer zero pass goals but they still occur, for example, after a clearance from the penalty area is regained and the player in possession dribbles past one or two defenders to score; a goal like this would be included in the Other Methods category. Therefore the new approach is totally inclusive and will reflect an accurate assessment of events preceding all goals in Open Play.
1.6 The aim and objectives of the thesis
The aim of the research reported here is: to provide coaches with information about goal scoring patterns in football that is easy to understand and that encourages them to use this information to transform players’ performance in training and competition.
The aim leads to five objectives for this study:
1) To provide new information about goal-scoring behaviours in football.
2) To enable comparisons to be made with goal-scoring patterns reported in the
literature.
3) To discuss the detail required in match analysis in order to make accurate
assessments of performance.
4) To evaluate whether previous analysis of events preceding goals may have
identified events that have been and will continue to occur regardless of a
‘preferred’ playing style. GOAL SCORING PATTERNS 13
5) To consider if football is a combination of the styles of play referred to as
‘possession-based’ and ‘direct play’ rather than a choice of one or the other.
1.7 Research Questions
In total there are eighteen research questions addressed in this thesis. Questions 1 and 2 are the primary questions. The remaining sixteen questions (3–18) add detail to the primary questions and address gaps in the extant literature. These sixteen questions focus on the debate about regained possessions in the various thirds of the field and number of passes in a sequence leading to a goal. A detailed discussion of these questions is presented in Chapter 4.
The questions are:
1) Are more goals scored from passing the ball behind opponents or to a player level
with the last defender than from Crosses or Other Methods?
2) In which area of the field do the majority of passes originate that lead to goals
from passes behind the opposing defence?
3) Do the top and bottom teams in each competition score goals in similar
proportions in each of the three categories and other aspects of performance?
4) Are more goals scored from passes along the ground or in the air from Zone 14+?
5) Which third of the field has the most regained possessions that lead to goals?
6) Are more goals scored from regained possessions in the teams’ own half or the
opponents’ half of the field?
7) What percentage of the total goals are scored from regained possessions in the
Back Third and are more goals a result of sequences of 6+ passes compared with
1–5 pass sequences? GOAL SCORING PATTERNS 14
8) What percentage of the total number of goals is from regained possessions in the
Final Third and which method of regaining possession is the most prevalent?
9) What percentage of the total number of goals is from a ‘Zero’ pass move and
sequences of 1–5 passes and 6+ passes?
10) What percentage of the goals from a ‘Zero’ pass move and sequences of 1–5
passes and 6+ passes is from the four designated areas of the field?
11) What percentage of goals from regained possessions in the Back Third is from
passing sequences of 1–5 and 6+ after long forward passes, goal kicks and drop
kicks are removed?
12) What percentage of goals in each category is scored from regained possessions in
the Back Third, own half of midfield and own half of the field?
13) What percentage of goals in each category, from passing sequences of 1–5 and 6+,
is from regained possessions in the opponents’ half of midfield and the opponents’
half of the field?
14) What percentage of goals in each category is from regained possessions in the
opponents’ half of midfield and the Final Third?
15) What percentage of goals from regained possessions in the Final Third is from
‘zero’ pass moves, and sequences of 1–5 and 6+ passes?
16) Which area of the field has the most regained possessions leading to goals from
Crosses?
17) What percentage of goals is scored from inside the penalty area, between 18 to 23
yards and further than 23 yards from goal?
18) What percentage of goals is scored from Set Plays and Open Play? GOAL SCORING PATTERNS 15
1.8 The scope of the research
It is not a difficult task to find an interesting research topic in football but gaining access to whole game videos can limit the scope of what can be achieved. In this respect, it was fortunate that for more than ten years I had recorded and retained videocassettes and digital video discs of English Premier League (EPL) matches, EPL highlight programs,
Australian ‘A’ League matches, FIFA World Cup and UEFA European Championship matches.
This thesis reports data collected from a strategic sample of three seasons from the
EPL and ‘A’ League matches starting with the oldest season that had a majority of whole- game recordings. Data from three international tournaments were analyzed. The available
World Cup database included every match from the finals tournament in 2002, 2006 and
2010 World Cup Finals and the European Championship database included every match from the Finals tournament in 2004, 2008 and 2012. The EPL seasons were 2001–02,
2005–06 and 2011–12 and the ‘A’ League seasons were 2008–09, 2009–10 and 2011–12.
The decision to limit the research to include three samples of each competition was made because it was felt the sample would be sufficiently large enough to provide answers to the research questions and the time it would take to find, capture and code the goals in the computer was unknown but within acceptable expectations.
1.9 Synopsis of the Thesis
The remaining four chapters of the thesis are: Literature Review (Chapter 2);
Methodology (Chapter 3); Results and Discussion (Chapter 4); and Conclusions (Chapter
5). Whilst a number of the topics on events preceding goals that can be counted or timed were listed earlier in this chapter, the Literature Review provides a much more detailed account of the literature and includes discussion of the major contributions by Charles
Reep and Charles Hughes. References to more recent techniques of behavioural analysis GOAL SCORING PATTERNS 16
applied to sport, including T-patterns are included. The value of possession is examined as an indicator of success, how possession of both teams is affected immediately after goals are scored and how time in possession may reflect a style of play. The Literature Review highlights the importance of operational definitions and aspects of goal scoring where additional information would improve understanding or where new research is required to add to the existing body of knowledge.
The Methodology chapter explains the operational definitions used in this research, what had to be captured and analyzed to answer the research questions as well as the procedure that had to be followed to ensure the accuracy and reliability of coding was of the highest standard. Data were collected with the Sportscode Elite software (Version 9;
Sportstec, Sydney, Australia). A detailed explanation of the use of the software is provided and this includes an account of creating a valid and reliable Code Window. The
Methodology chapter includes the data analysis and justification why the statistical tests chosen were used to answer research questions. There is a discussion of intra- and inter- operator reliability too.
The Results and Discussion chapter examines the outcomes of the analysis to answer the research questions and includes discussion on the comparisons of results between the international and league competitions overall, as well as the comparison of top teams with the overall results and the top teams with the bottom teams. The results for the international tournaments and the league competitions are compared to identify trends or changes during the years between the first and the most recent competition. Examples of particular aspects in the results, such as, ‘Regained possessions by the top four teams in
Euro 2012’, are provided in the text to serve as an illustration of what is included in the relevant Appendices. The Conclusions and final chapter of the thesis summarize the results of the investigation and includes practical and theoretical implications for coaches. The GOAL SCORING PATTERNS 17
chapter concludes by recognizing the limitations of the study and makes suggestions for future research in goals scoring patterns.
GOAL SCORING PATTERNS 19
Chapter 2 Literature Review
2.1 Introduction
The major focus of research in football over the past four decades has been on goal scoring patterns. Researchers have tended to focus on a combination of performance indicators. The catalyst for research can be traced back to Wing Commander Charles Reep, whose pioneer work on the number of passes made prior to goals being scored has been repeated more than any other aspect of goal scoring (Reep & Benjamin, 1968).
Early analyses of football performance focused on some basic performance indicators to explain how goals are scored; the number of touches taken by the scorer, which part of the body is used predominantly to score, (See Table 2.1, page 25) the position of the scorer on the field of play (See Figure 2.2, page 26) and the key events prior to a goal being scored. (See Table 2.2, page 30) Notational analysts expanded these data points to include where on the field the final pass originated, (See Figures 2.3, 2.4 and 2.5, pages 28 and 29) the number of passes made in the final sequence, (See Figure 2.7 and 2.8, pages 41 and 43 and Table 2.3, page 35) the type and length of pass that preceded the goal, which area of the field possession was regained (See Figure 2.9, page 46 and Table 2.5, page 52) and how long the team had possession of the ball (See Table 2.4, page 50) (Breen et al., 2006; Carling, Williams & Reilly, 2005; Garganta, Maia & Basto, 1995; Horn,
Williams & Ensum, 2002; Hughes, 1990; Olsen, 1986; Rees et al., 2010; Taylor, Ensum &
Williams, 2002; Turner & Sayers, 2010).
Temporal analysis of goals scored has been of ongoing interest to researchers and includes the number of goals scored in each half of the game and during smaller time intervals throughout the match, (See Figure 2.1, page 22) (Αbt, Dickinson & Mummery,
2002; Acar, Yapicioglu, Arikan, Yalcin, Ates & Ergun, 2009; Giampietro, Arcelli, GOAL SCORING PATTERNS 20
Cavaggioni & Rampinini, 2013; Grant et al., 1998; Grant, Williams, Lee & Reilly, 1999;
Jinshan, Xiaoke, Yamanaka & Matsumoto, 1993).
Temporal analysis also includes the length of time a team has possession of the ball prior to scoring (Breen et al., 2006; Carling et al., 2005; Taylor et al., 2002; Turner &
Sayers 2010; Wright, Atkins, Polman, Jones & Srageson, 2011).
The number of passes in the sequence preceding a goal has fuelled the debate about the most effective styles of play, described as ‘Direct Play’ or ‘Possession-based’ (Hook &
Hughes, 2001; Hughes, 1990; Hughes & Franks, 2005; Mitrotasios & Armatas, 2014; Reep
& Benjamin, 1968). The debate over playing styles has led to comparisons between successful and unsuccessful teams within a tournament and winners of the same tournament in different years (Bell-Walker et al., 2006; Castellano, Casamichana & Lago,
2012; Grant & Williams, 1998; Grant et al., 1998; Horn, Williams & Grant, 2000; Hughes,
Robertson & Nicholson, 1987; Hughes & Churchill, 2005; Lago-Ballesteros & Lago-Penas,
2010; Low, Taylor & Williams, 2002; Scoulding, James & Taylor, 2002; Taylor &
Williams, 2002; Tenga & Sigmundstad, 2011; Yates, North, Ford & Williams, 2006).
Possession is seen by many researchers to be fundamental to success in football but teams that dominate in this aspect of play do not always win. Attempts have been made to quantify the amount of possession a team has at different times during a match, how it might be influenced by the evolving score and if possession is a better indicator of success than shooting accuracy (Collett, 2013; Jankovic, Leontilevic, Pasic & Jelusic, 2011; Jones,
James & Mellalieu, 2004; Redwood-Brown, 2008; Ridgewell, 2011).
More recent analysis has included the effects of playing at home or away and how venue location influences the outcome of matches (Gomez, Gomez-Lopez, Lagos &
Sampaio, 2012) and how scoring the first goal influences the chances of winning GOAL SCORING PATTERNS 21
(Werlayne, 2013b). While this analysis is valuable and attempts to explain why some teams are more likely to win matches, these factors are beyond the scope of this research.
This literature review is divided into eleven themes.
2.2 When goals are scored
Research over several decades has investigated if more goals are scored in the first or second halves of games and in smaller time intervals within these halves (fifteen minute intervals and smaller units of five and three minutes). Studies have been conducted on competitions with different durations. These studies have included several seasons in different European Leagues (Giampietro et al., 2013) and the National Soccer League
(NSL) in Australia (Abt et al., 2002). They have reported on world and continental championships that take place over a shorter time period: FIFA World Cup tournaments
(Armatas, Yiannakos & Sileloglou, 2007; Breen et al., 2006; Grant et al., 1998; Taylor et al., 2002; Werlayne, 2013a) and European Championships (Werlayne, 2013b), which are completed in less than four weeks.
It has been established that more goals are scored in the second halves of World
Cup matches (Armatas, et al., 2007; Grant et al., 1998; Werlayne, 2013a). In a comparative study of three World Cup tournaments between 1998 and 2006, the reported differences between goals scored in each half were not statistically significant (Armatas, et al., 2007).
Results of scoring frequencies are not always analyzed statistically with just the actual data reported (Breen et al., 2006; Taylor et al., 2002). Figure 2.1, on the next page, is an example of how the information is presented in the literature. GOAL SCORING PATTERNS 22
Percentage of goals scored in intervals of 15 minutes 30 25 20
% 15 WC 1998 10 WC 2002 5 0 0-15 16-30 31-45 46-60 61-75 76-90 ET Time (mins)
Figure 2.1 Percentage of goals scored in intervals of 15 minutes Source: Taylor, S., Ensum, J. & Williams, M. (2002:28)
Goals scored in time added on – that is after the 45th minute in the first half and after the 90th minute in the second half – are normally excluded from the calculations of goals scored in the fifteen-minute intervals. However, one study of goals over four seasons in the Australian National League (NSL) did not exclude goals scored in time added on
(Abt et al., 2002). They accessed data from web-based sources and it was not clear if a goal listed as being scored in the 47th minute was scored at the end of the first half or at the beginning of the second half. Consequently the goals scored later than the 45th minute were included in the second half category and goals scored after the 90th minute were included in the last fifteen-minute period, 75th–90th. The different methods of accounting for goals scored in extra time at the end of each half leads to inaccurate results when comparisons are made between the numbers of goals scored in different fifteen-minute periods of the game.
The Chi-square analysis of 703 matches revealed a significant difference (p<.001) between 1st and 2nd halves with a 34% increase in the frequency of goals scored in the 2nd GOAL SCORING PATTERNS 23
half, which is far greater than differences reported for tournaments of shorter durations.
The analysis indicated an upward tend of 10% for the fifteen minute intervals over time and 6% for the five-minute intervals (Abt et al., 2002). Similar upward trends were found in a study of 4,560 matches in four European Leagues (Giampietro et al., 2013) for the fifteen-minute intervals, which may indicate that the data base has to be considerably larger than one or more international tournaments involving between 64 and 172 matches before the gradual upward trend will be noticed (Armatas et al., 2007). A longitudinal study of every FIFA World Cup since 1930 showed the second half produced more goals than the first half, which is similar to all reported findings in football and with the highest number of goals (19.6%) recorded in the final fifteen minute period (Werlayne, 2013a).
Jinshan et al. (1993) reported that in the 1986 World Cup, the highest numbers of goals were scored during the 61st–75th minute period. That was not an isolated case because the same outcome was evident in the World Cups in 1930, 1934, 1970 and 1982.
However since the 1991 study, the only time more goals have been scored in the 61st–75th minute period than the final fifteen minutes was in the 2002 World Cup (Werlayne, 2013a).
In contrast to most of the research findings the most prolific period for goal scoring in the
European Championships in 2008 was between the 61st – 75th minutes and between the 46th
– 60th minutes in 2012 (Werlayne, 2013b), which is unusual compared to other reports produced by UEFA (2004; 2008).
The analysis of goals scored within intervals of three and five minutes has not been extensive or reported in recent studies. At the World Cup 2002, the ‘Golden’ goal rule was introduced when extra-time was played, to decide the outcome of the match after scores were level at the end of full time. It was noted that more than fifty percent of all goals were scored in the opening and closing 6–9 minutes of each half and time added on. It was suggested that teams would rather take more risks to win the game in the closing stages of GOAL SCORING PATTERNS 24
the matches than play extra-time when the ‘Golden’ goal rule would be applied (Taylor et al. 2002) .The rule was that the first team to score in extra-time won the match. It proved to be an unpopular method of deciding the outcome of a match and was discontinued.
Researchers have tried to explain why more goals are scored towards the end of the game and particularly in the last fifteen-minute period between the 76th and 90th minute.
The most frequent explanations include, more risk taking to score by the team that is losing, or for the team that may need to win when the scores are level (Abt et al., 2002, Taylor et al., 2002). This strategy would make the same team more vulnerable to concede a goal, thereby adding to the number of goals scored. The importance of winning specific matches applies particularly in international tournaments because winning rather than drawing determines the progress of the team to the next stage of the competition. Loss of concentration due to fatigue is seen as another contributing factor as well as differences in physical condition between certain players who maybe opposing each other on the field
(Abt et al., 2002, Armatas et al., 2007, Werlayne, 2013a). It is not mentioned in the research papers but the introduction of a substitute in the attack, late in the game, could create a situation where a fatigued defender may be exposed.
2.3 How goals are scored
More goals are scored from inside the penalty area and as one might expect the majority of goals are scored with the feet, with different parts of the foot being more dominant from time to time. Goals from using the inside of the foot and instep were equal in World Cup 1998 (Grant et al., 1998), goals scored with the laces dominated in WC 2002
(Taylor et al., 2002) while the instep was dominant in WC 1986 and 1990 (Jinshan et al.,
1993) with the inside of the foot being dominant again in 2006 (Breen et al., 2006), suggesting that accuracy is as important as power. Goals scored from headers average around twenty percent of the total but constitute approximately fifty percent of goals GOAL SCORING PATTERNS 25
scored from Set Plays, when penalties are excluded (Acar et al., 2009; Breen et al., 2006;
Grant et al., 1998; Jinshan et al., 1993; Taylor et al., 2002). (See Table 2.1)
Table 2.1 Techniques used to score goals in World Cup 2006
Source: Breen et al., 2006
The number of touches taken by the goal scorer is reported frequently in research and the figures have not changed much over several decades. Goals scored with one touch including headers have ranged between 54% and 78% and if one and two touch goals are combined the range is between 76% and 90% (Breen et al., 2006; Grant et al., 1998; Olsen,
1986; Taylor et al., 2002). The percentage figures do not change much if a calculation is made for goals scored in Open Play because most goals from Set Plays are scored with one touch.
The change in dominance of the inside of the foot and the instep, to score goals, suggests that accuracy (using the inside of the foot) is as important as power, which is achieved by using the instep. However, there are two important considerations. Whenever the instep is recorded there is an assumption that the strike at goal was hit with power, but that is not always the case. The decision to use the instep or the inside of the foot will be influenced by the position of the player, relative to the goalkeeper and distance from goal, that is, if the player is in a central position there may be more emphasis on accuracy and placement compared with a shooting opportunity to the side of the goal where the emphasis may be on power. GOAL SCORING PATTERNS 26
2.4 The pitch location of the goal scorer
The majority of goals, between 80–90%, are scored from inside the penalty area according to analysis of FIFA World Cups, UEFA European Championships and English
Premier League games (Breen et al., 2006; Grant et al., 1998; Taylor et al., 2002; Wright et al., 2011). Some researchers have excluded goals from penalty kicks while others have included them (Mitrotasios & Armatas, 2014; Olsen, 1986; Wright et al., 2011; Yiannakos
& Armatas, 2006). It is evident that within the penalty area there are key areas from where goals are scored. The majority of goals are scored from a central position within the width of the goal area, from the goal line to the edge of the penalty area. This area is divided further into three equal sections measuring six yards. The first section is between the goal line and the goal-area (28.4% in the example below), the second from the edge of the goal area to the penalty spot (31.3%) and the third is from the penalty spot to the edge of the penalty area (10.5%) as shown in Figure 2.2.
Figure 2.2 Areas on the field from where goals were scored Source: Breen et al., 2006
The example (Figure 2.2) from the 2006 World Cup is typical of where the majority of goals are scored inside the penalty area. In studies of goals scored in World GOAL SCORING PATTERNS 27
Cups, it has been reported consistently that more goals were scored in the space between the goal area and the penalty spot, 31.3% in this 2006 example, than any other area (Bell-
Walker et al., 2006). This area is referred to as the ‘Prime Target Area’ for players who cross the ball from outside the penalty area (Hughes, 1990).
It is logical that the most effective pitch position of the scorer, inside the penalty area, is central rather than to one side of the goal area. The fact that up to 20% of goals are scored from shots outside the penalty area makes play highly predictable and why there is so much interest in what happens in the play leading to the final pass or cross into the penalty area. However, few attempts have been made to classify the types or length of passes that lead to goals, which may reflect the degree of difficulty associated with the task or it may indicate that other aspects of scoring are deemed more important.
2.5 The origin of ‘the final pass’
In order to identify the key areas of the field for regained possessions, attacking prowess and where final passes are made prior to goals being scored, some researchers have divided the playing area in to six equal areas from side to side (Grant & Williams,
1998), while other researchers have used eighteen areas or zones (Breen et al., 2006).
Some researchers define these areas as ‘equal’ in size (Figure 2.3), whilst others have areas of different sizes (Figure 2.4), GOAL SCORING PATTERNS 28
Figure 2.3 The 18 equal areas of the field Source: Breen et al., 2006
In Figure 2.3, the eighteen areas appear to be the same size and in some literature they are described as being equal but they are not. Figure 2.4 illustrates a different allocation of space in the eighteen areas. The central areas, C1–C6 (Figure 2.3) are the same areas as Zones 2, 5, 8, 11 and 14 in Figure 2.4. The central zones are the width of the penalty area, which is 44 yards, while the areas L1–L6 and R1–R6, the same as zones 3–18 and 1–16 are much narrower, between 14–18 yards in most cases, depending on the actual width of the field.
Figure 2.4 The 18 unequal zones on the field of play Source: http://www.footballscience.net/special-topics/performance-analysis/ GOAL SCORING PATTERNS 29
The area of the field where the final pass was made provides important information about the delivery of the ball into the penalty area, prior to a goal being scored. More goals are scored from inside the penalty area (C6, Zone 17), following a pass assist from the central area C5, or Zone 14, than any other area. The central areas in the attacking half of the field, C4, C5 and C6 account for between 66% and 73% of all events preceding goals.
(See Figure 2.5)
Figure 2.5 Areas on the field from where the final pass was played prior to a goal in Open Play Source: Breen et al., 2006
Passing is the most common event preceding goals followed by crossing. Other events do precede goals and include individual play, tackles or interceptions, rebounds and defensive errors but they are in a minority (Breen et al., 2006; Carling et al., 2005; Taylor et al., 2002); see Table 2.2.
GOAL SCORING PATTERNS 30
Table 2.2 Key events preceding goals in World Cup 2006
Source: Breen et al., 2006.
The 2010 FIFA Technical Report described events as: combination play; wing play; defence-splitting pass; diagonal ball into penalty area, which are a different way of describing actions that could be grouped as passing or crossing.
2.5.1 Area C5 or Zone 14
This particular area has received more attention than any other because it is important strategically in scoring goals. Information on the frequency of goals scored in several World Cup finals was presented in section 2.4. A study of France in the 1998
World Cup and European Championships in 2000 showed that more than half of their goals came from a pass assist in this area of the field (Horn et al., 2000). A detailed study of play in Zone 14 in ten English Premier League matches revealed that; teams had an average of thirty possessions per game, 73 per cent of all goals scored originated in Zone
14 and the average time in possession for an attempt at goal or scoring was 2.9 seconds
(Horn et al., 2002) . A more recent analysis of 12 matches in the English Championship, which is one division below the English Premier League, found the teams had fewer possessions per match in Zone 14, averaging 23 possessions, but this still accounted for
43% of all play in the final third. The study found that more goals were initiated in Zone GOAL SCORING PATTERNS 31
14 than any other area in the attacking half of the field, which is in line with other research findings (Rees et al., 2010).
Play in Zone 14 has to be quick; 64% of all possessions in Zone 14 lasted between
0.5 and 2.5 seconds. The average time for the 600 possessions in Zone 14 for the ten EPL matches was 2.9 seconds and more goals (58%) were scored from a one pass move (Horn et al., 2002). As Zone 14 is within shooting range it is to be expected that a player will not have much time on the ball before a defender will challenge or force the player into action.
More goals were scored from passes that went from Zone 14 into the penalty area (Zone
17) than in any other zone in the final third (Horn et al., 2002; Rees et al., 2010), which is not surprising as the pass would be made to another player closer to goal. One might assume that the players receiving a pass inside the penalty area would be there waiting but that may not have been possible because of the offside law. From a tactical perspective a limitation of the Horn et al. (2002) and Rees et al. (2010) studies is that the type of pass, either in the air or along the ground, is not recorded nor neither is the position of the player who received the pass. It would have been valuable to know if (a) the player receiving the pass was ahead of the ball or (b) the player was running forwards into the penalty area when the pass was made or (c) the player received the ball and took it past the last line of defence before scoring the goal. Passing and crossing are two key events that precede goal scoring.
2.5.2 The ‘Crossing’ areas
As might be expected the areas outside the ‘penalty area’ and to the sides of the field, L 5–6 and R 5–6 (Figure 2.3) were the areas, which supplied the majority of crosses for the goals with slightly more coming from the right side compared with the left (Breen et al., 2006). The reported number of goals scored from a cross in FIFA World Cups is not too varied, 18% in WC 98, 29% in WC 2002, 24% in WC 2006 and 26% in WC 2010 GOAL SCORING PATTERNS 32
(Breen et al., 2006; FIFA, 2010; Grant et al., 1998; Taylor et al., 2002). However there have been significant differences in reports on the 2012 European Championships ranging between 20% and 43% (UEFA, 2012; Mitrotasios & Armatas, 2014). Crossing is a term that may mean different things to different people depending on their interpretation of what constitutes a ‘cross’. This can lead to considerable confusion in the absence of a clear operational definition because in the absence of a video replay of an event, people will create an image in their own mind according to their interpretation of what a ‘cross’ is. For most people a ‘cross’ would have to be played from a position on the field, in the opponent’s half, outside the width of the penalty area. For others a ‘cross’ may be determined by the act of getting the ball in the air regardless of where it happens on the field. Breen et al. (2006:49), for example, propose:
“In 2006, more crosses came from the wide right areas R5 and R6, (n=59.1 percent) as well as the central area C6 (n=27.3 percent) compared to the wide left areas (n=9.1 percent)”; see Figure 2.6.
Williams (2003:35) clarified his opinion on the origin of crosses when he stated:
“In the past, crosses were mainly played into the penalty area from wide areas between the penalty area and the touchline or from ‘cut backs’ from the goal line; cross assists now tend to originate from all areas of the field. It appears that players are able to generate sufficient spin and swerve on the ball to deliver effective crosses even from central areas of the field.” GOAL SCORING PATTERNS 33
Figure 2.6 Origins of crosses leading to goals in Open Play in 2006 World Cup Source: Breen et al., 2006
Differences in the interpretation of what constitutes a ‘cross’ may account for the differences in the number of goals reported as scored from crosses in the 2012 European
Championships, where figures of 20% and 43% have been reported (UEFA, 2012;
Mitrotasios & Armatas, 2014). It is clear that operational definitions need to be included in research documents or results can be misleading or confusing for readers. When large differences are reported in studies of the same topic it creates doubt in the minds of coaches who read reports and research papers for information they might be able to use in a practical way with their teams. It is interesting to note that in the UEFA Technical Report
(UEFA, 2012) the term ‘cross from the wing’ is used in the analysis of goals scored but the
FIFA World Cup Technical Study Reports have used different descriptions: in 2002 the term ‘Passing run or breakthrough on the wing’ was used and from 2006 to 2014 the expression ‘wing play’ was used instead of ‘cross’ (FIFA, 2002; FIFA, 2006; FIFA, 2010;
FIFA, 2014). The FIFA, UEFA Technical Reports and literature mentioned in this section do not include an operational definition of ‘wing play’ or a ‘cross’. This makes the interpretation of data difficult and replication impossible. GOAL SCORING PATTERNS 34
2.6 The number of passes made prior to a goal scored
Charles Reep was the pioneer of football match analysis. His analysis of matches played between 1950 and 1967, mostly in the English Football League, demonstrated that the probability of scoring a goal reduced as more passes were made as a result of the defenders being more compact, closer to the ball and better positioned to intercept or tackle
(Reep & Benjamin, 1968). Reep and Benjamin proposed that the majority of goals were and would be scored with moves of three passes or less and that ‘chance’ played a major part in the outcome of games. The evidence of ‘chance’ occurrences and other predictable aspects of performance will be introduced at the appropriate juncture in this literature review.
A very clear definition of a ‘pass movement’ was provided in the research paper and a ‘zero pass’ move was said to occur after an interception of a pass or after a shot at goal was taken without a preceding pass, for example, from a penalty kick. This includes shooting opportunities that happened randomly and as a result of opposing defenders failing to clear the ball effectively. The concept of ‘zero pass’ goals started with Reep &
Benjamin (1968) and with such a high percentage (36%–44%) reported in every aspect of the research, which included FIFA World Cups between 1958 and 1966, it justified his claim that ‘chance’ plays a large part in goal scoring opportunities. The research also showed that passing moves that ‘broke down’ in a team’s own half contributed to approximately 50% of goals conceded. This fact combined with the supporting evidence that most goals were scored with three passes or less, led to the concept of ‘Direct Play’, the merits of which are debated continually.
Charles Hughes was the next researcher to advocate that ‘Direct Play’ was the key to scoring more goals. Hughes was a football coach of some repute, who was the Director of Coaching at The Football Association in England for many years. In the 1980s he GOAL SCORING PATTERNS 35
conducted considerable research into goals scored in 109 matches involving every national team that had won the World Cup, successful club teams in Europe and England’s youth team. Based on the evidence from the research, The Football Association, under the direction of Charles Hughes, produced a series of video-tapes and a book called ‘The
Winning Formula’ (Hughes, 1990).
Hughes showed in the study of 109 matches that 87 per cent of all goals were scored with 5 passes or less. The point where passing moves were classified as
‘Possession-based’ rather than ‘Direct Play’ was after 5 passes had been made. The distinction was based on the size of the decrease in the number of goals after one more pass in the sequence. When 6 passes were made in a sequence of play the percentage decrement between goals from 5 passes and 6 passes was greater than 50 per cent for the first time (57%) and the number of goals scored after 7 or more passes never exceeded the number of goals scored after moves of 6 passes. (See Table 2.3.)
Table 2.3 Goals from sequences between 0 to12 passes
Source: Hughes, C. 1990
Hughes made the point that of the 202 goals in the analysis of 109 matches only 12 per cent, 26 goals were scored with passing moves of six passes or more or as a result of
‘possession-based’ football. Hughes suggested that if a team tried to make six or more passes whenever it gained possession of the ball it would greatly reduce its chances of winning and that any strategy based on long passing movements was in complete contrast to the evidence, which showed that seven out of every eight goals were scored with moves GOAL SCORING PATTERNS 36
of five passes or less. Hughes supported his case for ‘Direct Play’ by referring to teams who are recognized for playing ‘Possession-based’ football. One example he gave was in the 1970 World Cup final when Brazil beat Italy 4–1, another proponent of possession football, four of the five goals were scored with five passes or less.
Hughes reported that in his study, 26 per cent of goals (53 out of 202) were from
‘zero’ pass moves and asked the question, “Could it be that, after movements featuring a high number of consecutive passes, there was a rebound, a partial interception by a defender or a free kick which led to a ‘zero’ pass goal being recorded?”
To answer the question, Hughes re-analyzed the number of passes involved in the moves that led to a ‘zero’ pass goal. The analysis showed that the percentage of ‘zero’ pass goals involving five or less passes was virtually the same as the original analysis of all goals, that is 88 per cent were from moves of five passes or less.
This evidence was powerful and supported his argument for a philosophy of ‘Direct
Play’. A question that has to be asked is, ‘In light of the overwhelming evidence to support the theory of ‘Direct Play’, why would anybody advocate making long passing sequences?’
A possible answer might be that keeping possession of the ball is one way of getting the ball and a number of players within striking range of the opponent’s goal. Another question would be, ‘Is it possible that Charles Hughes identified what happens in football, that is, more goals are scored with five passes or less, regardless of whether a team has a
‘Direct Play’ or ‘Possession-based’ philosophy?”
A criticism of the evidence Hughes produced in ‘The Winning Formula’ is that he did not make a clear distinction between the number of goals scored from Set Plays and goals scored in Open Play. The major thrust of his argument for playing ‘Direct’ football was the difference in the number of goals scored from moves of five passes or less, compared with moves of six passes or more, which he classified as ‘Possession-based’ GOAL SCORING PATTERNS 37
football. If the goals scored from Set Plays are taken out of the total goals scored, a more realistic comparison may be made between the number of goals scored with five passes or less and six passes or more, which was Hughes’ criterion. In his analysis of 109 matches a total of 202 goals were scored; 92 were from Set Plays, which means a total of 110 goals,
54 per cent were from Open Play. The number of goals scored with five passes or less was
84, which leaves 26 goals scored with six passes or more. The figure of 84 represents 76 per cent of the total goals in Open Play, which equates to approximately three goals out of every four; slightly less than the seven out of eight goals stated by Hughes when goals from Set Plays were included in the calculations. A further criticism would be that the number of goals is quite small and the matches were selected randomly, they were not from one particular event, such as a FIFA World Cup or a league championship from one of the top footballing nations.
It is questionable whether goals from Set Plays should be included when advocating a style of play based on the number of goals scored from sequences of five passes or six passes or more. Goals scored from penalty kicks or direct free kicks are the result of shots at goal not passing moves and it is debatable whether a corner kick would be classified as a passing move. The point about excluding goals from Set Plays is valid because the inference to Open Play is quite clear in Hughes’ statement:
“If there is a Winning Formula in soccer it must be based on the attacking strategy of direct play, that is on playing the ball forward whenever possible with the aim of achieving a shooting opportunity within five consecutive passes” (Hughes, 1990:173).
However, it must be noted that Hughes provided a ‘Winning Formula’ that was based on more than just trying to score with passing moves of five passes or less. The
‘Winning Formula’ included strategies for attacking and defending play and had to be viewed as total package of ideas, based on logical deduction and evidence. GOAL SCORING PATTERNS 38
Richard Bate supported the concept of ‘Direct Play’ and in his analysis of 106 goals from the 1982 World Cup showed that 84 goals came from moves of four passes or less, which represented 79 per cent of goals scored (Bate, 1988). Bate did not provide any reason for calculating the number of shots at goal or goals scored from four passes or less rather than five passes or less. He provided figures for the number of goals scored and the number of passes made so it was possible to calculate that 92 goals were scored with moves of five passes or less, which represented 98 per cent of all goals scored. The figure of 98 per cent, for goals scored with five passes or less is considerably higher than the 87 per cent quoted by Charles Hughes in his analysis of 109 matches. Bate reported that 49 goals, representing 48% of all goals, were scored from a ‘zero’ or one pass move.
Bate referred to Hughes’ analysis of ‘zero’ pass and one pass moves to ascertain the number of passes in the final passing sequence prior to goals in ‘zero’ and one pass categories and used a sample of eleven matches played by different English national teams.
Interestingly, the results showed that 51 of the 52 goals in the eleven matches, came from moves of 5 passes or less and of the 19 ‘zero’ pass goals, 17 of them had moves of 5 passes or less prior to possession being lost. Bate did not give a reason why the ‘zero’ pass goals from the 1982 World Cup were not analyzed and maybe the reference to international matches played by England was meant to infer the same outcomes could be expected. Bate also showed data from goals scored in the 1985–86 season by Notts County FC in the
English 3rd Division where 37 of 71 goals scored during the season were video recorded and analyzed. No reason was given why only 37 goals were recorded and analyzed but it may have represented the goals scored at home matches when the club would have been able to conduct its own research. The data showed that all 37 goals were scored with moves of 5 passes or less. To emphasize the importance of Set Plays, Bate showed they accounted for 33 percent of the 37 goals analyzed and 33 of the 71 goals, 46 per cent, GOAL SCORING PATTERNS 39
during the season. The reference to the importance of Set Plays is linked to the area of the field where possession is regained and this topic will be dealt with in the next section. Bate used evidence from Hughes’ analysis of sixteen England matches to show the number of passes that were made before a shot was taken and the number of goals scored. Passing movements of five or less produced 324 of the 354 shots taken and 51 of the 52 goals scored. Passing movements of between six and eleven produced 30 of the 354 shots taken and 1 of the 52 goals scored.
The evidence produced by Bate prompted him to conclude:
“So-called ‘possession football’ (the retention of ball possession at all costs) involving a high number of consecutive passes is not as effective playing method in terms of creating shooting chances and goals and is not supported by research conducted across a broad cross-section of soccer played at Senior and Youth levels” (Bate, 1988:298).
In contrast to Hughes, Bate stated that ‘zero’ and 1 pass moves included goals scored from Set Plays, although he failed to highlight how many were actually scored in the 1982 World Cup and in the evidence produced on the eleven England matches. Bate provided more detail than Hughes did in the ‘Winning Formula’ in that he produced data on the number of shots taken, the number of passes in the moves that led to the shots being taken and the number of goals scored in each category of passing moves. The evidence was compelling to support the concept of ‘Direct Play’ albeit from a small sample of matches involving Notts County FC and England teams at different levels, but mostly from the senior team.
Researchers can have different interpretation of published results. Hughes and
Franks (2005) proposed that the accepted fact that ‘80 per cent of goals resulted from a sequence of three passes or less’ may be misinterpreted. Their point was based on the mathematical principle of ‘normalizing’ the outcomes when unequal frequencies of events are compared. The process of ‘normalizing’ is done by dividing the number of outcomes GOAL SCORING PATTERNS 40
by the frequency of the events, a technique not generally applied to performance analysis data (Hughes & Franks, 2005). In their analysis of passing sequences preceding all goals in the 116 matches played in two FIFA World Cups in 1990 and 19994, their data showed that 80 per cent of goals were scored with moves of four passes or less, which was in agreement with earlier studies (Hughes, 1990; Reep & Benjamin, 1968). However, when the number of goals scored in each team possession was divided by the frequency of that particular sequence length, there was a higher conversion rate in the longer passing sequence of six passes, per 1000 possessions. This analysis showed there may be other interpretations of Reep and Benjamin’s (1968) data and suggested that teams that have the skill to sustain longer passing sequences have a better likelihood of scoring.
Hughes and Franks (2005) examined the frequency of shots to passing sequence length and found it to be similar to the profile obtained for goals, with 80 per cent of the shots in World Cup 1990 occurring with passing sequences of four or less and 77 per cent for the corresponding data in the World Cup 1994. When these data were normalized the sequence length of five passes produced the highest number of shots, which was similar to the passing sequence length of six that produced the highest number of goals per 1000 possessions. (See Figure 2.7.) GOAL SCORING PATTERNS 41
Figure 2.7 Analysis of the mean number of goals scored per 1000 possessions for the 1990 and 1994 World Cups Source: Hughes and Franks, 2005
However the peak for shots per passing sequence for each World Cups (1990 and
1994) occurred at seven and four passes respectively, making the latter equivalent to a decrement of 43 per cent between 1990 and 1994. If the trend in football is more towards possession based football as the years pass by, one might have expected the results in 1990 and 1994 to be reversed.
Hughes and Franks (2005) normalized the conversion ratios of shots to goals per passing sequence too. The mean for sequences of zero to four passes per possession, for the total goals in both World Cups was 9.5 shots, which was almost identical to the figure reported by Reep and Benjamin (1968). The mean for sequences of five to eight passes was 15.1 shots, which was significantly higher. The normalized data suggest that longer passing sequences produce more shots than shorter passing sequences but the conversion ratio of shots to goals is better for shorter passing sequences. If the analysis had been done on goals scored in Open Play the overall percentage of goals from passing sequences of five or six passes or more would have been higher, which would have supported the evidence in favor of longer passing sequences, but no distinction was made between goals GOAL SCORING PATTERNS 42
scored from Set Plays or in Open Play. Of the 244 goals in the study 63 goals (26 per cent) were classified as ‘zero’ pass goals many of which, would have been scored from Set Plays.
In studies of passing sequences and possession based football it would be appropriate to discard goals scored from Set Plays because their inclusion will affect the results adversely.
A further weakness in this paper is that it only included 244 of the 256 goals that were scored in the FIFA World Cups of 1990 and 1994. It was also reported that in 1994 the number of teams increased from 24 to 32 with a subsequent increase in the number of matches, when in fact the number of teams in the FIFA World Cup Finals did not increase from 24 to 32 until France 1998.
The authors stated, “The difficulty facing coaches is to determine the skill of their team and then the appropriate tactics, but it would seem that the principle of direct play is only applicable where the skill of the team is insufficient to sustain possession of the ball”
(Hughes & Franks, 2005:513). This statement does not recognize that the dynamics of the game do not always demand a long passing sequence to score a goal. Many goals are scored from regained possessions close to goal, which do not require a pass to be made or may only require one pass to convert the opportunity into a goal. To do otherwise would be considered poor play.
In a study on goals scored in the EPL, data were normalized for the shots to goals ratio for passing sequences of less than four and for passing sequences of five to eight
(Wright et al., 2011). The findings were the same as results from Hughes and Franks
(2005) with a ratio of 9.52 goals per 1000 possessions for four passes or less, but the opposite to Hughes and Franks for passing sequences of five to eight, 7.28 compared with
15.1 goals per 1000 possessions. This may reflect the difference between results in a domestic professional league, which is played over a period of nine months, compared with the results from a World Cup tournament, which is played over a period of four weeks. GOAL SCORING PATTERNS 43
While it is interesting to use different analytical techniques to provide alternative interpretations of data, the fact is the majority of goals were scored with four passes or less in the 1990 and 1994 FIFA World Cups. Similar findings were reported in 1998 (Grant et al., 1998) where 75 per cent of goals in Open Play were scored with moves of four or less passes. The trend to isolate goals scored in Open Play continued in the research literature for the FIFA World Cups in 2002 and 2006, where 75 and 56 per cent of goals respectively, were from moves of four passes or less (Breen et al., 2006; Taylor et al., 2002); see Figure
2.8.
Figure 2.8 Number of passes preceding goals in 2002 and 2006 World Cups Source: Breen et al., 2006
Breen et al. (2006) acknowledge the importance of normalizing data. Their research highlighted the number of goals from passing sequences of five passes or more had increased from 26 per cent in 1998, to 34 per cent in 2002 with a high of 43 per cent in
2006, providing evidence for advocates of possession based football.
Analysis of goals in the European Championships in 2000 showed that 80 per cent of goals came from moves of four passes or less, which is consistent with other research findings (Hook & Hughes, 2001). Further and more extensive research of the European
Championships in 2004 showed similar results with 84 per cent of goals coming from GOAL SCORING PATTERNS 44
moves of four passes or less with 70 per cent coming from moves of two passes or less.
The latter study (Hughes & Snook, 2006) examined the differences between three categories of teams in the 2004 Championship and used the average figures for all teams, groups of teams and individual teams to make comparisons. The overall average for the shots to goal ratio was 1:12.7, which is higher than overall figures reported in earlier research (Reep & Benjamin, 1968) and higher than the shots to goal ratio for goals scored with four passes or less. The average number of passes per shooting possession was 2.69.
The figures were calculated by dividing the total number of passes made in possessions that finished with a shot at goal, for example, 74.5 passes divided by 28.36 possessions, equals 2.69 passes per possession.
An average score for passes can be misleading if a large section of data includes possessions where a pass was not made. In this study, ‘loose ball’ possessions accounted for 42 per cent of all shooting possessions in a number of these there would not have been any passes before a shot was taken. If Set Plays were also included in the total of shooting possessions, they would have lowered the average figure even further because most Set
Plays are from direct shots or one pass moves. This clarity of information was not provided, which leads the reader to question the results. When research is conducted and possession is the variable, a case could be made for not including possessions when a shot is taken and a pass is not made. Jones et al. (2004:101), stated:
“In the current study we excluded possessions less than 3 seconds in duration because they typically involved events such as tackles or clearances, which seemed to occur randomly irrespective of the team playing and as such do not contribute to an understanding of strategic possession”.
The importance of an operational definition to clarify when a team is deemed to be in possession of the ball or not, is critical when the number of passes in a sequence is used to make calculations and when quantifying the total number of possessions. Tackles and GOAL SCORING PATTERNS 45
clearances may not be counted as possessions by some researchers because the definition of possession implies having sufficient control over the ball to make a pass to a teammate
(Reep & Benjamin, 1968). When the data were normalized the higher passing sequences produced more shots at goal than lower passing sequences, which has been a consistent finding with research by Hughes and Franks (2005).
There has been more agreement than disagreement about the lower number of passes preceding goals, but it is not universal. In a study involving Norwegian professional clubs it was reported that longer passing sequences of five or more were more effective than passing sequences of two or less and particularly when a team had ‘imbalance’ in its defense (Tenga, Holme, Ronglan & Bahr, 2010). This is slightly confusing because in most research papers the notion of ‘longer passing sequences’ is attributed to five, or six, or more passes and shorter passing sequences are reported as less than four, or five passes. To exclude the number of goals from passing sequences of three and four passes, when comparing outcomes of passing movements of different lengths, is potentially misleading and to qualify the type of defense as balanced or imbalanced makes the interpretation even more difficult.
2.7 Regained possessions – the areas of the field
Reep and Benjamin (1968) identified the advantage of regaining possession of the ball in the vicinity of the opponent’s goal. They divided the field into four quarters laterally and the two quarters containing the goals were referred to as the ’shooting areas’.
The analysis showed that approximately 50 per cent of goals came from passing moves that originated in the ‘shooting area’, presumably they would have included free-kicks, corners and throw-ins and 50 per cent of those came from regained possessions as a result of poor clearances by opposing players. The return for regaining possession in the
‘shooting area’ was high with 30 per cent resulting in a shot at goal. They showed that GOAL SCORING PATTERNS 46
moves originating in the ‘shooting area’, which is comparable with the attacking third in modern parlance, accounted for one third of all attacks that reached that area of the field.
In contrast they showed that the teams in the analysis conceded 50 per cent of their goals from losing possession of the ball in their own half of the field.
Hughes advocated the defensive strategy of regaining possession in the attacking third of the field, when players were not outnumbered and based this philosophy on the evidence he produced (Hughes, 1990). The analysis showed the ratio of goals to regained possessions in the attacking third was 1:34 and accounted for 106 of the 202 goals in the study of 109 matches, which represents 52 per cent of all goals. The ratio of goals from regained possessions in the middle third was 1:147, which accounted for 60 of the 202 goals and 30 per cent of the total. The figure for the defending third was 1 goal for every
235 possessions, 36 of the 202 goals and 18 per cent of all goals; see Figure 2.9.
Figure 2.9 Ratio of goals from regained possessions in each third of the field Source: Hughes, C. 1990
GOAL SCORING PATTERNS 47
Based on this evidence Hughes proposed the defensive strategy of attempting to win back the ball as soon as possible and as near as possible to the opponents’ goal. The figures Hughes produced were hard to refute but after close examination of the evidence a different interpretation emerged.
Hughes’ argument for trying to win the ball back as quickly as possible and as close as possible to the opponents’ goal would have been sound if all regained possessions had been in Open Play, but they were not. Of the 202 goals analyzed in the study of 109 matches, 92 goals were scored from Set Plays, which would most likely have been regained in the attacking third. Set Plays include penalty kicks, corners, throw-ins and free- kicks, which are a result of fouls, saves by the goalkeeper and challenges by the opposition and not a result of a defensive strategy to win back the ball. One might argue that some throw-ins may be conceded as a result of pressurizing defenders when they have the ball but that would represent a small number and maybe none at all in any sample of matches.
If the 92 goals from Set Plays are deducted from the total goals scored it leaves a figure of
110 goals in Open Play, when the strategy of trying to win back the ball would be applied.
The analysis showed that 60 goals were from regained possession in the middle third and
36 goals were from regained possessions in the defensive third, which leaves a total of 14 goals from regained possessions in the attacking third. If the percentages for goals are recalculated it shows that 13 per cent of goals scored would have been from regained possessions in the attacking third, 54 per cent from the middle third and 33 per cent from the defending third. This slightly different but realistic interpretation of the data portrays a very different set of circumstances that might alter one’s strategic thinking. The percentage of goals from regained possession in the attacking third drops from 52 to 13 percent when goals from Set Plays are taken out of the calculations. This different interpretation of
Hughes’ evidence highlights the importance of knowing if goals from Set Plays are GOAL SCORING PATTERNS 48
included in any analysis of goals scored. Analysis by Bate (1988) supported Hughes’ work but he included the goals scored from Set Plays in the attacking third when making calculations. He stated that 50–70% of all movements leading to shots and goals originate in the attacking third of the field as a result of Set Plays and regained possessions. In the analysis of seven senior international matches involving England 62 shots and 14 goals came from 291 regained possessions in the attacking third, including Set Plays. In contrast, from 411 regained possessions in the defending third only 122 made the attacking third, which produced 14 shots at goal and no goals. These facts supported his philosophy of trying to regain possession of the ball as early as possible and as close as possible to the opponents’ goal. Charles Hughes and Richard Bate provided sound arguments for keeping players compact in the opponents half of the field as a result of ‘Direct Play’ and quoted high figures of goals from regained possessions in the attacking third, but both failed to acknowledge the number of regained possessions that were a result of pressing opponents or from poor clearances by the opposing players in Open Play.
The interest in where teams regain possession of the ball has continued because the outcomes may reflect the tactical strategies of teams in terms of where they try to win the ball as well as their ability to keep possession if it is regained in a team’s own half or defending third. A study of goals in the 1998 World Cup and 1997–98 EPL season showed that in the EPL 14.3 % of goals in Open Play were from possessions regained in the defending third compared with 29.6 % in the World Cup. The number of goals from regained possessions in Open Play in the attacking third was 50% in the EPL compared with 22.2% in the World Cup (Grant et al., 1999). This analysis highlights the importance of comparing the outcomes in different types of football, i.e. league football and international football. The difference in figures for goals from regained possession in the attacking and defending thirds of the field reflects several factors, (a) The style of football, GOAL SCORING PATTERNS 49
(b) the duration of different competitions and (c) the potential impact of the environment on the playing style. Teams in the World Cup have the very best players from each nation and one would expect the most skillful. As a result the teams are better equipped to withstand pressure from opponents, which in hot conditions would discourage opposing teams from pressing high up the field at great cost to the energy systems of the body. In the
EPL, which is played over a period of nine months and not in summer, the intensity of matches and effort to regain possession in the attacking third may be higher due to the lower temperature when most matches are played. Each team in the EPL plays 38 matches so risk taking is tolerated more than in the World Cup competition, which is knock out after the first three group matches and the finalists only play a total of 7 matches. In the
EPL there are more teams for talent to be distributed, which may result in a lower standard of player in club teams compared with international teams. This may account in some part for the lower number of goals from regained possession in the defending third of the field plus the playing style of some teams in the league, where playing out from the back may be deemed too risky, so players choose to pass the ball quickly up-field with long passes when under pressure from opponents. When comparisons are made between World Cup finals and EPL seasons it is worth noting that in World Cups the number of goals scored in
Open Play may range between 80 and 110, whereas in the EPL the number would be between 600 and 800 goals.
Other studies of goals scored in World Cups from regained possessions in the three thirds of the field have reported similar results. For example, the number of goals from regained possessions in the defending third of the field in the 1998, 2002 and 2006 World
Cup finals was 29.6, 29 and 34.4 per cent respectively (Breen et al., 2006; Grant et al.,
1998; Taylor et al., 2002); see Table 2.4.
GOAL SCORING PATTERNS 50
Table 2.4 Goals from regained possessions in each third of the field in World Cup 2006
Note: Defending Areas 1 and 2 combined represents the Back Third, Areas 3 and 4 represent the Middle Third and Areas 5 and 6 represent the Final Third Source: Breen et al., 2006
In each of the three World Cups, more goals were scored in Open Play from regained possessions in the defending third than in the attacking third and more than 50 per cent from regained possessions in the defending half of the field. This evidence highlights the importance of being able to play the ball forwards and keeping possession once it has been regained. What the evidence does not provide is the number of passes that were made for each goal scored in each third or each half of the field and by which teams. The reports from World Cup Finals have shown that the majority of goals regardless of where possession was regained have been scored after passing sequences of between zero and four passes. This information alone shows that many goals must be scored from quick forward transitions because over 50 per cent of possessions have originated in the defending half of the field (Breen et al., 2006; Grant et al., 1998; Taylor et al., 2002).
2.8 Time in possession
This aspect of analysis is linked closely to the number of passes preceding goals scored and is influenced strongly by the definition of possession, which is often absent in published reports and journal articles. Time in possession is not reported in many recent studies (Bekris, Gioldasis, Gissis, Komsis & Alipasali, 2014; Werlayne, 2013b) so the data available may not be a true reflection on current football played at the elite level. The GOAL SCORING PATTERNS 51
majority of papers report goals scored between 0–5 seconds, 6–10, 11–15, 16–20, 21–25 and onwards. In studies of goals in Open Play in World Cups from 1998 to 2006, the highest number of goals, 25% – 31% were scored within 6–10 seconds. When categories are combined, the number of goals scored with possessions lasting between 6–15 seconds is consistently higher (45–55%) than goals scored between 0–10 seconds (36–49%) but the difference is not great (Grant et al., 1998; Taylor et al., 2002).
France and Brazil scored the majority of their goals to win the 1998 and 2002 FIFA
World Cups respectively with possessions lasting between 6–15 seconds, which is in agreement with the averages for all teams in the competitions. France scored 24% of their goals with possessions lasting between 16–20 seconds, which was considerably higher than the figure of 4%, which was recorded for Brazil and the overall average in 2002 and
2006 (Breen et al., 2006; Horn et al., 2002; Taylor & Williams, 2002). Another study of goals in the 2006 World Cup reported that 61% of all goals, excluding penalties, were scored with possessions lasting between 0–10 seconds, which is considerably higher than figures in other studies on the World Cups (Acar et al., 2009).
In all of the studies on time in possession it is clear that no matter which single or pair of time frames is used, 0–5, 0–10 or 6–15 seconds, the fact is that most goals are scored in under fifteen seconds regardless of where possession is regained. There are reported discrepancies, which may be explained by the absence of operational definitions.
Acar et al. (2009) reported a figure of 61% for goals scored between 0–10 seconds in the
2006 World Cup compared with Breen et al. (2006), who reported 36%. The difference in this example is most likely because Breen et al. excluded all Set Plays, which included penalties, while Acar et al. only excluded penalties, so goals from free kicks and corners would have been included; see Table 2.5.
GOAL SCORING PATTERNS 52
Table 2.5 Time in possession preceding goals in World Cup 2006
Source: Breen et al., 2006
Analysis of time in possession prior to a goal being scored in domestic league football has not been reported widely. An early study in 1998 compared the time in possession of the ball in the English Premier League (EPL) in the 1997-98 season (Grant et al., 1999) with the time in possession of goals scored in the 1998 World Cup (Grant et al.,
1998). The percentage of goals scored in the EPL between 0-5 seconds (55.7%) was nearly three times higher than goals scored in the World Cup in the same period (19.5%). This is a substantial difference, which may be explained by the different styles of play seen in
EPL clubs at the time and the teams at the top level of international football. Another explanation might be the nature and timing of the competitions, which are very different.
The EPL is played over ten months of the year in a cool climate, while the World Cup is invariably played in hot conditions in less than four weeks. The cool conditions and longer recovery times between matches may have encouraged a more robust, high tempo and direct style of play in the EPL compared with the hot conditions and shorter recovery times between matches at the World Cup, which would tend to lead to a slower tempo in matches and a more patient approach to scoring goals. The number of goals scored in the World GOAL SCORING PATTERNS 53
Cup with possessions in excess of 12 seconds (54%), compared with the EPL (18.6%) would support this theory. An interesting point to consider is that if 54% of goals scored in the World Cup were with possessions over 12 seconds, it would be logical to expect more goals to be scored with sequences of more than four or five passes. It was reported that the highest frequency of goals from open play came from three and four pass moves (Grant et al., 1999). On that basis the tempo of play had to be quite slow allowing the player in possession plenty of time on the ball to either run with it or have time to decide what to do with it. One might assume wrongly that goals scored with less than four passes would come from regained possession closer to their opponent’s goal. In the 1998 World Cup
57.4% of goals came from regained possessions in a team’s own half of midfield or defending third (Grant et al., 1999).
A study of 104 goals in Open Play from 44 matches involving five top European club teams from different countries, recorded considerably higher figures for four of the teams (50% –64%) for the number of goals scored with possessions between 0-10 seconds, with Paris St Germain (PSG), as high as 80% (Garganta et al., 1995). These figures are considerably higher than those reported in studies of World Cups but are mostly comparable with the study in the EPL in 1998, referred to earlier. When there is one recorded figure that stands out from all other reported data, it may be explained by the sample size and how the data is reported or it might just be an isolated case. In the example of PSG recording 80% of goals in Open Play being scored in 0-10 seconds, the number of goals scored is not stated but the analysis was done on five matches, which is a small sample (Garganta et al., 1995).
2.8.1 Possession in 5 minute intervals
Analysis of time in possession has been extended beyond the final sequence of passes that lead to a goal. One study recorded the number and success of passes made by GOAL SCORING PATTERNS 54
both teams in the 5 minutes leading up to a goal and the 5 minutes that followed the goal
(Redwood-Brown, 2008). There were 285 goals from 120 EPL matches in 2004–05. If a goal did not have a full five minutes of possession, either before or after it was scored, it was excluded. The possession for both teams, in terms of number and success of passes, for each half of the match was scaled to be equivalent for five minutes of play.
Comparisons were made between the team that scored and the team that conceded for the 5 minutes before and after the goal, for the number and percentage of accurate passes, with the average for whichever half of the match the goal was scored. In the 5 minutes that preceded a goal, the scoring team played a significantly higher percentage (72.4% compared with 70.2%, p<.001) of passes accurately than for the average for the half the goal was scored. In the 5 minutes after the goal was scored the scoring team played a significantly lower percentage of passes (67.3% compared with 72.4%, p<.001) were played accurately than the average for the half the goal was scored (Redwood-Brown,
2008). Speculation to explain why this should happen included a lower level of intensity and concentration and the mindset to protect the lead if the team went ahead. For the scoring team there was a significant difference in the number of passes made compared with the average for the relevant half of the game (p<.001). The conceding teams were found to pass less often and less successfully in the five minutes before and after conceding a goal, compared to their average for that half of the match. There was a significant difference in the number of passes made in the 5 minutes preceding a goal (19.3 compared with 22.9, p<.001). The study examined how the variables (accuracy and number) of passing for both teams were affected for different scores during the games when the teams were winning, losing or drawing. The differences were reported at every score-line from 0–0 to 4–1 indicating the trends but showing there were exceptions at various score-lines. The score-line of change from 1–1 to 2–1 for the scoring team was the GOAL SCORING PATTERNS 55
only time the difference in the accuracy percentage of passes was significantly different
(p<.001) to the average for the half the goal was scored as indicated by the Bonferroni adjusted post hoc Wilcoxon signed rank test.
A similar study was conducted to analyze 121 goals from the 64 matches at the
2010 World Cup. In addition the time in possession by the scoring and conceding team, for the overall match and for five-minute periods before and after a goal was scored, was calculated for three areas of the field referred to as the Defending, Middle and Attacking
Third (Ridgewell, 2011). The results supported the work done by Redwood-Brown and showed the scoring team made significantly more passes, with a higher percentage of success in each third of the field than in the average five-minute period for the half of the match the goal was scored. In the five-minute periods after the goal was scored the scoring team made significantly fewer passes in all thirds of the field than the average for the five minutes of the half in which the goal was scored. The conceding team had significantly more possession in its own defending third before conceding when compared to the five minutes after a goal was scored, despite having less overall possession before conceding.
However the possession had by the conceding team in both the middle and attacking thirds was significantly less in the five minutes before they conceded when compared to both five-minute periods after they conceded and the average for the half.
Redwood-Brown’s study showed the scoring team made fewer passes in the five minutes before scoring than in the average five-minute period for the half, whereas the study by Ridgewell showed the teams made more passes in the same period compared with the average for the half. The difference in the number of passes made might be explained by the contrasting playing styles of international teams and club teams playing in different types of competitions. There were fewer significant differences found in the latter study GOAL SCORING PATTERNS 56
(Ridgewell, 2011) when individual score lines were looked at, which may be explained by the smaller sample of goals at the World Cup, 121 compared to 285 in the EPL.
Explanations have been offered as to why possession and the success rate of passing are generally worse, by the team that scores, for five minutes after the goal. The team that conceded the goal restarts the match and may want to keep possession of the ball rather than risk losing it and possibly conceding another goal very quickly; this would reduce both variables for the team that scored. After scoring, which often follows extended periods of intense effort and sustained pressure, the team might stay in it’s own half of the field and be willing to let the opponents have the ball, free of pressure, especially if the team that scored has taken the lead. The idea of protecting the lead is a real one in football.
However, evolving game score might encourage a team that has just equalized to press its opponent immediately after scoring in order to win back the ball and attempt to score again.
The evidence suggests that this does not happen or if it does, the pressing team does not regain possession of the ball sufficiently to increase possession. It is a fact that many goals are conceded within two minutes of a team scoring. There have been attempts to explain why this happens, particularly loss of concentration by the defending team, but it still happens on a regular basis. It is possible that teams do not actually practice what they should do immediately after scoring and only talk about it.
2.9 Counter attacks and transitions
All teams attack to score goals when they have possession of the ball. Where teams regain possession on the field is normally described as the defending, middle or attacking thirds of the field. Possession is referred to as happening in ‘Open Play’ or from ‘Set
Plays’, throw-ins and goal kicks. In the past, when a team regained the ball in its own half and attacked quickly the term, ‘counter attack’ was used to put that type of attack into a special category and it was ‘understood’ that it happened from within a team’s own half of GOAL SCORING PATTERNS 57
the field. The term ‘counter attack’ was not used to describe the change of possession in the opponent’s half of the field. ‘Transition’ is a more recent term to describe the transfer of possession from one team to another and when the team attacks quickly, regardless of where possession was regained, the expression ‘quick transition’ is often used. The result is that the terms ‘counter attack’ and ‘quick transition’ may mean the same thing or something quite different depending on the interpretation.
The number of goals reported to be scored from counter attacks, compared with organized attacks and set plays, is inconsistent, which may be a result of different operational definitions or the lack of them in the literature. Yiannakos and Armatas (2006) reported that 20.3% of goals were a result of counter attacks compared with 44.1% from organized attacks in European Championships 2004, while Mitrotasios and Armatas (2014) and Werlayne (2013b) reported 20% and 17% from counter attacks with 60% and 53% respectively coming from organized attacks in the European Championships in 2012.
Armatas, Yiannakos, Ampatis and Sileloglou (2005) reported that the frequency of counter attacks was low (4.9%) but that form of attack was more effective than organized attacks because the strike rate of scoring was 16.9% compared with 11.1% in organized attacks in high standard soccer games. While the strike rate might indicate more success the reality of translating those figures into the number of goals scored would encourage teams to use organized attacks more than counter attacks. These contrasting reports illustrate the difficulty of categorizing goals without operational definitions.
One study that did have an operational definition of a ‘transition’, examined the speed of transition, the number of players involved and where the team regained possession (Turner & Sayers, 2010). The authors analyzed the number of transitions one team had in 27 matches in the Australian ‘A’ League in 2009/10 season (N=1105,
Mean=41). The definition for a transition was that at least two passes had to be made or a GOAL SCORING PATTERNS 58
distance of 15 meters had to be covered by running with the ball. Once the data were analyzed the term ‘counter attack’ was used to refer to ‘transitions’ in the ‘short passing category (0–3 passes) that had the highest transition speed (4.9 m/s.). The results showed significant differences in mean transition speed for each pass sequence length; short (0–3 passes), medium (4–6 passes) and long (7+ passes) but no differences in mean transition speeds between positive and non-positive outcomes both overall and for each pass length.
The mean transition speed was greatest when possession was regained in the defensive third of the field (4.2m/s.) and the slowest in the attacking third (2.5m/s.). Anova testing showed significant differences in mean transition speed between all player groupings: 1–3 players (4.9m/s.), 4–6 players (3.2m/s.) and 7+ players (1.7m/s.). Standard residual analysis showed that transitions that ended centrally were more likely to have a positive outcome.
The residual analysis suggests that attacks with up to 3 players (R=2.0) and with moves of up to three passes will produce more positive outcomes (R=3.6) than moves with four or more players and passes. This was reported to be in line with other studies
(Garganta et al., 1995; Hughes & Franks, 2005). The point was made quite clearly that the speed of transition and number of players involved did not produce a significant difference in the number of positive or non-positive outcomes and that this was not supported by other studies. However, in the Garganta et al. study, for example, the suggestion that more positive outcomes would come from sequences of three passes or less was based on the comparison of goals scored with passing sequences of 4 or more passes and time in possession (less than 5 and 10 seconds); no attempt was made to count or define a non- positive outcome. Unfortunately in this study there was not a comparison of ‘transitions’ with the number of total possessions the team had or information to differentiate between positive outcomes from transition starting in the defending, middle or attacking thirds. GOAL SCORING PATTERNS 59
2.10 Characteristics of successful and unsuccessful teams
Success can be measured in different ways. Japan qualified for the first time in the
1998 World Cup and their aim, or measure of success, was to score at least one goal in one of the three group matches. Making progress from the Group Stage to the Knockout Stage
(final 16 teams) or reaching the Quarter Finals of a tournament or the World Cup Final itself are all measures of success. Inevitably the performance indicators of successful and unsuccessful teams are analyzed and compared to identify where the difference might be and so the less successful teams know where they must try to improve if they are to increase their chances of being more successful in the next tournament. This section will review different methods of comparing successful and unsuccessful teams.
Some studies have compared successful teams that reached the semi-finals of the
World Cup with teams that did not progress from the initial group stage of the tournament, described as the unsuccessful teams (Bell-Walker et al., 2006; Grant et al., 1998; Low et al., 2002). Comparisons have been made also on statistics for winners of the World Cup and the average figures for the rest of the teams in the tournament in a variety of variables, for example, the shot to goal ratio, or the number of goals scored from regained possessions in the defending third (Horn et al., 2000; Yates et al., 2006). Other studies have been conducted to identify statistics that discriminate between successful and unsuccessful teams, compare successful teams (top 3), with unsuccessful teams (bottom 3) in domestic leagues in Europe and seeded teams in the 2004 European Championships
(Castellano et al., 2012; Hughes & Franks, 2005; Hook & Hughes, 2001; Lago-Ballesteros
& Lago-Penas, 2010; Tenga & Sigmundstad, 2011).
When the results of successful teams are compared with unsuccessful teams it is not surprising the outcome is usually better for whichever variable is analyzed, for example, the shots to goal ratio, the total number of shots, or the conversion rate. One of GOAL SCORING PATTERNS 60
the weaknesses of comparing successful and unsuccessful teams in a tournament such as the FIFA World Cup is the sample size of matches; five for teams that reach the quarter finals and three for teams that do not progress beyond the group stage. Comparisons of teams that have won the World Cup show there are variations in the way teams play. For example, Italy conceded only two goals in 2006 because they were defensively strong and scored six of their ten goals in Open Play from passing sequences or four or less with only one goal from a cross (Yates et al., 2006). In contrast Brazil won the World Cup in 2002 and scored eighteen goals; twelve in Open Play with nine from passing sequences of five passes or less and seven from crosses. Although a definition of a ’cross’ is not provided in any of the referenced papers it was noted that four of the goals were from a cross slightly
‘infield’ and further out than the edge of the penalty area and only two of the seven goals from crosses between the penalty area and the bye line (Taylor & Williams, 2002). This type of analysis is interesting because it shows that teams with different characteristics in the way they play football can be successful, even if there are some similarities, for example, the number of goals from low passing sequences.
An alternative approach to comparing a successful team with an unsuccessful team was adopted by comparing the winners of the 2002 World Cup, with a team that failed to progress beyond the group stage but had been a previous World Cup winner in 1998
(Scoulding et al., 2002). The teams were compared for the type and length of pass, whether possession was lost or retained and whether the pass was made to feet or to space. There was no difference in passing ability between the two teams even though they differed markedly in terms of success in the tournament. It was suggested that either the criteria used were not sensitive enough to detect differences in passing or the teams were of a similar standard and other factors determined match outcomes, which was obviously the case. It could be argued that the definition of success based purely on the result of a match GOAL SCORING PATTERNS 61
may not be a true reflection of what might be deemed successful play or attacking strategy.
Other indicators including the ratio of shots to goals or the ratio of shots from possessions with shorter or longer passing sequences may provide information to discriminate between successful and unsuccessful teams.
Normalizing data to provide a more informed analysis of the frequency of events during a match has been strongly recommended (Hughes & Franks, 2005). In their analysis of passing sequences, shots and goals in the 1990 and 1994 World Cup finals they normalized the data and expressed the outcomes as goals per 1000 possessions. The mean for sequences of five to eight passes, referred to as possession play, was much higher at
15.1 shots per goal confirming the conversion rate was better for lower passing sequences between 0 and 4, referred to as direct play. When successful teams (eight quarter-finalists) were compared with unsuccessful teams (eight first round losers) in the 1990 World Cup, they had more shots at every sequence between 0 and 9 passes with one exception. The unsuccessful teams had more shots from sequences with 7 passes. The evidence shows that successful teams created more shots from passing sequences of 5 or more passes than from sequences between 0 and 4 passes but the difference was not statistically significant.
Hughes and Snook (2006) took a different approach and compared teams that were seeded for the European Championships by analyzing five matches played between teams in category A, (top four) B, (middle eight) and C (bottom four) so their hypotheses could be tested with teams from each category playing the same number of matches. In their extensive analysis they showed that teams in category A had the best shot to goal ratio of
1:8.4, category B had the 1:12.6 and category C teams had the worst at 1:17. All teams had average shooting possessions of less than five passes, with an overall average of 2.69 passes. Germany had the highest at 4.28 passes, while eventual winners Greece had an average of 1.37 passes. Greece was a category C team which had the lowest number of GOAL SCORING PATTERNS 62
shooting possessions when compared with category A and B teams. The overall number of passes per possession per game was 2.56, which is amazingly low. Category A teams had an average of 2.45 passes per possession, which is below the overall average of 2.56, while category B and C teams had an average of 2.93 and 2.12 passes, respectively. The fact that a category C team won the 2004 European Championships with very low figures in a number of performance indicators shows that in a low scoring game the difference between success and failure is hard to identify. It was reported that no significant differences were found between the three groups and at the elite level it is not possible to differentiate between ranked teams on variables such as, when and how possession is gained, the number of passes in possessions, the type or position of the assist or the type or location of shots at goal.
Hughes and Snook (2006) do state the total number of possessions per team per game, but they do not distinguish between Set Plays and Open Play. Some of the data presented in the paper raise some issues about how to record performance. For example,
England had 191 possessions in two games and made 549 passes, producing an average of
274 passes per game and 2.88 passes per possession. The average number of possessions per game is 95, which is quite low for any international match and would indicate a match played at a very low intensity. When a match is played at a high tempo, as a result of players working hard to win the ball, the number of possessions will go up because it is harder to keep the ball under pressure from opponents. During most matches with teams of a similar standard each team will have a similar numbers of possessions because when one team loses the ball the other team gains it. A team may have slightly more possessions if it has more free kicks or throw-ins than the other team because there will be consecutive possessions. There might be a considerable difference in the total time in possession between two teams in the same match but generally there is not much difference in the GOAL SCORING PATTERNS 63
total number of possessions. So if England had 95 possessions in a slow tempo game it is quite surprising that their average number of passes per possession was less than three. If we compare Sweden’s results with England’s it is a complete contrast but with a similar number of passes per possession. Sweden had 754 passes in 344 possessions, which equates to 172 possessions per game, nearly double the number of possessions for England, who averaged 95 possessions per game. Sweden had an average of 2.2 passes per possession, virtually the same as England who had an average of 2.88 and yet the game must have been played at the most frantic pace imaginable for a team to have 172 possessions. This is hard to explain and one has to question the interpretation of what constitutes ‘possession’.
In the conclusion to the Hughes and Snook (2006:73) paper, reference is made to
Jones et al. (2004), “that possessions under three seconds should possibly not be counted because many of them involve tackles and clearances, which seem to occur randomly”
(2004:101). By including the quote from Jones et al. (2004) and suggesting that possessions of less than three seconds might be left out of future research on possession,
Hughes and Snook (2006) imply that possessions involving clearances and tackles were included in the study or may have been in the matches involving Sweden. Certainly the high number of possessions for Sweden would suggest that. The range for the number of possessions per team, excluding Sweden, is between 95 (England) and 130 (Spain) with an average of 115. In matches with a lower number of possessions as in England’s (95) case, the ball is generally in play for a longer period than in a high tempo game when the total number of possessions would be higher. If an average time of 60 minutes is applied for when the ball is in play, the average time in possession for each team is 30 minutes, with an average of 19 seconds per possession for England. If the same criteria is applied to the figures produced for Sweden, that is, 172 possessions in 30 minutes, the average time for GOAL SCORING PATTERNS 64
each possession is 10 seconds but both teams had an average number of passes per possession of less than three. It is hard to understand how one team could have an average of 19 seconds per possession and have virtually the same number of passes, 2.2 compared to 2.8 per possession, as a team with an average of 10 seconds per possession.
Unfortunately the comparative figures of England and Sweden raise questions about interpretation of possession and the use of averages to compare the number of possession per game or number of passes per possession. In the 2006 FIFA World Cup most games had an actual playing time of less than 60 minutes (FIFA, 2006). The average number of passes per game in the 2004 European Championships was 304. The lowest average was by Latvia with 157 passes per game and the highest was Holland with an average of 447.
Category A teams had an average of 309 passes, Category B teams had an average of 339, well above the overall average and higher than Category A teams, while Category C teams averaged 248 passes per game, well below the overall average of 304. Interestingly,
Greece who won the title of European Champions had an average of 193 passes per game and was a Category C team. A weakness of this format of comparing teams is that each team only played between one and three games. The figures from one game may not be a true reflection of what happened in other matches and the figures from one game were used as an average, in Holland’s case. The average number of 304 passes per game is comparable with figures in the 1998 World Cup, where successful and unsuccessful teams averaged 362 and 308 passes per game, respectively (Grant et al., 1998). In comparison with the figures in the 2002 World Cup the average of 304 passes per game was higher than for successful and unsuccessful teams who recorded 271 and 252 respectively (Low et al., 2002).
A more recent study of successful and unsuccessful teams in Norwegian football compared the top three clubs, the bottom three clubs and the clubs in-between (Tenga & GOAL SCORING PATTERNS 65
Sigmundstad, 2011). The study included four possession characteristics; the duration, the number of passes, the area of the field where the attack started and the type of possession, described as ‘counter attack’ or ‘elaborate attack’. The counter attack referred to the speed and directness of the attack while the elaborate attack was used to describe attacks with a slower build up and safer passing, regardless of where the attack started. The mean number of goals scored for each variable was compared between the three table positions using a series of Kruskal Wallis H tests with P values of under 0.05 indicating significant table- position effects. Where a significant table-position effect was found, individual pairs of variables were compared between the two table-position groups using a series of
Bonferroni adjusted Mann-Whitney U tests, with P values of under 0.017 indicating significant differences. The results showed a significant difference between the three bottom teams and the teams in the two other groups, in the number of goals scored by
‘counter attack’ (p=.003), the number of goals scored with possessions of 12 seconds or more (p=.002), the number of goals from 4 passes or less (p=.003) and the number of goals starting in the middle third of the field (p=.001).
The authors reported that longer possessions with five passes or more failed to distinguish between successful and unsuccessful teams when the individual pairs of variables were compared, but there was a significant table position effect (p=.043) between the top three and bottom three teams. This was reported to be in disagreement with other studies where longer passing sequences were found to be more effective in goal scoring than shorter passing sequences (Hughes & Franks, 2005; Hughes & Snook, 2006).
However, the authors failed to mention that when longer passing sequences have been shown to be more effective than shorter passing sequences the data were normalized, which is a very important distinction that should have been made to avoid a misunderstanding. GOAL SCORING PATTERNS 66
If research is to provide information for coaches to use, it is more important to highlight performance indicators that show a significant difference between teams at the top and teams at the bottom when differentiating between successful and unsuccessful teams rather than differentiating between the group at the bottom and those in-between, some of whom, by definition, have to be very close to the teams at the top.
Another study that involved the comparison of the top three teams in the EPL with the bottom three teams calculated how much time each team had possession of the ball when the team was drawing, winning or losing during the match. A total of 24 matches were analyzed from the 2001–02 season (Jones et al., 2004). The results showed that successful teams had significantly longer possessions than unsuccessful teams, irrespective of match status. The study used a definition that possession was deemed to start when a player had sufficient control over the ball to enable a deliberate influence on its direction.
Possessions of less than three seconds were eliminated from the analysis because they consisted of insignificant events not thought to be indicative of a team’s strategy, for example, tackles and goalkeeper clearances.
Clearly the definition is not explicit enough to capture the concept of ‘possession’, when strategic play is considered. Possession should be linked to the idea of keeping the ball, not kicking the ball out of play or merely clearing the ball with a header or a kick.
Hughes and Snook (2006:73) suggested that, “possessions of less than 3 seconds could be removed from the results to see if there were any more differences in the possessions occurring”. It is particularly interesting that Jones et al. (2004:101) “excluded possessions of less than 3 seconds in duration” because they “do not contribute to an understanding of strategic possessions”. The average number of possessions per game in the 2004 European
Championships was 115, after Sweden’s high of 172 was taken out of calculations; the figures included possessions of less than 3 seconds. Jones et al. (2004) reported possession GOAL SCORING PATTERNS 67
totals between 201 and 262 in the analysis of the top and bottom three teams in the EPL in
2001–02 season, before possessions of less than 3 seconds were eliminated from the study.
Unfortunately, Hughes and Snook (2006) did not provide a definition of possession. The differences in the number of possessions between the EPL and the European
Championships may be a result of different styles of play, the difference in the number of matches played by the participating teams or by different interpretations of ‘possession’, but clearly a better definition of possession needs to be created.
It is interesting to review the criteria of success, the various performance indicators that have been used to differentiate between successful and unsuccessful teams and how different statistical techniques have been used to analyze data. Some performance indicators of successful teams are predictable, such as shots to goal ratio or the conversion rate, while ‘possession’ is still difficult to quantify and explain even though research shows that successful teams tend to have more of it overall and make more passes than unsuccessful teams, but not always. Quite clearly there is a need for all studies to include operational definitions, which also need to be specific to the research being undertaken.
For example, a definition for ‘possession’ may be different if the intention is to record the total number of times a team makes contact with the ball and with enough control to determine its direction, compared with a definition to measure the number of possessions a team has where the specific intention is to keep the ball in play by passing or running with the ball. Several studies have reported that successful teams tend to play more through the central areas of the field, pass more on the ground, use more penetrative passes between defenders and balls over the top (Hughes et al., 1987; Hughes & Churchill, 2005) but there is not one research paper that has examined the success rate of scoring goals from passes made behind the opposing defence. GOAL SCORING PATTERNS 68
2.11 Tactical and strategic analysis
Possession is considered a key indicator of success in football. This has encouraged researchers to examine the quality and type of attacks that have been more successful in producing shots at goal and success in tournaments. Attacking play has been described as
‘organized’ or ‘elaborate’ with implications of patience or lack of directness in the approach to scoring. In contrast ‘counter attacks’ or ‘quick transitions’, imply speed in the attack with an emphasis on directness of play. Goals from Set Plays are frequently included to account for the total number of goals scored but excluded from analysis when possession is the theme. Several studies have shown agreement in that ‘organized’ attacks produce the greatest number of goals, compared with ‘counter’ attacks or goals from Set
Plays, sometimes described as ‘stopped balls’, (Mitriotasios & Armatas, 2014; Tenga &
Sigmundstad, 2011; Werlayne, 2013b; Yiannakos et al., 2006). A major criticism of all the research papers on types of attacking play is the absence or clarity of operational definitions. Counter attacks, or quick transitions, are normally associated with speed of movement and few passes involving a small number of players (Turner & Sayers, 2010).
The research on ‘possession type’ shows that more goals are scored from ‘organized’ or
‘elaborate’ attacks than ‘counter’ attacks. The evidence shows the majority of goals in
Open Play are scored with less than five passes (Hughes & Franks, 2005; Hughes & Snook,
2006), many of which would be classified as ‘counter’ attacks; which makes it difficult to explain how ‘organized’ attacks, presumably with more than five passes, produce more goals than ‘counter attacks’.
Two separate studies of the 2012 European Championships reported the majority of goals were from ‘organized’ attacks, but the reported figures were quite different.
Werlayne (2013b) reported 53.95 per cent of goals were from elaborate attacks and
Mitrotasios and Armatas (2014) reported 60%. The comparison of goals from ‘counter’ GOAL SCORING PATTERNS 69
attacks showed closer figures, 17.1 per cent and 20 per cent respectively, but the difference in goals from Set Plays is hard to explain (28.95% compared with 20%) because the figures show a considerable difference and Set Plays arise from stoppages in play. A small difference might be explained by the inclusion or exclusion of second phase goals in the
Set Play category, that is, a goal scored within a pass or two after the initial Set Play.
In a study of professional clubs in Norway, involving 997 goals scored during three seasons between 2008 and 2010, Tenga and Sigmundstad (2011) reported that teams in the top three league positions, middle of the table and the bottom three positions all scored more goals from elaborate attacks than counter attacks, which was similar to the findings in international football. In their study Tenga and Sigmundstad (2011:547) defined counter attacks as:
“Possessions with a high degree of offensive directness, which tend to approach the opponent’s goal directly by using forward passes and dribbles once possession of the ball has been won”. The definition for an elaborate attack was, “possessions with low degree of offensive directness which progress with low risk of losing the ball often by using safe passes and dribbles either backwards or sideways”.
In an earlier study of 163 professional matches during the 2004 season in Norway,
Tenga et al. (2010) reported that 203 of a total 476 goals were a result of ‘elaborate’ attacks (97 goals = 48%) or ‘counter’ attacks (106 goals = 52%), which is the opposite to results in other studies. However, the 203 goals only accounted for 43% of the 476 goals scored so there was most likely another category of goals apart from Set Plays that was not included. It is unlikely that the remaining goals (273 of the 476 = 57%) were scored from
Set Plays. It is possible that goals from zero pass moves were not included in either the counter attack or elaborate attack categories. If goals from Set Plays accounted for 40 percent of the total, which is extremely high, the remaining 17 percent of the total would still represent 80 goals, which is realistic. Unfortunately, operational definitions to describe GOAL SCORING PATTERNS 70
the types of attack were not included. In addition, to highlight the fact that teams have different types of possessions, a random sample of twenty possessions were taken from every match as a control group, to ascertain if teams generally had more ‘elaborate’ attacks than ‘counter’ attacks. From a total of 3260 team possessions a random sample of 1688 were used as the control group and categorized as ‘elaborate’ or ‘counter’ attacks. The analysis showed that there were more elaborate attacks (n = 1002, 59%) than counter attacks (n = 686, 41%) but more goals were scored from counter attacks (52%) than from elaborate attacks (48%) which supports the results from Armatas et al. (2005) who reported a greater success rate of scoring from counter attacks compared with elaborate attacks. The results of this study (Tenga et al., 2010) were in contrast to other studies
(Mitrotasios & Armatas, 2014; Werlayne, 2013b) in that more goals were scored through counter attacks than elaborate attacks. It was suggested that the different results might be a reflection of the different standard of football compared with other top international tournaments. While the breakdown of goals into elaborate or counter attacks provided insight, the study failed to account for all the goals scored and did not include operational definitions for the types of possession.
Other studies have looked at the overall time in possession and the number of passes made during the game, which is a reflection of the domination one team might have over another or how a team might dominate opponents on a regular basis (Castellano et al.,
2012; Collett, 2013; Jankovic et al., 2011; Papadimitriou, Aggeloussis, Derri,
Michalopoulou & Papas, 2001). It is logical to expect that a higher number of passes will coincide with a longer total time in possession and ultimately lead to success in winning.
There are examples where the results favor the teams that dominate opponents (Jankovic et al., 2011; Jones et al., 2004) and there are examples where teams have no significant differences in either possession time or in number of passes but have opposite outcomes in GOAL SCORING PATTERNS 71
terms of success or failure (Papadimitriou et al., 2001). In a study of total possession time after 90 minutes there was no significant difference between teams that won, drew or lost in all the matches played in three World Cups between 2002 and 2010 but when individual tournaments were analyzed there was a difference between teams that won and lost in 2006 and 2010 (Castellano et al., 2012). Based on this evidence, studies involving larger samples of tournaments where the majority of teams are of a similar standard, such as a
World Cup, teams have a similar amount of time in possession. Analysis of the total possession time of teams using a smaller sample, for example, one World Cup shows that total possession time is a key indicator of success but not always. In the 2006 World Cup, for example, there was a statistically significant difference between successful and unsuccessful teams but not in the 2002 World Cup. One might expect the winners to exhibit a higher level of possession but with Italy, the eventual winners in 2006 there was no significant difference in possession time compared with their opponents. The evidence shows that total possession time can discriminate between successful and unsuccessful teams but there will be examples when it does not. The fact that more possession can and does lead to success is evidence enough for coaches to strive to dominate opposing teams.
The data used in this analysis (Castellano et al., 2012) were sourced from the FIFA website using information contained in the technical reports from three World Cups between 2002 and 2010. The FIFA technical reports do not explain how total possession time is calculated or provide any evidence of reliability studies, which is a weakness. One might think it is a simple task to determine when a team has possession of the ball, but it is not.
Until researchers use the same operational definition there will always be differences of interpretation that result in different outcomes.
The strategy of keeping the ball to increase possession time and the number of passes a team makes has been a feature of successful club teams in Spain and the Spanish GOAL SCORING PATTERNS 72
national team between 2008 and 2012, where analysis has shown total domination in FIFA
World Cups and UEFA tournaments (FIFA, 2010; UEFA, 2012; UEFA, 2013). Spain won the European Championships in 2008 and 2012 and the World Cup in 2010. To illustrate the point, their percentage time in possession at the 2010 FIFA World Cup ranged between a low of 55% and against Germany in the semi-final, to a high of 69 percent against
Switzerland where they lost the group match 1–0. In the 2012 European Championship, time in possession ranged between 52% in the final against Italy to a high of 67 percent against Republic of Ireland. The number of passes per game ranged between 664 and 929 with an average of 778. Barcelona developed a reputation for playing possession-based football during the same period and won several domestic titles and UEFA Champions
League tournaments. Figures 2.10 and 2.11 show the comparative data for accuracy and number of passes in the UEFA Champions League in 2012–13 with Barcelona at the top of each table as well as the highest average percentage for possession.
Figure 2.10 Accuracy and number of passes in UEFA Champions League 2012–13 Source: UEFA Technical Report (2013:45) www.uefa.org/documentlibrary/competitions/uefachampionsleague/
GOAL SCORING PATTERNS 73
Figure 2.11 Average total time in possession in UEFA Champions League 2012–13 Source: UEFA Technical Report (2013:43) www.uefa.org/documentlibrary/competitions/uefachampionsleague/
In a comprehensive study involving more than 6000 club matches in domestic
European Leagues, European Champions League, Europa League matches and 299 national teams games between 2007 and 2010, the researcher tried to find a positive relationship between total time in possession and number of passes to successful outcomes.
The results showed that in club matches the longer teams kept possession the more points they won irrespective of the league. The same trend was evident in international matches
(Collett, 2013). However, when the top teams from each league were omitted from the study, possession time remained an indicator of success in most leagues, but not as strong in England, Italy France and Germany and in Spain the effect disappeared. Time in possession was not an indicator of success for individual matches involving club teams because there is a greater differential of ability in clubs and the elite teams dominate at GOAL SCORING PATTERNS 74
home or away. However, when the elite teams were omitted from the study the effect of time in possession was not significant. The link between number of passes and points gained was positive in some countries but not in all. The conclusions drawn from the study
(Collett, 2013) were that time in possession and number of passes as performance indicators are driven by the top teams but when they are omitted from the analysis the influence is marginal or not at all. Possession and a high number of passes are not indicative of success unless they translate into shots at goal and the best indicator of success at club or international level is the shots to goal ratio. Superfluous passing or passing for the sake of keeping possession was directly tested by regressing points per game on the passes to shots on goal ratio and the results were in the expected negative direction and significant. A weakness of this study is that the data were obtained from websites, which do not provide operational definitions or reliability tests so the results have to be viewed with that in mind. The author did not attempt to validate any of the information by doing any analysis of a club or international team. Studies involving large numbers of teams and matches at different levels tend to provide general trends in the interpretation of results, which are often as one might expect. The research was unique in that it looked at the effects of total possession time and the number of passes when the top teams were included and excluded from the study.
The outcome of the research from a strategic perspective did not provide anything new about the top teams in the domestic leagues in Europe or the top teams in the world in international football but it did show that for all the other teams total possession time and number of passes may not be a good indicator of success. The limitations of this analysis are that information on the length of passing sequences or the number of passes leading to goals and shots at goal, two of the most strategic discussion points were not included. GOAL SCORING PATTERNS 75
Anderson and Sally (2013) analyzed the value of having ‘more possession’ in the
EPL between 2008–09 and 2011–12. They stated, “The data also show that, whatever possession statistic you look at overall – completion percentage, volume – having more, rather than less, possession of the ball increases offensive output” (2013:154). In 2,280 team performances, the teams that did a better job of keeping the ball away from opponents had more shots and scored more goals; they also conceded fewer goals and shots at goal, outscoring opponents by 1.44 to 1.19 goals per game. They concluded,
“Clubs with more possession will not win every match –far from it – but they will win more and lose less” (2013:159). Anderson and Sally (2013) had a unique method of comparing the relevant success of playing ‘possession-based’ football with the ‘direct approach’ or ‘long ball’. They calculated the ratio of long passes to short (less than thirty- five yards) for each team. The teams at the top of the league had a lower ratio than the teams at the bottom of the league and more attempts at goal.
2.12 Football Analytics
Interest in sports analytics has increased enormously in the past ten years, resulting in a proliferation of websites, blogs and performance data available for secondary analysis.
There has been a growing academic interest in sport analytics that combines data science insights and domain-specific knowledge (see, for example, Anderson & Sally, 2013;
Coleman, 2012; Sumpter, 2016; Wright, Carling and Collins, 2014; Wei et al., 2013). Both trends have transformed the way sport is discussed and analysed.
The momentum around football analytics gained pace during my research. I was mindful that this momentum was transforming the ways in which data are visualized and discussed. During the 2010 FIFA World Cup, I integrated some of these data and visualisations into my work as analyst for the Australian team. Thereafter I tracked developments in the literature but made a strategic decision not to change the focus of my GOAL SCORING PATTERNS 76
thesis. It was conceived as a decision-support resource for coaches. As a result, I am mindful of and aware of an emerging tradition of research (Gudmundsson & Horton, 2016) and I have monitored the literature to triangulate my own research in the analysis of goal scoring in football and address some of the spatiotemporal issues raised by discussions about large-scale analysis of formations in football (Wei et al., 2013; Bialkowski et al.,
2014; Wang et al., 2015; Memmert, Lemmink & Sampaio, 2016; Sha et al., 2016).
I note Carling et al.’s (2015:3) observation: “Performance analysts can unfortunately all too easily drown themselves, practitioners and players in large amounts of cross-tabulated data, potentially leading to rejection of their work.”
I recognize that, as Lames and McGarry (2007) suggest, any analysis of performance in football must recognize “the dynamic nature of game sports” and be sensitive to the limitations of descriptive statistics. McKenzie and Cushion (2013), amongst others, have pointed to the limitations of the performance analysis literature. They assert that research must be contextualized. The approach I have taken is to provide contextual information about my data and to use longitudinal data to strengthen conversations about context. I am aware that Borrie, Jonsson and Magnusson (2002) provided valuable insights into temporal structures in games through their discussion of T- patterns. Their insights enabled within-game and between-game comparisons as
“additional perspectives” on performance. Camerino et al. (2012) reported their use of a T- pattern methodology to analyse ten matches of Barcelona FC from the 2000–2001 season.
I have spent a lifetime on professional football and am extremely sensitive to context in real-time and lapsed-time analysis of performance. Like Carling et al. (2005), I believe there are good reasons for not considering oppositions’ defensive strategies in an analysis of goals from a tournament(s) or league competition(s). The first is, they might change during the game depending on match outcomes, the second is they may be GOAL SCORING PATTERNS 77
considered to have failed whenever a goal is conceded and thirdly they are predictable because all defensive strategies are based on the application of the principles of defence; delay, depth, balance and concentration. The primary consideration is where on the field will they be applied and how the players will be positioned to implement them. For these reasons, in agreement with my supervisors, no attempt was made to consider the defensive strategies of any team in the current study.
The introduction of three categories of goals in Open Play, combined with the number of passes preceding each goal, the location of the scorer and where the team regained possession provides some contextual information about each goal. Identifying where on the field possession was regained provides information about the distance the ball traveled before entering the goal, the categories of goals provide information about the position of the defenders relative to the attackers when the goal was scored and the specific area on the field where the final pass was made when the ball was passed behind the defence or to a player level with the last defender.
One benefit of using computerized systems for analysis is that ‘chunks’ of information can be viewed according to the aspect of play under scrutiny, for example, passing moves from regained possession in the Back Third of the field that result in goals.
The incidents are a collection of successful moves from a specific area of the field and may be further separated by the number of passes made, each of which will provide more contextual information about the goals. Incidents of specific events from one match or from a collection of matches, such as World Cups, do not have to contain every detail possible because coaches will observe the incidents and form their opinions about the quality of performance and which aspects are deemed to be successful or not based upon their opinion of what constitutes ‘good’ performance. Similarly when full games are coded as opposed to selected events from a whole series of matches a coach can view ‘chunks’ of GOAL SCORING PATTERNS 78
a game to determine successful or unsuccessful moves, such as the number of times a team does not reach the Final Third after regaining possession in the Back Third. This is another example of how the context of the game can be provided with frequency data and how coding to highlight specific aspects of performance can be adapted to provide coaches with the ‘snapshot’ of their requirements.
Were I to start my research today, I am acutely aware that innovations in visualization would play an important role in supervisory conversations, data collection and analysis. I am clear that my thesis would remain a coaching-led discussion of performance. I acknowledge a literature in and the growing use of machine learning approaches. The coaches for whom I have researched this thesis are coaching in data rich environments. The challenge is to find ways to translate these data into sustainable coaching behaviours. My approach to this thesis is my choice of translation for a community of practice in which many of them are my peers.
2.13 Summary
This review of the research literature highlights some of the issues that need to be addressed by researchers when goal scoring is investigated. These include: the importance of operational definitions; the distinction between goals scored in Open Play from Set
Plays; and much more clarity about what constitutes ‘possession’. The review has identified how this study will add to the existing body of knowledge by providing a different approach to conventional performance analysis (PA). Mackenzie and Cushion
(2013:640) in their review of performance analysis in football stated,“…despite the emergence of PA, it would appear that there has been little evolution in the research, nor a development of the research areas within the PA research landscape”. They referred to the work of Reep and Benjamin, Charles Hughes and Egil Olsen as examples of PA research GOAL SCORING PATTERNS 79
that influenced football practice and that beyond these examples there is little or no recent evidence for the application of PA findings in coaching practice.
The type of data included in MacKenzie and Cushion’s review could be classified as ‘useful’ and would include topics such as: ‘When goals are scored’, ‘Total time in possession’, ‘Shots to goal ratios’, ‘Success rates of counter attacks versus elaborate attacks’ and ‘Characteristics of successful or unsuccessful teams’, depending on the definition of success. It might also include comparative data, which show if the performance indicators have changed over time in FIFA World Cups for example. This type of information might influence a coach’s thinking about playing style or recruitment of players or provoke additional analysis of the overall team performance to discover, for example, how many goals were scored from regained possessions in the defending third of the field compared with the other areas of the pitch. Data in the ‘useable’ category would cover the topics that a coach might choose to implement in practice such as, ‘Where goals are scored and how’, ‘Time in possession preceding goals scored’, or ‘The number of passes made prior to goals scored’. This type of information can be implemented at training. For example, the majority of goals are scored inside the penalty area with one touch so targets can be set for the players to implement at training to reflect what happens in the game. More examples could include trying to score within ten seconds of regaining possession, or using the inside of the foot when trying to score, to improve the shots to goal ratio, or passing into the penalty area rather than shooting from outside, these are all realistic targets linked to the implementation of research data. All of the topics in the
‘useable category have been analysed in previous research, further supporting the views of
Mackenzie and Cushion that little in PA has changed.
The interpretation of research data is difficult when operational definitions are either missing or not sufficiently explicit. A lack of operational definitions has been a GOAL SCORING PATTERNS 80
weakness in many research studies to date. Mackenzie and Cushion (2013: 648) noted,
“Evidence from the review reveals that there seems to be a lack of transparency and published operational definitions in scholars’ work (James 2006)”. An example from this study is the analysis of goals scored from ‘crossing’ the ball. If the definition of a ‘cross’ is not included, but it is reported that crosses are made from five different areas on the pitch the research is less than helpful (Breen et al., 2006). A similar criticism could be made with regards to research into Zone 14 or area C5, when it is not made clear if Zone 14 is the width of the penalty area or one third of the width of the field. Unfortunately the research of actions in Zone 14 does not include information about passes behind or in front of defenders, which are relevant to the offside rule in an area of the field where it is strategically very important. This is an illustration of the difference in the quality of research that provides useful information and research that would provide useable information.
Another issue that has been highlighted in this review is the need to differentiate between goals scored in Open Play and from Set Plays, especially when the theme of the research is ‘possession’, or where possession is regained, or when the number of passes per possession is an integral part of the study. If information from Set Plays is included in a study that examines any aspect of goals or possession or where possession is regained the results will be misleading, unless it is stated that Set Plays have been included. One aspect of many studies is the number of goals attributed to ‘zero pass’ or ‘one pass’ goals without clarification if the goals were scored in Open Play or from Set Plays. The number of goals scored from ‘zero pass’ goals is what fuels the notion of ‘chance’ in the game. Hughes and
Franks (2005) used the term, ‘loose ball’ to describe the situation, which led to the majority of shooting possessions. The classification of ‘zero pass’ goals is a result of the interpretation of ‘possession’. If possession is deemed to be lost when an opponent makes GOAL SCORING PATTERNS 81
contact with the ball or the goalkeeper saves a shot or a shot rebounds off the goalposts, the preceding events are not usually considered. A shot that rebounded from the post may have been produced after a sequence of seven passes originating in the defensive third of the field, but because of the definition of possession the passage of play leading up to the shot is eliminated from the research.
A definition of possession, created by Pollard and Reep (1997), is often used in research, for example Tenga et al. (2010).
“A team possession starts when a player gains possession of the ball by any means other than from a player of the same team. The player must have enough control over the ball to be able to have a deliberate influence on its subsequent direction. The team possession may continue with a series of passes between players of the same team but ends immediately when one of the following events occurs: a) the ball goes out of play, b) the ball touches a player of the opposing team (e.g. by means of a tackle, an intercepted pass or a shot being saved). A momentary touch that does not significantly change the direction of the ball is excluded; c) an infringement of the rules takes place (e.g. a player is offside or a foul is committed).”
The definition of possession may be distorting the outcome of research into goal scoring and there is a case for changing or modifying the definition depending on the type of information that is sought from the research. The first point to make is that the majority of goals are scored inside the penalty area. For example, 89% of goals in the 2014 World
Cup were scored inside the penalty area (FIFA, 2014) so the lead up play, which describes how the ball entered the penalty area is of vital strategic importance when passing or possession is examined. This type of information may be excluded because of the definition of possession. The penalty area is the only place where a goalkeeper can stop a shot. If researchers are particularly interested in the build up play that leads to the creation of a scoring opportunity in Open Play there is a case for ignoring the save made by the goalkeeper, or the rebound from the goalposts and possibly anything that happens once the ball has entered the penalty area, until the opposition have possession of the ball or the ball goes out of play. Hughes (1990) and Bate (1988) analyzed the number of passes GOAL SCORING PATTERNS 82
immediately preceding a ‘zero’ or ‘one pass’ goal but they are isolated cases. It is a fact that between 23% and 45% of goals have been recorded as ‘zero pass’ goals in the literature, some of which would have been scored in Open Play and attributed to ‘chance’
(Bate, 1988; Grant et al., 1998; Hughes, 1990; Hughes & Franks, 2005; Reep & Benjamin,
1968).
From a strategic perspective, current research shows that football is fairly predictable in that more goals are still scored with five passes or less despite the evidence in some studies that the top teams tend to dominate opponents in terms of total possession time and number of passes completed (UEFA, 2012; Anderson & Sally, 2013). Analysis of events preceding goals scored has not embraced the offside law, which has a major effect on the timing and direction of runs made by players to receive a pass. Furthermore, the last line of defence could be anywhere between the halfway line and the goal line so it is appropriate to cater for this eventuality in match analysis. Reducing the number of traditional zones on the field and changing the shape of the new zones did this. Zone 14, for example, was extended and changed to Zone 14+. Analysis of passes from Zone 14+, which extends from the half way line to the edge of the penalty area (see Figure 3.1,
Chapter 3), will provide a better understanding of where passes originate prior to goals being scored and if passing the ball behind the opposing defence is the most effective strategy compared with ‘Crossing the ball’ and ‘Other Methods”.
Mackenzie and Cushion (2013:647) also suggest that researchers engaging in research concerned with attributing performance outcomes to performance variables should be mindful of contextual information such as the period of the season the data were collected, the quality of the opposition faced and match location should arguably be provided in order to bring context to research data and its subsequent conclusions. This view is commendable but not always applicable or practical depending on the type of GOAL SCORING PATTERNS 83
analysis undertaken. For example the match location at a World Cup may not favour any team other than the host nation, the quality of opposition is subjective and in any competition is beyond the control of any team. It could be argued that the defensive strategies of teams should be taken into account to provide contextual information but that would only provide more conjecture unless the coaches of all the teams were interviewed to explain their strategies.
The research undertaken in this project will provide contextual information, which according to Mackenzie and Cushion (2013:647) future researchers should be mindful of, and make a significant contribution to the understanding of goal scoring patters. It will (a) differentiate between goals scored in Open Play and from Set Plays, (b) provide explicit operational definitions of three categories of goals scored in Open Play, (c) provide data on the most successful category of scoring goals in recent international tournaments and compare it with data from two professional football leagues; and (d) introduce five new areas of the field to identify the most productive area of the field to pass the ball behind opponents or to a team mate level with the last defender. Item (c) uses an amended definition of ‘possession’ and item (d) provides useable information that can be implemented by coaches in football generally but may be more appropriate in a professional club environment.
GOAL SCORING PATTERNS 85
Chapter 3 Methodology
3.1 Introduction
From the outset, the aim of this project was to establish a research methodology that would ensure validity of content captured, accuracy of data entry, and reliability in observation in order to answer, with a high degree of confidence, the research questions posed. This chapter explains how the coding system was created and how data were captured and analyzed using specific software. This process is shown in a logical sequence and justified, as is the use of Yule’s Q test for intra-operator and inter-operator reliability.
Particular attention is given to explain operational definitions in this investigation, in the light of criticisms I have made of other research in Chapter 2.
3.2 The software for coding, capturing and analyzing
It is standard practice to use one or more software packages for an extensive performance analysis research project for many reasons. These include the volume of video footage to be viewed, the ease with which footage could be retrieved for viewing and the capacity for arithmetic calculations to be made automatically. The volume of video footage captured and analyzed in this investigation was in excess of 1.5 terabytes. In order to have instant review of selected sections of these data, it was decided to use a Sportscode
Elite System (Version 9; Sportstec, Sydney, Australia). This software allows a database and a code matrix to be created to enable a coach or analyst to click on any element of the performance to view the associated video clips (Butterworth, O'Donoghue & Cropley,
2013). SPSS Version 19 software was used for statistical analysis of the numerical data created in the Sportscode Elite software to test these research questions:
1. Are more goals scored from passes behind the last line of defense, or to a player
level with the last defender than from Crosses or Other Methods? GOAL SCORING PATTERNS 86
2. In which area of the field do the majority of passes originate that lead to goals from
passes behind the opposing defence?
The decision was made to have three categories, for goals scored in Open Play. If research outcomes are to be understood, accepted by coaches and applied with players it is easier and more effective if the information is succinct in the opinion of the author.
3.3 The three new categories of goals
The three categories were chosen to represent how goals are scored in the most basic of contexts including ‘Crosses’, which is universally accepted even if it has different connotations to coaches or researchers. Other goals may be described as being scored either from in front of defenders or from level with or behind defenders, so the three categories of goals were determined as (1) From Crosses, (2) From Other Methods, which would include any number of other situations preceding a goal, and (3) From passes behind the last line of defence, or to a player level with the last defender who is able to shoot or take the ball forwards. The number of zones on the field was reduced from eighteen to seven to denote where on the field the final pass was made. See Figure 3.1.
GOAL SCORING PATTERNS 87
Figure 3.1 Areas of the field for passes behind the defence and crosses
Researchers have used a numbering system from 1–18 or L (Left) 1–6, C (Centre)
1–6 or R (Right) 1–6 to denote the areas of the field where the final pass was made prior to a goal being scored, see for example, Figure 2.3 and Figure 2.4 in Chapter 2. The areas C1 to C6 or Zones 2, 5, 8, 11,14 and 17 are the central areas from end to end of the field.
These areas are sometimes described as being the same size as all other areas or zones
(Breen et al., 2006; Rees et al., 2010; Taylor et al., 2002) or bigger than the other areas, i.e. unequal (Bell-Walker et al., 2006). The decision to use seven areas on the field was for ease of recognition for the operator when recording the events, to take the off-side rule into account and for coaches to relate to easily identifiable areas on the field. Zone 14, or area
C5 has been identified as the most productive or important area for passes leading to goals GOAL SCORING PATTERNS 88
(Rees et al., 2010) but previous research has not accounted for the position of the last line of defence, which could be anywhere between the half way line and the goal line as a result of the off-side law and the tactical strategies that different teams might apply during a match. Since Zone 14 is central on the field and recognized as a significant area the decision was made to describe the central area between the half way line and the edge of the penalty area as Zone 14+. The width of the penalty area was used as the demarcation for ease of recognition to improve accuracy of recording and reliability. The Penalty Area,
Wide Right and Wide Left were the remaining areas in the opponents half of the field to record where the final pass might originate. The zones Wide Right and Wide Left are outside the width of the penalty area and from the half way line to twenty yards from the goal line. Passes from either of these zones to the back of the defence would be considered a’ pass’ not a’ cross’. In the absence of an accepted operational definition of a ‘Cross’ the final two areas in the opponents half, the Crossing areas, are from outside the penalty area and up to twenty yards from the goal line. The decision to extend the Crossing area lengthwise beyond the edge of the penalty area to twenty yards from the goal line was due to the fact that the edge of the penalty area is often used as a guideline for defenders to hold their position, when play is in front of them. This results in some crosses being made from slightly deeper positions, between eighteen and twenty yards from the goal line. It was considered misleading and unacceptable to the author that a ‘Cross’ could originate from central areas of the field, inside the penalty area and almost anywhere from a wide position in the Final Third of the field (Breen et al., 2006). The seventh area to record where the final pass might originate was from inside a team’s own half of the field. Once the categories for the goals and the areas of the field were established the next stage was to manage the data collection at different levels of professional football, followed by data and statistical analysis. GOAL SCORING PATTERNS 89
3.4 Data collection
Due to the extensive video resources that were available for this research project, it was possible to collect data from international and professional club matches according to the operational definitions, for comparative purposes with previous research and to facilitate answering questions that might arise as a result of the research outcomes. These factors were taken in to account when the code window was designed. The collected data was stored in external hard drives and backed up as a precaution against data loss.
3.4.1 Video resources
The video resources available for this project comprised whole game and highlights recordings of international and club matches. It was important to repeat the analysis of goal scoring patterns in different competitions so that comparisons could be made between and within competitions.
Complete data sets were collected from FIFA World Cup Finals in 2002, 2006 and
2010, UEFA European Championship Finals in 2004, 2008 and 2012 and the Australian ‘A’
League in 2008–09, 2009–10 and 2011–12 seasons. A complete data set was collected from the EPL in season 2011–12 comprising 1066 goals, but in 2001–02 and 2005–06 the data set was incomplete comprising of 858 goals out of 1001 (86%) and 813 out of 944
(86%) respectively. Table 3.1 shows the total number of matches, the total of all goals from each type of competition and the number of games in which the analysed goals were scored.
GOAL SCORING PATTERNS 90
Table 3.1 Total games and goals in each competition World Cups European English Premier A League Championships League Total Games 453 94 1140 354 Games Covered 453 94 1043 354 No.Games 100 100 91 100 Covered % Total Goals Scored 304 230 3011 960 Goals Analysed No. 304 230 2737 960 No.Goals Analysed % 100 100 91 100
With the exception of the 2001–02 EPL matches, which were recorded on VHS tapes all other video resources were recorded on digital video discs (DVDs). The DVDs of full matches and highlights programs of the EPL and ‘A’ League had been recorded from television broadcasts. Additional DVDs, which had been purchased, were used to collect data from the 2005–06 season. These included whole season reviews of the Arsenal,
Chelsea, Liverpool and Manchester United football clubs. To complete the total number of goals scored in the 2011–12 EPL season, permission was granted from Murray Shaw,
Executive Producer of Football at Fox Sports to review EPL video footage in their Sydney studio and create Quick Time video files of goals from matches that were missing from the available resources in Canberra.
3.4.2 Essential data
To answer the research questions posed, each passage of play that showed the goal being scored had to include the events that preceded it, that is, was the goal a result of: a
‘Cross’ from outside the penalty area and within eighteen meters of the goal line; ‘a pass behind the opposing defense or to a player who was level with the last defender and able to take the ball forwards before shooting or passing to a team mate to score’; any other action that might be in included in the category of Other Methods. The length of each incident started at the point when the team in question regained possession of the ball, plus the GOAL SCORING PATTERNS 91
preceding action for at least two to three seconds. There were examples when it was not possible to determine when a team regained possession of the ball due to replays of other events during the broadcast. To overcome this problem during the coding of the EPL goals in 2011–12 the available resources at Fox Studios in Sydney were utilized to collect information that was not readily available during the regular television broadcast.
3.4.3 Additional data
Since a predetermined amount of footage had to be recorded for each goal, the plan was to code any other information that might be of interest in the passage of play, referred to as the ‘incident’, which is a term used as a convention by operators of Sportscode software. These data included information that could be used to compare the events that preceded the goals with other research projects as well as additional information that might be of value if the results of the research raised further questions. This additional information included; the name of the team, the different types of goals from Set Plays, for example, corners, free kicks, long throws and penalties, the number of passes made in each sequence of play, where the scorer was positioned on the field, the number of touches taken by the goal scorer, if the goal was from a header or from the use of the instep or the inside of the foot, the type of cross, was it an in-swinger or an out-swinger or if the ball was passed into the penalty area. Where a team regains possession of the ball is of interest and particularly in the attacking third. Hughes (1990) was an advocate of regaining possession of the ball quickly and as close as possible to the opponent’s goal based on the evidence he produced. Unfortunately, Hughes (1990) did not explain if possession was regained in Open Play or from a Set Play, so that level of detail was recorded in this project. GOAL SCORING PATTERNS 92
3.5 Operational definitions
The importance of operational definitions was highlighted in Chapter 2. The rationale to minimize the categories of goals in this project, provided in the Introduction to this chapter, elicited the challenge to make the operational definition for each category easy to understand and apply. However, before the definitions for the categories are explained the conventional definition of ‘Possession’ will be discussed and the rationale for amending it will be explained later in the text.
3.5.1 Definition of ‘Possession’
In some research papers the definition of possession is not included, it is assumed the reader understands (Acar et al., 2009; Hughes & Snook, 2006). Possession is sometimes referred to without being defined, for example, “the ball is passed from player to player among the eleven members of a side until a particular player loses possession of the ball either by interception or tackle on the part of a member of the defending team or by an infringement of the rules of the game or by himself shooting at the defending side’s goal” (Reep & Benjamin, 1968:581). The following definition by Pollard and Reep (1997) is used frequently when a definition is included: “A team possession starts when a player gains possession of the ball by any means other than a pass from a player of the same team.
The player must have enough control over the ball to be able to have a deliberate influence on its subsequent direction. The team possession may continue with a series of passes between players of the same team but ends immediately that one of the following events occurs:
(a) The ball goes out of play
(b) The ball touches a player of the opposing team (e.g. by means of a tackle, an
intercepted pass or a shot at goal being saved). A momentary touch that does
not significantly change the direction of the ball is excluded. GOAL SCORING PATTERNS 93
(c) An infringement of the rules takes place (e.g. a player is offside or a foul is
committed)” (Ridgewell, 2011:565; Tenga & Sigmundstad, 2011:238).
The last part of the definition acknowledges that an opponent may make contact with the ball and it will be ignored as long as it does not significantly change the direction of the ball. One might ask about the situation when a defender tackles a player and does not win the ball but it changes direction, contact is made but the defender does not gain sufficient control over the ball to be ‘in possession’? Would possession deemed to be lost in such a situation? Does it really matter if the goalkeeper saves a shot or the ball hits the post when the focus of the research is on the actions that precede goals being scored?
Definitions can and ought to be changed when there is a good reason to do so. For the purposes of this research, an amendment was made to the first sentence and part (b) of the definition mentioned above.
Namely:
The team possession may continue with a series of passes between players of the same team but ends immediately that one of the following events occurs outside the penalty area;
(b) a player of the opposite team makes contact with the ball to alter the direction of the ball for a minimum of 10 yards, or if contact is made with the head and the flight path of the ball changes at least 90 degrees.
This amendment to the definition of ‘possession’ will increase the total number of goals included in the analysis and provided a more accurate representation of the events preceding the goals. It will eliminate all but the legitimate ‘zero’ pass goals scored, for example, when a player intercepts a back pass and scores. If the amended definition has an effect it will be to increase the number of goals scored from longer passing sequences, which may encourage and support advocates of possession based football. This may be GOAL SCORING PATTERNS 94
evident when the results of goals scored from different passing sequences in this study are compared with other studies.
3.5.2 Definition of a goal from a ‘Cross’
Any pass from the ‘Crossing Area’ into the penalty area, either in the air or along the ground will be defined as a ‘Cross’. The description of the ‘Cross’ will be recorded as either ‘In-swinging Cross’, ‘Out-swinging Cross’, or ‘Passed in from the Crossing Area’
(PCA).
3.5.3 Definition of a goal from passing the ball behind the last line of defence or to a player level with the last defender who is in a position to take the ball forwards and shoot (Ball behind and strike – BB&S) or pass to a team mate. (Ball behind, pass and strike - BBP&S)
A goal will be allocated to either category if:
(a) A pass is made behind the last line of the defence, or behind the defender
marking the player who receives the ball and the player in possession takes
the ball forwards before shooting at goal. If the goalkeeper saves the shot and
a different player scores it will not alter the allocation of the goal.
(b) A pass is made to a player who is almost level with the last defender and the
player takes the ball forwards before shooting or passing to a teammate to
score.
(c) A pass is made inside the penalty area and the pass is in the air to a player
who is either level with or behind the last line of defence and the player
scores with the header or heads the ball to a teammate to score.
Note: There are many occasions when a player receives a pass behind the last line of defence and a recovering defender will make a desperate effort to block the impending shot by throwing his body between the ball and the goal. The player in possession will GOAL SCORING PATTERNS 95
often change the position of the ball, dummy to shoot and drag the ball into a position where the defender cannot block the shot, before scoring a goal. When this happens inside the penalty area, the example will not be classified as dribbling to get beyond the last line of defence because the opportunity to shoot came from a pass behind the defence in the first instance.
3.5.4 Definition of a goal from ‘Other Methods’ (OM)
Any goal in Open Play that is not allocated to one of the other two categories will be included in ‘Other Methods’. Examples of such goals would include, a shot from anywhere on the field where an opponent or opponents had a chance to block it, a goal that was a result of a player dribbling past the last line of defence followed by a shot or a pass to a team mate, as long as the pass was not behind another defender. Goals scored after intercepting a back pass, or goals scored as a result of an opponent giving the ball away would be included in ‘Other Methods’. Once the categories of goals had been determined the next step was to create a Code Window.
3.6 Creating the code window
The Sportscode Elite software enables the operator to record and analyze events that are of particular interest in any sport. There are two buttons generated by the
Sportscode software to record data that can be used in creating a Code Window, the first is the Code button and the second is the Label button. (See Figure 3.2) GOAL SCORING PATTERNS 96
Figure 3.2 Example of a code window using two generic buttons
A Code button is identified with a red triangle in the upper left hand corner of the button and a Label button is identified with a blue circle in the upper right hand corner of the button. The green lines in Figure 3.2 indicate ‘Activation’ links. If the ‘Other Methods’ button is pressed it will automatically activate the ‘Open Play’ button as well. The
‘Activation’ function can also be used with a Label button. For example, the ‘Corner’ button is linked to activate the Code button, ‘Set Plays’. The reason ‘Activation’ links are utilized is to save the operator time by reducing the number of times buttons have to be turned on or off when coding information. This is particularly important when an operator GOAL SCORING PATTERNS 97
codes in real time and may need to refer to the information at half time or immediately after a match. Each time a button is pressed or activated a numerical counting system is updated. The Label buttons may be placed on top of the Code button to save space in the
Code Window and the green lines or colored identifiers can be switched off, as shown in
Figure 3.3.
Figure 3.3 Code and Label Buttons and their identifiers
GOAL SCORING PATTERNS 98
3.6.1 Code buttons, Label buttons and the Code Matrix
Once a Code button is activated every Label button that is pressed will be recorded automatically as additional information to do with the Code button, until the Code button is deactivated. For example, in Figure 3.3 the Code button ‘Crosses’ would be activated by placing the cursor over the button and left clicking the mouse, followed by additional information to be recorded using the Label buttons, for example, did the cross come from the Right or Left side of the field, was it an In-swinging Cross or Out-swinging Cross or was it Passed in from the crossing area (PCA), was the goal scored inside the penalty area,
(Inside Box) and was it after 1–5 passes or 6–9 passes.
A Code button is different to a Label button also in the way information is presented in the Timeline, which is created automatically (see Figure 3.4). A Timeline shows the sequential creation of incidents, which are linked to the timing of the event. A separate line is created for every Code button and the data that is linked to each Code button appears inside each incident when the cursor is placed on it. In Figure 3.4 there is a row for each category of goals and the cursor is on incident number 5 in the row labeled
‘Open Play’. Just above the Open Play row there is another row with the content of the highlighted incident. It has the # 5, CZE the country that scored the goal, the number of passes (6–9) preceding the goal and all the other Label data that was recorded for that particular incident. GOAL SCORING PATTERNS 99
Figure 3.4 The Timeline with highlighted incident #5
When a Code Matrix is created a separate horizontal line appears for each Code button along with a separate line vertically for each label button, see Figure 3.5.
Figure 3.5 The Code Matrix – Code buttons Horizontally and Label buttons vertically
In Figure 3.5, a sample of the Code Matrix is provided to illustrate the Code buttons, which are on the left side in rows from top to bottom, starting with Open Play and each category of goal in Open Play, for example, BB & S and BBP & S which are the abbreviations for Ball behind and strike and Ball behind pass and strike. Crosses and Other
Methods are self-explanatory with Set Plays at the bottom. At the top of the Code Matrix the vertical columns indicate the Label button data, which were coded for each incident created by a Code button. In the ‘Crosses’ row, the total of 12 (far right) incidents coded is GOAL SCORING PATTERNS 100
broken down to show 1 goal was from an In-swinging cross, 10 goals were from Out- swinging crosses and 1 goal was from when the ball was Passed in from CA, the crossing area. Reading from left to right in the ‘Crosses’ row, 2 came from the Left side and 10 came from the Right side, 6 of the goals came after from a sequence of between 1–5 passes,
5 of the goals came from a sequence of between 6–9 passes and 1 goal came after a sequence of between 10–15 passes. All 12 goals were scored inside the penalty area, 6 goals came from headers and 11 of the goals were scored after 1 touch and 1 goal after 2 touches. Under the Spain heading there is nothing recorded in the ‘Crosses’ row, indicating
Spain did not score one goal from a ‘Cross’. It should be noted that the figure at the top in the ‘Open Play’ row is the sum of the figures in each of the Open Play categories below and the totals for each category on the right hand side provides the total for the number of goals scored in Open Play, which is 61. The goals from Set Plays are a different category and shown to account for the total number of goals scored in the tournament.
It is important to understand how data will be presented in the Code Matrix when creating a Code Window. An important function of the Code Matrix is to determine when different combinations of Label buttons have been coded. For example, if the question was in relation to the number of goals scored from headers after Out-swinging Crosses from the
Right side of the field, this could be answered in a matter of seconds by combining the relevant Label buttons at the top of the Code Matrix (see Figure 3.6). Four goals were scored with headers from Out-swinging crosses on the right side of the field. GOAL SCORING PATTERNS 101
Figure 3.6 Code Matrix – Combining label buttons
The next step in creating the Code Window was to create a Label button to record the data that would be specific to each Code button or data that might be relevant to every
Code button (see Figure 3.7). GOAL SCORING PATTERNS 102
Figure 3.7 The Code Window – Grey Code Buttons and Coloured Label Buttons
To record the areas of the field where the ball may be passed behind the opposing defence a Label button was created to record Own Half (OH), Wide Right (WR), Zone 14+
(Z14+), Wide left (WL) and Inside the Penalty Area (IPA), see Figure 3.7. These Label buttons are specific to the Code buttons, Ball behind and strike (BB&S) and Ball behind, GOAL SCORING PATTERNS 103
pass and strike (BBP&S). The Label buttons, Inside Box, Shots 18–23 yards and Shots outside 23 yards are relevant to every goal scored. The Label buttons at the top of the Code
Window include abbreviations for each participating country in the tournament so coding each goal scored by each country will allow comparisons to be made between teams and between individual teams and the average figures for all teams in the tournament. The number of passes preceding each goal is of special interest so five buttons were created to determine the parameters, for example, 0 Pass, 1–5 Passes or 6–9 passes. The number of touches taken by the scorer was taken into account along with the area of the field where possession was regained. The field was divided into three equal sections with two buttons to divide the middle third into Own Half (OH) and Their Half (TH). The buttons were dark blue and when it was extremely difficult to determine which third of the field should be used, two buttons were created [Reg. BL B3 & OH and Reg. BL TH & F3] to record those incidents. The abbreviations stood for Borderline Back Third and Own Half and similarly for the division in the opponents half of the field, Borderline Their Half and Final Third.
This was done to highlight the difficulty of accurately identifying the areas on the field
(Hughes & Franks 2005) and in the event that, for example, 4 incidents are recorded in, BL
B3 & OH, they might be evenly distributed to the Back Third and Own Half of midfield when the coding is completed. The CONFM and UNCON buttons were to confirm if the exact location where possession was regained could be observed or not in the available footage. Regained possessions in the Attacking Third or Final Third of the field are of special interest so green Label buttons were created to account for the different ways possession might be regained. All other relevant information about Set Plays was accounted for and by continually adding Label buttons the Code Window slowly takes shape as shown in Figure 3.8. Once the Code Window had been created the next step in the process was to record the data. GOAL SCORING PATTERNS 104
Figure 3.8 The Code Window for the European Championships 2012
3.7 Capturing the data
The video resources used in this study included full matches and highlights programs of the English Premier League (EPL), the Australian ‘A’ League, FIFA World GOAL SCORING PATTERNS 105
Cups and UEFA European Championships. The database included the following number of goals from each international tournament or league championship:
Table 3.2 Number of matches and goals analyzed in each competition 123 Number of Goals Analyzed Percentage of matches total FIFA World Cup 2002 61 161 100 2006 57 147 100 2010 57 145 100 EURO Championships 2004 27 77 100 2008 29 77 100 2012 29 76 100 English Premier League 2001–02 297 858 86 2005–06 300 813 86 2011–12 353 1066 100 Australian ‘A’ League 2008–09 80 249 100 2009–10 128 346 100 2011–12 128 365 100
3.7.1 Equipment
A Sony DVD player, model number DVP-NS728H was used to play the discs and an LG 6 Head Video Cassette Recorder was used to play the VHS tapes. The standard definition signal was transferred from each piece of equipment, through RCA cables connected to a Canopus signal converter, model ADVC-55 and then by Firewire cable
(400 out to 800 in) to the Apple Macbook Pro computer.
1 In 2009-10 season the number of teams increased from 8 to 10 in the ‘A’ League competition 2 The number of matches excludes 0-0 results and missing games in EPL seasons 2001-02 and 2005-06 3 The number of matches excludes 0-0 results and missing games in EPL seasons 2001-02 and 2005-06 GOAL SCORING PATTERNS 106
3.7.2 Method
Once the video feed was transferred to the computer hard drive, each goal was viewed before being captured. The fast forward function was used to identify when each goal was scored and then the operator watched it, sometimes repeatedly, to determine the category the goal should be placed in. When the operator was not sure to which category the goal should be assigned to the operational definitions were referred to until a decision was made. The essential part of the incident had to include when possession of the ball was regained and that was determined by the amended definition of possession, described earlier in the text in 3.5.1. Once possession was established the capturing of each incident started prior to when possession was regained. In some cases it was not possible to identify exactly where or when the team regained possession of the ball because of close-ups or video replays of earlier incidents in the game. When the regained possession could be confirmed it was coded as such with a label button CONFM or CONHL to differentiate between FM (full match) and HL (highlights). The Label button UNCON was used to record when possession and therefore the number of passes preceding a goal could not be confirmed (see Figure 3.8). To determine which third of the field should be used to record where possession was regained an estimate was made based on the number of mown sections on the field, when they were visible, or a visual approximation was made. Most football fields have ten mown strips in each half so one third was equal to approximately just over six and a half mown strips, for example, the Middle Third comprised of three and a half mown strips either side of the halfway line. The operator coded the additional information, using the Label buttons that were relevant to the category of goal. The operator paused capturing after each goal to eliminate player celebrations and crowd scenes and then captured replays of the goal from different camera angles, which often helped the observer confirm the details that needed to be coded. GOAL SCORING PATTERNS 107
3.7.3 Accuracy of coding data
The interpretation of regaining possession needs to be understood clearly because this is the point where each incident begins. Sometimes it is difficult to describe precisely or practically in words what is being observed. If a player heads the ball and it goes to a teammate, the point at which the header was made is determined to be the area of the field where possession was regained. Headers are counted as the first pass in a sequence because it is impossible to determine if it was intentional or not for the ball to go to a teammate. If a player makes a tackle and the ball goes to a teammate, possession is regained where the tackle was made but it is not counted as a pass. When an interception is made and the ball goes to a teammate the operator has to decide if there was intent to pass the ball, which is possible. When the goalkeeper regains possession of the ball and throws it to a teammate, the throw is not counted as a pass and the same criteria is applied when a throw-in is taken to restart the game. If a goalkeeper uses his foot to pass the ball to a teammate it will be counted as a pass. If a player attempts to make a pass and the ball makes contact with an opponent but goes to another player on the attacking team only one pass is counted, unless the ball travels more than 10 yards in which case possession will have been deemed to be lost, in accordance with the definition of possession. If a player attempts to make a pass and the ball rebounds off an opponent and comes back to him the attempted pass is not counted in a sequence. Each incident that is created by activating and deactivating a Code button is counted automatically so in the timeline there is continuous calculation of the number of incidents created in each category of goals.
The ‘Activation’ link, which was explained earlier, provides an opportunity to improve accuracy when coding because when two buttons are linked, only one button has to be activated, which eliminates error of omission. The Code Matrix, see Figure 3.9, provides the opportunity to cross reference the total number of incidents recorded in, for GOAL SCORING PATTERNS 108
example, the Open Play category (61) with the totals of the goals allocated to the other categories, that is, BB&S (21) BBP&S (9), Crosses (12) and Other Methods (19). The total for Set Plays (15) can be added to the total in Open Play (61) to make sure every goal is counted (76). Creating a Code window so that cross-referencing is possible also allows the operator to find information that has been omitted, which becomes apparent when the totals do not add up. This is a common occurrence when coding a lot of information. For example, Figure 3.9 shows a sample of Label buttons from the Code Matrix for goals in the UEFA 2012 European Championships. The totals for the categories of passes, 0 Pass
(7), 1–5 Passes (30), 6–9 Passes (17) and 10–15 Passes (5) add up to 59, but it should add up to 61, which is the total for the number of goals in Open Play. The ‘square’ for each category of pass has been clicked, which will highlight all of these incidents in the
Timeline, shown in Figure 3.10. Incidents, which have not had the number of passes coded will not be highlighted, which in this example are instances 9 and 26. Also, the totals for the number of touches taken by the scorer do not add up to 61, so the same process would have to be applied to find the incidents, which have not been coded. This process of cross- referencing was employed for every group of Label buttons to ensure accuracy. GOAL SCORING PATTERNS 109
Figure 3.9 The Code Matrix – the number of goals from different passing sequences
Figure 3.10 The Timeline with the highlighted incidents from the Code Matrix
Once the incidents have been identified, the next step is to add in or code the missing information. The information will most likely be different for each incident so they have to be dealt with separately. In Figure 3.11 the mouse has been placed on incident number 26 and the coded information appears at the top of the Timeline. GOAL SCORING PATTERNS 110
Figure 3.11 Highlighted incident #26 in the Timeline
The video clip of the incident will then be played to analyze the information that needs to be coded, which in this instance is the number of passes preceding the goal. The category of passes that needs to be coded is 1–5 Passes. To code the missing information, the incident number 26 has to be highlighted in the Timeline, as in Figure 3.11 and then a function called Label mode is used. The Label mode has to be activated by clicking the third button from the left at the top of the Code Window, shown in Figure 3.12. The next step is to click on the Label button 1–5 Passes.
Figure 3.12 The Label Mode Function button, 3rd from left in the Code Window
GOAL SCORING PATTERNS 111
The same process needs to be completed for incident number 9 and for the number of ‘touches’ as there are two incidents where that information has been omitted. Once the information has been coded the Code Matrix is checked to make sure the totals for each category of Label buttons is correct as in Figure 3.13.
Figure 3.13 Updated Code Matrix
Figure 3.13 is a sample of the full Code Matrix, which is shown in Figure 3.14, in which the text is too small to read. A readable version of the code matrix is in Appendix 25
(1), page 326.
GOAL SCORING PATTERNS 112
Figure 3.14 A Code Matrix with all Label Buttons included
When the EPL and ‘A’ League goals were captured they were done so in batches because of time constraints and to check for accuracy of coding without having to deal with a high volume of incidents. A typical batch included the goals scored over one or two rounds of matches of the season. The goals were captured, coded and then the Code Matrix was checked to ensure 100% accuracy. If a batch was not completed because coding had to stop for a reason, there is a function called ‘Append’, which allows the capturing to continue in an existing file at a later date. Once a batch of goals had been completed and checked for accuracy it was added to the Database, which would eventually contain every goal captured. The Database file is created from the Timeline and saves the edited incidents along with coded information. A Database provides an option to delete the original files or to use it as a back up of the original files, which is how it was used in this project. When all the goals for a particular tournament have been entered in the Database, a
Code Matrix can be created that includes every detail that has been coded through the
Code and Label buttons as shown in Figure 3.14. GOAL SCORING PATTERNS 113
3.8 Creating a database
The procedure to create a database with the Sportscode software is: open the ‘File’ menu on the toolbar in, scroll down to ‘New’, then across to ‘Database’, which presents a request to name and locate where the database will be saved, shown in Figure 3.15.
Figure 3.15 Creating a database
Once the database has been created the incidents from the Timeline need to be highlighted as shown in Figure 3.16 before clicking on the Database Icon located on the extreme right hand side in the timeline toolbar.
Figure 3.16 Highlighted incidents to be transferred to the database
GOAL SCORING PATTERNS 114
Once the Database icon in the Timeline has been clicked a drop down menu
(Figure 3.17) will ask if the incidents are to be transferred to the Database.
Figure 3.17 Transfer the selected instances to the database
After selecting the ‘Yes” button, the next menu offers the type of file to be created,
(Figure 3.18) of which ‘standalone’ is the preferred option for back up purposes. The last step in the process is to select the ‘Export’ function at the right of the screen and the files will be created in the Database.
Figure 3.18 Exporting the standalone movie file of incidents to the database
3.9 Data analysis
Two software programs were used in this study. The Sportscode Elite software created the coded raw data and SPSS software was used to apply statistical tests. GOAL SCORING PATTERNS 115
3.9.1 Analysis using Sportscode Elite
The Sportscode Elite software maintains a continuous calculation of incidents, which represent Code buttons and appear in the Timeline as each incident is created, shown in Figure 3.19.
Figure 3.19 Sample of a Timeline showing Code Buttons and Incidents in each row
The total number of incidents or goals in ‘Open Play’ is recorded in the top line in
Figure 3.19 and in this example 14 of the incidents are visible. The number of goals in each category is shown in the corresponding line, for example in line 5 there are 5 incidents in ‘Other Methods’, in line 2 there are 6 incidents in BB & S. The totals for the number of incidents in each line, 6 in line 2, 1 in line 3, 2 in line 4 and 5 in line 5 add up to the total of 14 in Open Play. The incidents in line 6 are Set Plays so they are not included in Open Play. All incidents from the Timeline can be viewed in the Code Matrix.
Figure 3.20 shows a sample of the Code Matrix, which has the Code buttons listed from top to bottom on the left side and each Label button that has been created and coded is shown horizontally from left to right at the top. GOAL SCORING PATTERNS 116
Figure 3.20 The Code Matrix showing Categories on the left, labels at the top and the frequency of recordings in the corresponding boxes
In this example the Label buttons start with Reg. in B3 (17), which shows that 17 of the total goals in Open Play came after possession had been regained in the Defending
Third or Back Third of the field. The diagram also shows that 8 of the goals from playing the ball behind the opposing defence, followed by a strike (BB&S) came from a regained possession in the Back Third, as did 3 goals from a ball behind the defence followed by a pass and strike (BBP&S), 2 goals from Crosses and 4 from Other Methods. Each column with Reg. (short for Regained) at the start depicts a different area of the field where possession was regained for each goal scored in each category. This example also shows the number of passes made before the goals in each category, 0, 1–5 Passes, 6–9 Passes and 10–15 Passes and the number of touches, 1 Touch, 2 Touch, or 3+ Touches, taken by the scorer. The last Label buttons in this example show how many goals in each category were scored Inside the Box, from 18–23 metres and outside 23 metres, which should have been reported in yards. The Code Matrix allows label buttons to be combined, by clicking and dragging the column to the top row to find additional information about the goals in GOAL SCORING PATTERNS 117
each category. For example Figure 3.21 shows the number of goals in each category that were scored after possession had been regained in the Back Third and after 6–9 passes; 5 in Open Play, 2 from BB&S, 1 from a Cross and 2 from Other Methods.
Figure 3.21 Two Label buttons combined in the Code Matrix
Figure 3.22 Three Label buttons combined in the Code Matrix GOAL SCORING PATTERNS 118
Figure 3.22 shows the number of goals from regained possession in the Back Third, from 6–9 passes and from 1 touch. Figure 3.23 shows the number of goals from Regained possession in the Back Third, from 6–9 Passes, from 1 touch and scored Inside the Box.
Figure 3.23 Four Label buttons combined in the Code Matrix
The Code Matrix provides the opportunity to answer questions that might arise after the initial research questions have been answered and is a unique feature of the very sophisticated software. Once the data were collected for each category of goals and for each tournament, for example, the EPL seasons 2001–02, 2005–06 and 2011–12 the next step in the analysis process was to use SPSS Version 19 software.
3.9.2 Statistical Analysis
The data for goals scored from a pass behind the opposing defence followed by a strike (BBP&S) and goals scored from a pass behind the opposing defence followed by a pass and strike (BBP&S) were combined and entered as BBSPS. Combining BBSPS with
Crosses and Other Methods created three categories of goals. The data for goals in the EPL, GOAL SCORING PATTERNS 119
‘A’ League, European Championships, and World Cups were normalized for statistical tests but actual figures for events in each competition were entered for Pearson’s Chi
Square tests.
The data for each tournament were entered in SPSS Version 19 and assessed for normality, equal variance and reliability. The level of significance was set at <0.05.
Parametric tests (Kolmogorov-Smirnov and Shapiro-Wilk) were used to assess normal distribution. If the significance score was less than <0.05 the Kruskal-Wallis test was utilized for non-parametric data and the post hoc Bonferroni test was used to identify where the significant difference was in the data. A one-way ANOVA was used to assess the data for equal variance and Levene’s statistic determined if the variation was within a range of acceptability. Proportions were analysed using Pearson’s Chi Square with the corresponding residual scores.
3.9.3 Reliability
Intra-operator and inter-operator reliability were assessed using Kappa
Measurement of Agreement, Cronbach’s Alpha Reliability Test, by calculating the percentage of error and Yule’s Q Test.
3.9.3.1 Intra-operator reliability
Intra operator reliability was calculated by comparing the results of coding the goals in the UEFA European Championships in July 2012 and six months later. In other studies a period of four to six weeks was given to operators to recode ‘5 minute’ samples from 6 matches or 45 minutes of one match (Jones et al., 2004; Scoulding et al., 2002;
Turner & Sayers, 2010).
The length of the movie file containing the edited incidents of the 76 goals from the
2012 European Championships was 58 minutes in total. The goals were scored in 29 of the GOAL SCORING PATTERNS 120
31 matches because 2 matches were decided with a penalty shoot-out after the matches ended with a 0–0 score-line. The information to test for reliability of coding was selected in order of importance to answer the research questions and required Operator 1 to code between nine and 12 discrete events for each goal scored in Open Play and between 5and 7 events for goals from Set Plays. The retest required Operator 1 to code a minimum of 700 events. The first task was to allocate the total of 76 goals to either Set Plays or Open Play.
The second task was to allocate goals into one of the three categories of goals in Open Play,
BBSPS, Crosses or Other Methods. The third task was to determine where each pass was made for goals in the BBSPS category, from one of the five areas of the field, Own Half
(OH), Wide Right (WR), Zone 14+ (Z14+), Wide Left (WL) or Inside the Penalty Area
(IPA). Other information for comparative purposes is important to the investigation into goal scoring patterns so the final task for goals in Open Play was to code the:
1) number of goals from passes of 0–5 and 6+ in each incident
2) number of touches taken by the scorer and type of strike, e.g. instep, header,
passed in
3) type of Cross and which side of the field it came from
4) number of confirmed and unconfirmed incidents
5) ‘third’ of the field where possession was regained
6) distance from goal they were scored, either inside the penalty area, between 18–23
yards or outside 23 yards, and
7) type of Set Play, and type of Corner or Free-Kick
A total of 21, 2x2 tables (See Table 3.3) were used for intra-operator reliability using the Yule’s Q Test. Examples of how Yule’s Q Test was calculated in a 2x2 table are provided in Table 3.7 and Table 3.8. The data for calculating the intra operator reliability GOAL SCORING PATTERNS 121
scores for the 21, 2x2 are in Appendix 19(1) and the results of Yule’s Q Test are in
Appendix 19(2).
Table 3.3 The combined factors for categorizing data in Yule’s Q Test4 Open Play & Set Play Own Half & Wide Left Reg. in Own Half of MF & Their Half of MF BB&S & BBP&S Zone 14+ & Wide Right Reg. in Their Half of MF & Final Third BBSPS & Other Methods Zone 14+ & Wide Left Inside Penalty Area & 18- 23 yards BBSPS & Crosses Cross & Wide Right 18-23 yards & outside 23 yards Other Methods & Crosses Cross & Wide Left 1 Touch & 2 Touch Own Half & Wide Right 0–5 Passes & 6+ Passes 2 Touch & 3 + Touches Own Half & Zone 14+ Reg. in B3 & Own Half of MF Confirmed & Unconfirmed
3.9.3.2 Justification for the use of Yule’s Q Test
The following tests of reliability were considered in terms of suitability for the type of data collected: Kappa’s Measurement of Agreement; Cronbach’s Alpha Reliability Test;
Yule’s Q Test; and a calculation of the percentage of error (Hughes, Cooper & Nevill,
2002). A comparison of the scores recorded from each test of reliability is shown later in the text.
The Kappa Test was used for intra-operator reliability with 13 variables in two tests, listed in Table 3.5. In the tests, 5 variables had complete agreement, 4 disagreed by 1 and 4 variables disagreed by 2. If there is disagreement in the placement of one variable there will automatically be disagreement in the totals of two categories. The Kappa test for
Version 1 (V1) and Version 2 (V2) in Table 3.4 produced a score of .350, which is an unacceptable level of agreement for reliability. Even when there is a high degree of
4 BB&S - Ball behind and strike, BBP&S - Ball behind pass and strike, BBSPS – BB&S and BBP&S combined GOAL SCORING PATTERNS 122
agreement, Kappa will produce a relatively low score because it takes ‘chance’ into consideration, which it could be argued, is not really appropriate in this study.
Cronbach’s Alpha Reliability Test is a correlation test, which has been proven to be insensitive to changes in non-parametric data (Hughes et al., 2002). The V1 and V2 data in this example were compared using SPSS V21 and then V3 and V4 data were manipulated between 20 and 30 per cent but similar results were found, in that the changes to the data were not reflected in the correlation scores: V1 and V2 = 0.999, V1 and V3 = .0998 and
V1 and V4 = 0.996 The data could not be manipulated to reflect a consistent percentage of change between V1, V3 and V4 because of the low number of frequencies in some variables. Both tests were deemed to be inappropriate in this study.
Table 3.4 Calculations of Kappa and Cronbach’s Alpha Actual Manipulated V1 & V2 V1 & V3 V1 & V4 Kappa Scores .350 .106 .055
Cronbach’s Alpha .999 .998 .996
GOAL SCORING PATTERNS 123
Table 3.5 Calculation of Percentage Error Variable Actual Manipulated V1 V2 V3 V4 Open Play 61 61 61 61 BBSPS 30 28 27 26 Cross 12 13 14 11 Other Methods 19 20 20 24 Own Half 3 1 2 0 Wide Right 2 2 1 1 Zone 14+ 22 22 21 23 Wide Left 3 3 2 4 Inside Penalty Area 53 55 56 58 <23 Yards 6 4 3 2 >23 Yards 2 2 2 1 0–5 Passes 38 39 40 41 6+ Passes 23 22 21 20 274 272 270 272 Sum [V1–V2] = 2 Sum of [V1–V3] = 4 Sum [V1+V2]/2 = 273 Sum of [V1+V3]/2 = 272 % Error 0.73% % Error 1.47%
Data can be treated in different ways to calculate the percentage of error so analysts need to be very explicit when presenting results (Hughes et al., 2002). The simplest method to measure percentage error is to calculate the difference between two attempts at coding the same events, for example, in Table 3.5 the ‘Other Methods’ totals of 19 and 20 were recorded, showing 95% agreement (19/20 x 100 = 95%) or 5% error. If the accepted % error is set at 5% or 10% this method can be applied to each of the categories to determine the level of intra-operator reliability. There are issues with comparing categorical data, which will be discussed in the next section.
Another method to calculate percentage error is to use the following formula. This was applied in two ways with the sample data in Table 3.5.
(Sum of (mod [V1–V2]) (Vmean) 100
GOAL SCORING PATTERNS 124
The second method is to divide the difference in the total number of incidents
(274–272 = 2) after two attempts to code the same data (V1 and V2), by the mean of both attempts (273) multiplied by 100, (2/273) x 100 = 0.73%. This method was deemed to be unsuitable because it did not identify differences in the coding of individual variables. The second method was to divide the difference between each variable, for example, Inside 53
– 55 = 2, by the mean of the two coding attempts, multiplied by 100. (2 / 54 x100) = 3.7% error. This was done for each variable where a difference was recorded, which produced mixed results. Some examples are in Table 3.6.
Table 3.6 Alternative methods of calculating Percentage Error5 V1 V2 % Agreement % Sum (V1–V2) % Error V1/V2*100 Error (V mean)*100 BBSPS 30 28 28/30*100 =93% 7% 2/29*100 = 7% CROSS 12 13 12/138100 =92% 8% 1/12.5*100 = 8% OM 19 20 19/20*100 = 5% 1/19.5*100 = 5% <23 YARDS 6 4 95% 33% 2/5*100 = 40% 4/6*100 =67%
The use of percentage error to measure reliability was rejected for two reasons, (1) percentage scores show big differences when the data set involves small numbers, for example 3 out of 4 shows 25% error and (2) the results are the same for the simplest and most complicated method of calculation, as shown in Tables 3.5 and 3.6. The use of percentage error with small numbers of data might encourage analysts to accept a higher level of error of reliability, for example 10%, or consider the level of reliability to be unacceptable.
5 BBSPS – Ball behind & strike combined with Ball behind pass and strike OM – Other Methods GOAL SCORING PATTERNS 125
The Yule’s Q Test is considered by James, Taylor and Stanley (2007) to be the most appropriate to measure the collection of categorical data on the nominal scale, that is, where events are recorded within defined categories.
“Reliability measures should reflect the way in which the notation data is analyzed so that assessments can be made regarding the extent to which each variable presented in the results was coded accurately.” (James et al., 2007:1)
The Yule’s Q statistic provides a measure of probability, or odds ratio that two analysts will agree or disagree on the categorization of events or that one analyst will code consistently. For intra-operator reliability the following example in Table 3.7, is given to rate the coding of passes in the 0–5 and 6+ categories using data from Table 3.5 for two attempts, V1 and V2.
Table 3.7 Yule’s Q Test for coding two categories of events V1
0–5 6+
Agree (a) Disagree (b) 0–5 38 1
V2 Disagree (c) Agree (d) 6+ 1 22
The following formula is used to calculate the odds ratio of coding two categories, when (a) is the figure agreed for coding 0–5 passes, (d) is the agreed figure for 6+ passes and (b) and (c) are the number of disagreements for 0–5 and 6+ passes.
�� �� Yule’s Q Test Formula � = �� ��
Q = (38x22) – (1x1) = 836 – 1 = 835 = .997 Yule’s Q = 0.997 (38x22) + (1x1) 836 + 1 837 GOAL SCORING PATTERNS 126
Yule’s Q score of 0.997 indicates there is a probability of 99% that the operator will code accurately the number of passes in the two categories, 0–5 and 6+. There is a second more complex formula, which can be used if there are three categories to compare.
Of the two formulae it is easier to compute three pairs of variables than it is to compare one variable with two others combined. For example, to calculate the reliability of coding the three most important categories in this study, BBSPS, Other Methods and Crosses, it may be done by comparing any two of the three variables using the Yule’s Q Test formula; see Table 3.8.
Table 3.8 Yule’s Q Test for BBSPS and Other Methods6
V1 BBSPS OM
Agree (a) Disagree (b) BBSPS 28 2
V2 Disagree (c) Agree (d) OM 1 19
�� �� Yule’s Q Test Formula � = �� �� Q= (28x19) – 2 = 530–2 = 528 = .992 Yule’s Q = 0.992 (28x19) + 2 = 530+2 532
BBSPS and Crosses
BBSPS (V1–30, V2–28) and Crosses (V1–12, V2–13)
BBSPS – (Agree 28, Disagree 2), Crosses – (Agree 12, Disagree 1)
Q = (28x12) – 2 = 336–2 = 334 = .988 Yule’s Q = 0.988
6 BBSPS – Ball behind and strike combined with Ball behind pass and strike OM – Other Methods
GOAL SCORING PATTERNS 127
(28x12) + 2 336+2 338
Other Methods and Crosses
Other Methods (V1–19, V2–20) and Crosses (V1–12, V2–13)
Other Methods – (Agree 19, Disagree 1), Crosses – (Agree 12, Disagree 1)
Q = (19x12) – 2 = 228–2 = 226 = .982 Yule’s Q = 0.982 (19x12) + 2 228+2 230
This easy calculation to compare the reliability of coding any two of the three variables shows the operator has between 98% and 99% probability of correctly coding each category.
“A Yule’s Q (gamma) value of 0.95 or above should be deemed an acceptable value to determine confidence in the analyst’s ability to place an event into a category reliably.” (James et al., 2007:10)
3.9.3.3 Comparing categorical data
Table 3.9 Comparing data of equal numbers7 Variable Score V1 V2 0–5 Passes 38 39 Total number of passes equals 61 in 6+ Passes 23 22 V1 and V2 Own Half 3 1 Totals in this group V1 (30) and V2 Wide Right 2 2 (28) are equal to the number in Zone 14+ 22 22 BBSPS V1 (30) and V2 (28) Wide Left 3 3 Inside Penalty Area 53 55 The totals for V1 and V2 are 61, <23 Yards 6 4 equal to the number of goals scored >23 Yards 2 2 Open Play 61 61 The total number of goals in Open Play BBSPS 30 28 The totals for V1 and V2 are 61, Cross 12 13 equal to the number of goals scored Other Methods 19 20 in Open Play
7 BBSPS – Ball behind and strike combined with Ball behind pass and strike
GOAL SCORING PATTERNS 128
The thirteen variables in Table 3.9 are divided into groups that have a total of 61, which is the total number of goals scored in Open Play in this set of data. Each group contains data for an aspect of the goals scored, for example, the number of passes that preceded the goal or the position on the field where the goal was scored. When Yule’s Q
Test is applied using the figures in each group with a total of 61 the probability score will reflect the reliability of the operator to code each variable. However, the group that includes the areas of the field where the pass was made behind the defence, that is, Own
Half, Wide Right, Zone 14+ and Wide Left has different totals, (V1 – 30, V2 – 28) because they show the breakdown of goals in the BBSPS category. There is a difference of two in the coding of goals from the Own Half area of the field, V1 = 3, V2 = 1, but this is not a coding error because there was only agreement on 28 goals in the BBSPS category.
Therefore only 28 goals, not 30, should be used to test for reliability of coding in the four areas of the field. It was obvious in this example that the two goals of disagreement were both in the Own Half area of the field because there is complete agreement in Wide Right,
Zone 14+ and Wide Left.
If the identification of the goals in disagreement is not obvious, opening and comparing both timelines can easily find them. The two incidents of disagreement, in this example, are obvious and will be removed from the reliability test. The point of the reliability test is to assess the operator’s ability to place an event into a category so in situations where a second level of analysis is applied, it is vitally important that the same number of instances is used, which will be the number of ‘agreed’ instances.
The Yule’s Q Test shows the operator has a probability of between 98% (0.982) and 99% (0.992) to code the categories of goals, BBSPS, Other Methods and Crosses and a probability of 100% in coding the areas of the field, Own Half, Wide Right, Zone 14+ and
Wide Left. When Yule’s Q Test is applied to the appropriate number of instances for each GOAL SCORING PATTERNS 129
variable and a score of 0.94 or less is recorded, it will highlight variables where either further training or clarification of the operational definition is required to improve the reliability of coding, or simply recognize a situation where an accurate decision is difficult to make. When a score of 0.94 or less was recorded for a particular variable, the relevant incidents were replayed and viewed by both operators until an acceptable level of agreement was reached.
Examples where this occurred (and is most likely to happen) was the imaginary or real line on the field between two areas. For example an inter-reliability score of 0.90 was recorded for placing the point where a pass was made behind the defence in either the
‘Own Half’ or ‘Zone 14+; the halfway line separates both areas and sometimes the ball is played on the line. Similarly an inter-reliability score of 94.9 was recorded for assigning goals from regained possession in ‘Their Half’ of midfield or in the Final Third, which is very difficult because the demarcation line is imaginary.
There are expert opinions on reliability issues in performance analysis
(O’Donoghue, 2007) procedures to be followed for analyzing different types of data
(Hughes et al., 2002; James et al., 2007; Nevill, Atkinson, Hughes & Cooper, 2002;
O'Donoghue, Papadimitriiou, Gourgoulis & Haralambis, 2012) and methods for assessing the reliability of data entry (Cooper, Hughes, O’Donoghue & Nevill, 2007), which make the selection of the most appropriate statistical and reliability tests extremely difficult for analysts who are not experts on ‘statistics’. So, it is highly likely that no matter which tests are selected for statistical analysis and reliability there will not be total agreement among the experts. This was a major reason why complete data sets were used in this investigation.
When research studies involve data from a whole season or from several seasons
O’Donoghue et al. (2012:154) suggested: GOAL SCORING PATTERNS 130
“There may even be a case in such investigations for not using statistical tests and p values at all. If the scope of an investigation is a particular tournament in a particular season, the data may be the whole population of relevant matches rather than a sample. In such cases, the results can be reported as factual if reliable methods are used without need to reason about sampling error.”
3.9.3.4 Inter-operator reliability
Inter-operator reliability was tested with the assistance of an experienced coach,
Operator 2, who was familiar with the Sportscode Elite software. Initial training was given to explain the purpose of the research, the operational definitions for the categories of goals and how the Code Window was designed so the ‘Label buttons’ provided all of the additional information that was relevant to the study. Operator 2 was shown how to code the goals using footage of the EPL 2011–12 and this provided an opportunity to discuss and explain the operational definitions, which was recommended when complex observational tasks have to be performed by (O’Donoghue, 2007). Once Operator 2 felt he was sufficiently informed to start coding goals a World Cup database was copied and the information in each incident in the timeline was deleted. Operator 2 was given the opportunity to practice coding the goals, initially with supervision and then without supervision but where he could ask for clarification when necessary. Operator 2 had access to a Sportscode Elite system so he was provided with a portable hard drive containing a blank database of the 76 goals scored in the 2012 European Championships, the Code window and two movie files that contained examples of the three categories of goals from the EPL, as a source of reference. After two weeks Operator 2 presented a timeline with his first attempt at coding the goals from the 2012 European Championship. The Code
Matrix (Figure 3.24) is included to show the full extent of coding but it cannot be read in this form. A readable copy of the Code Matrix is in Appendix 25. Figure 3.25 shows a close up of coded events from the Code Matrix. The matrix was checked for accuracy of coding by Operator 1 and feedback was provided to Operator 2 to identify incomplete GOAL SCORING PATTERNS 131
coding and how to find errors of omission. The technique of finding errors made in coding information was explained earlier in the text, see pages 99 and 100.
Figure 3.24 The Code Matrix with all Label buttons at the top
Figure 3.25, shows extracts of the full Code Matrix (Figure 3.24).
Figure 3.25 Extracts of the full Code Matrix shown in Figure 3.24
GOAL SCORING PATTERNS 132
The totals of the passing categories, 0 Pass to 16 Passes+, add up to 53 and the number of touches, from 1 Touch to 3+ Touches add up to 60 but both totals should add up to 61, the total number of goals scored in Open Play. This shows there are errors of omission. There was confusion in the use of the Label buttons, WR, Z14+ and WL because the totals of these three labels add up to 35 so these buttons have been used inappropriately to describe goals other than those from BB&S and BBP&S, which total 27.
Clarification was needed for coding goals from passing the ball behind the opposing defence, when technically the goal could be placed in either of the two categories,
BB&S or BBP&S. An example of this is when a pass behind the defense is made from
Zone 14+, followed by another pass inside the penalty area that goes behind the last line of defence. This raises the question, ‘should the goal be coded as a ball behind from inside the penalty area (IPA) and a strike at goal (BB&S) or as a goal from a ball behind from
Zone 14+, (Z14+) followed by a pass inside the penalty area (BBP&S)? Clarification was given that the first pass behind the defense should be used to determine which category, so in this example the goal should be allocated to the BBP&S category. To assist Operator 2 a modified Code Window, Figure 3.26, was provided which had the same color coded Label buttons as the relevant Code buttons. For example the Code button ‘Crosses’ is black in color so the Label buttons that were only relevant to ‘Crosses’ were also colored black. GOAL SCORING PATTERNS 133
Figure 3.26 A colour coded Code Window
The Label buttons to record which third of the field possession was regained are dark blue. The dark green label buttons record how possession was regained in the Final
Third. Operator 2 was asked to go through his Code Matrix and check for accuracy and then review the coding of goals into the categories, which he did alone in his own time. GOAL SCORING PATTERNS 134
The complete second Code Matrix presented by Operator 2 is in Appendices 25 (3) and
25(4) and extracts are shown in Figure 3.27.
Figure 3.27 Extracts of the 2nd Code Matrix by Operator 2
By cross referencing the totals in each category, for example, the number of passes, number of touches and where possession was regained, the accuracy of coding can be confirmed. A total of thirteen factors were independently checked for inter-operator reliability using the Yule’s Q Test.
3.10 Justification of methodology
The actions or events that precede each goal in Open Play are the main focus of this research. Due to the small number of goals scored in international football tournaments the aim was to include every goal in this study and not exclude any of them. ‘Own Goals’ are often excluded in research and ‘zero’ pass goals are normally referred to as just that, with little or no information regarding the preceding events. In this study ‘Own Goals’ were included in the analysis by applying the operational definitions for each category of goals. GOAL SCORING PATTERNS 135
An accepted definition of possession was amended to reduce the number of ‘zero’ pass goals, which would have the effect of including more of the actions preceding goals in the analysis. Researchers have been criticized for not performing reliability tests and or using inappropriate tests to analyze data, so particular attention was paid to these aspects.
3.10.1 Inclusion of ‘Own Goals’
An ‘own goal’ may be conceded by a defender who is trying to prevent the ball going into the goal or it might be the result of a miskick or a deflection after a shot at goal or an attempted clearance; there are multiple causes. In research studies, ‘Own Goals’ are usually identified and place in a separate category and excluded from any further analysis.
(For example see, Breen et al., 2006; Grant et al., 1998; Taylor et al., 2002). In this study, the operational definitions were applied and each ‘own goal’ was categorized accordingly.
For example, if an ‘own goal’ was scored by a defender while trying to intercept or clear a cross, the goal was recorded in the ‘Crossing’ category. If a defender accidentally scored an ‘own goal’ while passing the ball to his goalkeeper, or a shot was deflected into the goal it would be recorded in ‘Other Methods’. If a defender scored an ‘own goal’ while trying to clear a pass behind the defense it would be recorded as a goal from a ball behind and strike (BB&S).
3.10.2 Amendment to the ‘Definition of Possession’
The problem with the currently accepted definition of possession (Pollard & Reep,
1997) is that it eliminates some goals from the research because goals that occur after a save from the goalkeeper, or from a ball contact by a defender are categorized as ‘zero’ pass goals. This definition of possession has the potential to distort the analysis of goal scoring patterns in three ways. The first example is because there will be an increase in the number of ‘zero’ pass goals, which in turn will increase the number of goals scored from GOAL SCORING PATTERNS 136
five passes or less. Secondly, it has the potential to increase the number of ‘1 pass’ goals because when a ball rebounds from a goalkeeper, a post or a defender it may be passed to a team-mate to score the goal. This level of detail is not reported in research papers. The third example is because ‘zero’ pass goals are usually eliminated from analysis of possession ‘type’. For example, in a study of goals from three different types of possession, described as counter attacks, elaborate attacks or set-play attacks, 203 of the 476 goals were allotted to either counter attack or elaborate attack (Tenga et al., 2010). That left 273 goals to be accounted for and it would be unprecedented for that many goals to be scored from set-play attacks. If 30% of the 476 goals had been scored from set play attacks, which is reasonable, (142 of the total 476 goals) that would still leave 131 goals (28% of the total) unaccounted for, or goals that may have been designated as ‘zero’ pass goals and therefore eliminated from the study. If the 142 goals had been accounted for it may have changed the outcome of the research.
Another study of scoring opportunities in the EPL tracked the origin of possession and type of feed preceding each goal (Wright et al., 2011).
“In a number of cases ‘NO Position of Feed’ was evident, which would indicate that ‘NO Feed’ occurred, that is a tackle, interception, turn-over or a deflected pass etc., in the area from which an attempt was made. This resulted in 47 goals, which accounted for 28% of all goals and 30% of all attempts on goal.” (Wright et al., 2011:442).
A total of 169 goals were analyzed in this study of which 57 were scored from Set
Plays including penalties. If the study had been on goals scored in Open Play the 47 goals with ‘NO Feed’ would represent 42% of the remaining 112 goals. It is accepted that there are legitimate ‘NO Feed’ or ‘zero’ pass goals, for example, when a player dispossesses an opponent and runs with the ball towards goal before scoring, or when a player intercepts a back pass and scores but a number of ‘NO Feed’ goals are a result of the accepted definition of possession. The point to be made is that a slight amendment to the definition GOAL SCORING PATTERNS 137
of possession will provide more detailed information about events preceding goals and more goals will be included in the research.
Hughes (1990) analyzed the events preceding the 53 ‘zero’ pass goals in his analysis of 109 matches, to count the number of passes in the move immediately before the goal. The results showed that 88% of the preceding passing sequences were from five passes or less, which made no difference to the outcome and point of his research. The fact that he failed to report how many goals were scored from Set Plays has been dealt with in the Literature Review, Chapter 2. However, if a study of goals from possession types
(Tenga et al., 2010) applied the same retrospective analysis as Hughes (1990), many of the goals from two, three or four pass moves would be included in the study and would most likely affect the results, because counter attacks are typically from a low passing sequence.
The evidence is indisputable that most goals are scored from inside the penalty area so how teams develop attacks to get there is of primary importance to coaches. What happens after an attack has reached the inside of the penalty area is of secondary importance when analyzing, for example, time in possession or where possession was regained or the categories of goals in a research project such as this one. These reasons justify making a slight amendment to a commonly accepted definition of possession. A more inclusive and accurate set of data, coupled with high degree of reliability of coding will ultimately improve the quality of results from the research.
3.11 Summary
This chapter has provided the rationale for three categories of goals in Open Play, which led to the creation of new areas on the field to record data. A detailed explanation was provided on the data collection process using Sportscode Elite software, how the data would appear in the Timeline and Code Matrix and the systematic approach to create a
Code Window that would allow accuracy of coding to be checked and guaranteed. The GOAL SCORING PATTERNS 138
main reason why so much detail was provided in the creation of the code window and the use of Sportscode Elite software was to enable other coaches who have access to the software to replicate the research methodology across different levels of football.
A thorough explanation was provided on the operational definitions for the coding of the goals and why the accepted definition of possession by Pollard and Reep (1997) was amended.
Justification was given for the use of Yule’s Q Test for intra-operator and inter- operator reliability.
In the next chapter, the results of the data analysis will be presented to answer the research questions with discussion to explain the outcomes of the investigation.
GOAL SCORING PATTERNS 139
Chapter 4 Results and Discussion
4.1 Introduction
This chapter presents the data collected with the methods described in Chapter 3.
These data are discussed here too. The presentation of results is in order of priority. The most important task was to answer the primary research questions 1 and 2 and to show comparative data, for the same tournaments and between tournaments, for example, the
FIFA World Cups or the English Premier League (EPL) and between the World Cup and the EPL. Sections 4.1 to 4.4.3 relate to Research questions 1, 2 and 3.The second task was to answer the secondary research questions that relate to the primary questions, for example, ‘Do the top and bottom teams score goals in similar proportions in each of the three categories and other aspects of performance? ‘Is there a statistically significant difference between goals in the three categories in each competition?’ or ‘Are more goals scored from 1–5 passes or 6 or more passes from regained possessions in the Back Third?’.
The third task was to answer questions of a comparative nature, for example, ‘Are more goals scored from regained possessions in the Final Third compared with previous research studies?’ The fourth task was to report unforeseen outcomes that emerged from the results of this research and the final task was to present the results of the tests for inter-operator and intra-operator reliability. The remaining sections in this chapter relate to the research questions 4–18, which are referred in the text. To complete the tasks and apply statistical tests the data needed to be prepared and analyzed.
4.2 Data preparation
This study reports data from four different competitions. Seven of the nine data sets compiled here are complete records of goal scoring. The two exceptions are data from the
English Premier League (EPL). In the 2001–02 and 2005–06 seasons, the total number of GOAL SCORING PATTERNS 140
goals in the analysis comprised 86% (858/101 and 813/944 respectively) of the total goals scored in those seasons.
The data for each of the three categories of goals in the World Cups and the
European Championships were recorded as the actual scores and as percentage scores for comparative reasons. The data for each of the five sub categories of goals in the BBSPS category were treated in the same way. The data for the EPL and ‘A’ League were normalized using percentages because of the incomplete data sets in the EPL and the lower number of games played in the ‘A’ League in 2008–09 season compared with the other two seasons when the number of teams increased from eight to ten. When comparisons were made between the competitions, for example, to compare the number of goals scored from regained possessions in each third of the field, all data were normalized into percentages before the statistical tests were implemented. The actual data for each competition were used for Pearson’s Chi Square Test.
4.3 Reliability test results
All goals in the 2012 European Championships were analyzed in June and
December 2012 by Operator 1 (the researcher) to test Intra-Observer Reliability. A test for
Inter-Observer Reliability was conducted in December 2012 between Operator 1 and
Operator 2.
4.3.1 Intra-Observer Reliability Test
The data entries for (a) categories of goals, (b) the number of passes leading to the goals, (c) the position on the field where the goals were scored (d) the areas of the field where possession was regained and (e) the areas of the field where passes were made for goals in the BBSPS category are in Table 4.1 and Appendix 19. The entries are shown in the extracts from the code matrices in Figure 4.1 and Figure 4.2. GOAL SCORING PATTERNS 141
Table 4.1 Intra-Reliability Test data for Operator 18 Data June 2012 Dec 2012 Data June 2012 Dec 2012 Set Plays 15 15 0–5 Passes 38 39 Open Play 61 61 6+ Passes 23 22 BB&S 21 21 Inside PA 53 55 BBP&S 9 7 <23 yards 6 4 BBSPS 30 28 >23 yards 2 2 OM 19 20 Reg. B3 18 16 Crosses 12 13 Reg. OH 9 11 OH 3 1 Reg. TH 18 19 WR 2 2 Reg. F3 16 15 Z14+ 22 22 WL 3 3
Figure 3.1 The Code Matrix for the European Championships in June 2012
8 BB&S- Ball behind and strike, BBP&S- Ball behind pass and strike, BBSPS- BB&S and BBP&S combined, OM-Other Methods, WR-Wide Right, WL-Wide Left, Z14+-Zone 14+ GOAL SCORING PATTERNS 142
Figure 4.2 The Code Matrix for the European Championships in December 2012
It is important to understand that there is a logical sequence of events for coding information. The first is to separate goals into Open Play or Set Plays. The allocation of goals to Set Plays becomes difficult when a defender has cleared the initial pass. If the ball is cleared well beyond the boundaries of the penalty area the following sequences of play are usually determined to be in Open Play but if the clearance is poor and the ball is still in the vicinity of the penalty area, before it is passed or shot into the goal, the sequence of play will be allocated to the Set Play in question, be it a corner or free kick. It is common for analysts to refer to first or second phase of play when describing goals from Set Plays.
In this example there was total agreement on the allocation of goals to Set Plays and Open
Play.
The second stage of analysis was to allocate the goals in Open Play to one of three categories, which included (the combined total of BB&S and BBP&S) BBSPS, Other
Methods and Crosses on the agreed total of 61 goals. In Appendix 19, the Yule’s Q-Test result shows the operator has an accuracy score of 99.3% to discriminate between goals in the BBSPS and Other Methods categories, when there was agreement on 28 out of 30 goals in BBSPS and 19 out of 20 goals in Other Methods. GOAL SCORING PATTERNS 143
The next stage was to code the areas of the field for goals in the BBSPS category.
There was agreement on 28 goals on both dates but in June a total of 30 goals were coded.
The columns in Appendix 19 show agreement on both dates for Wide Right (2), Zone 14+
(22) and Wide Left (3) but disagreement for Own Half (3 and 1). The figures of 3 and 1 do not represent a coding error; they actually represent the number of goals in the BBSPS category on the two dates in question. There is total agreement in the coding of the areas of the field where the pass was made in the BBSPS category.
The purpose of an Intra-reliability test is to measure the ability of the operator to discriminate between each category of events so the process of comparing pairs of categories was applied when it was appropriate to do so. For example it would not be appropriate to compare the allocation of goals scored Inside the Penalty Area with goals from more than 23 yards because the areas are separated by a space of 5 yards. Similarly, it would not be appropriate to compare Regained Possessions in the Back Third with
Regained Possessions in the Final Third because the space between each area is approximately 40 yards or 35 meters.
The most appropriate comparisons are between categories of goals and adjacent areas of the field, for example Wide Right and Zone 14+, or the Back Third and Own Half of Midfield. The most difficult decisions are when an event takes place on the border of two areas and there is not a marking on the field to assist the operator, for example between the Back Third and the Middle Third or Wide Right and Zone 14+. The solution in this investigation was to code and group those situations when they occurred, review them a second time and if a decision could not be made, divide the number of events and allocate half to each category, for example Back Third and Own Half of Midfield. In such situations it has to be acknowledged that there is not a satisfactory solution to the problem and an acceptable accuracy score may need to be lower than the conventional 95%. For GOAL SCORING PATTERNS 144
example, when Regained Possessions in the Back Third and Own Half of Midfield were compared there was agreement for 16 incidents in the Back Third and 9 in Own Half of
Midfield and disagreement for 2 incidents in the Back Third and 2 in Own Half of
Midfield, producing an accuracy score of 94.6%. Due to the difficulty of accurately locating an incident precisely on the field of play, because of camera angles and close up views of the action an accuracy level of 90 percent should be acceptable in a reliability test using television footage.
It should be recognized that Yule’s Q-Test produces a score of 100% when two categories are compared and one of them has complete agreement, due to the arithmetic formula. For example, comparing the goals scored between 18 and 23 yards (<23) with goals scored outside 23 yards (>23), there was disagreement for 2 goals between 18 and 23 yards but complete agreement for goals scored outside 23 yards; Yule’s Q-Test produced an accuracy score of 100%, so this test should only be used when there is not complete agreement in two categories. If Percentage Error was used to measure accuracy for 4 out of
6, a score of 67% would result due to the very low number of goals, but when goals between 18 yards and 23 yards (6 and 4) were compared with goals scored Inside the
Penalty Area (53 and 55) when there was disagreement in both categories, Yule’s Q-Test produced an accuracy score of 96.3%.
When measuring accuracy of coding events with less than ten incidents, one mistake will produce an accuracy score of 90% or less. An operator should not be condemned in this situation if an accuracy score of more than 95% is recorded when coding several events with more than ten incidents. An operator should be assessed on the overall performance as well as on isolated events.
In summary, the Intra-Observer Reliability Test showed that there were six categories of events where 100% accuracy was reached when the European GOAL SCORING PATTERNS 145
Championships were coded in June and December 2012. The accuracy of coding the remaining events ranged between 94.6% and 99.8%, which was within the accepted range as indicated in the literature.
4.3.2 Inter-Observer Reliability Test
The data entries by both operators for (a) categories of goals, (b) the number of passes leading to the goals, (c) the position on the field where the goals were scored, (d) the areas of the field where possession was regained and (e) the areas of the field where passes were made for goals in the BBSPS category are in Table 4.2 and Appendix 20. The data entry for Operator 2 is shown in the extract of the code matrix; see Figure 4.3
Table 4.2 Inter-Reliability Test data for Operator 1 and Operator 29 Data Operator 1 Operator 2 Data Operator 1 Operator 2 Dec 2012 Dec 2012 Dec 2012 Dec 2012 Set Plays 15 15 0–5 Passes 39 38 Open Play 61 61 6+ Passes 22 23 BB&S 21 19 Inside PA 55 53 BBP&S 7 8 <23 yards 4 4 BBSPS 28 27 >23 yards 2 4 OM 20 22 Reg. B3 16 16 Crosses 13 12 Reg. OH 11 12 OH 1 2 Reg. TH 19 21 WR 2 2 Reg. F3 15 12 Z14+ 22 21 WL 3 2
9 BB&S- Ball behind and strike, BBP&S- Ball behind pass and strike, BBSPS- BB&S and BBP&S combined, OM-Other Methods, WR-Wide Right, WL-Wide Left, Z14+-Zone 14+
GOAL SCORING PATTERNS 146
Figure 4.3 The Code Matrix for the European Championships in December 2012 for Operator 2
The results of the initial coding by both operators are shown in Table 4.2 and in five categories there was complete agreement. Appendix 20 has the results of Yule’s Q-
Test, which was used to calculate the accuracy of coding of events when there were differences in the number of incidents.
With the exception of one event the accuracy of coding between both operators for the remaining events ranged between 94.9% and 99.8%. The event, which had the lowest coding accuracy score of 90.9%, was between the areas of the field, Own Half and Z14+.
In summary an accuracy score of 95% and higher was reached in twelve of the fourteen comparisons with two between 90.9% and 94.9%.
4.4 Research question 1
The first question is:
Are more goals scored from passing the ball behind opponents or to a player level
with the last defender than from Crosses or Other Methods? GOAL SCORING PATTERNS 147
Tables 4.3 and 4.4 show the results for each category of goals in three FIFA World Cups between 2002 and 2010 and three UEFA European Championships between 2002 and
2012.
Table 4.3 Goals Scored in FIFA World Cups Categories of Goals World Cup World Cup World Cup 2002 2006 2010 Ball Behind & Strike (BB&S) 41 36 39 Ball Behind Pass & Strike 8 9 15 (BBPS) Ball Behind Combined (BBSPS) 49 43% 45 48% 54 51% Crosses 32 28% 12 13% 14 13% Other Methods 34 29% 36 39% 38 36% Goals in Open Play 115 93 106 Goals from Set Plays 46 54 39 Total Goals 161 147 145
The percentage for each category of goals was calculated in relation to the total number of goals scored in Open Play. The figure for goals from Set Plays is provided to account for all goals scored in each tournament.
GOAL SCORING PATTERNS 148
Table 4.4 Goals Scored in the European Championships Categories of Goals Euro 2004 Euro 2008 Euro 2012 Ball Behind & Strike (BB&S) 17 23 21 Ball Behind Pass & Strike 9 11 9 (BBPS) Ball Behind Combined (BBSPS) 26 53% 34 57% 30 49% Crosses 7 14% 17 28% 12 20% Other Methods 16 33% 9 15% 19 31% Goals in Open Play 49 60 61 Goals from Set Plays 28 17 15 Total Goals 77 77 76
Table 4.5 Goals scored in the English Premier League (EPL) Categories of Goals EPL 2001–02 EPL 2005–06 EPL 2011–12 Ball Behind & Strike (BB&S) 221 209 226 Ball Behind Pass & Strike 50 47 81 (BBPS) Ball Behind Combined (BBSPS) 271 42% 256 44% 307 39% Crosses 106 17% 96 16% 132 17% Other Methods 264 41% 230 40% 353 44% Goals in Open Play 641 582 792 Goals from Set Plays 217 231 274 Total Goals 858 86% 813 86% 1066 All
Tables 4.5 and 4.6 show the results for each category of goals in three English
Premier League (EPL) season between 2001–02 and 2011–12 and three Australian ‘A’
League seasons between 2008–09 and 2011–12.
GOAL SCORING PATTERNS 149
Table 4.6 Goals Scored in the Australian ‘A’ League Categories of Goals A LGE 2008– A LGE 2009– A LGE 2011–12 09 10 Ball Behind & Strike (BB&S) 55 73 80 Ball Behind Pass & Strike 17 23 31 (BBPS) Ball Behind Combined (BBSPS) 72 43% 96 39% 111 42% Crosses 35 21% 53 21% 55 21% Other Methods 59 36% 98 40% 97 37% Goals in Open Play 166 247 263 Goals from Set Plays 83 99 102 Total Goals 249 346 365
The combined total for goals from passing the ball behind opponents or to a player level with the last defender (BBSPS) is higher than the figure for the other two categories,
Crosses and Other Methods in two of the three seasons in both the EPL, 2001–02 and
2005–06 and the ‘A’ League, 2008–09 and 2009–10.
4.4.1 Test results for categories of goals
In two of the four tests there was a normal distribution and in each test a significant difference was identified in two of the three categories of goals. In the World Cup and
European Championships, there was a statistically significant difference between goals from BBSPS and Other Methods but not in the EPL and ‘A’ League. In all four competitions there was a statistically significant difference between goals from BBSPS and
Crosses. In the European Championships there was a significant difference between goals from BBSPS and Other Methods but not in the other three competitions. In the EPL and the ‘A’ League there was a statistically significant difference between Other Methods and
Crosses but not in the World Cups and European Championships. In all tests the assumption of equal variance was true. Appendices 1 and 2 show the results of each test on GOAL SCORING PATTERNS 150
the data. Two examples are provided for data normally distributed and data not distributed normally.
Example 1 – Data normally distributed
The data for the European Championships between 2004 and 2012 (Table 4.4) were normalized and the assumption of normality and equal variance between the scores was assessed through the Shapiro-Wilk Test (Statistic = BBSPS, 1.000, df = 3, p–value = 1.000,
Statistic = OM, .832, df = 3, p–value = .194, Statistic = Crosses, .993, df = 3, p–value
= .843) and Leven’s Test (Statistic = 1.767, df1 = 2, df2 = 6, p–value = .249). The one-way
ANOVA result (p-value = .004) indicated a statistically significant difference between the categories so Bonferroni’s Test was applied and found the differences were between
BBSPS and Other Methods, (p-value = .013) and BBSPS and Crosses, (p-value = .005); see Table 4.7.
Table 4.7 Statistical Test results for categories of goals in the European Championships10 Kolmogorov - Smirnova Shapiro-Wilk
category Statistic df Sig. Statistic df Sig. Scored BBSPS .175 3 . 1.000 3 1.000 OM .349 3 . .832 3 .194 CROSS .204 3 . .993 3 .843 a. Lilliefors Significance Correction
Test of Homogeneity of Variances scores Levene df1 df2 Sig. Statistic 1.767 2 6 .249
10 BBSPS- BB&S and BBP&S combined, BB&S- Ball behind and strike, BBP&S- Ball behind pass and strike OM-Other Methods
GOAL SCORING PATTERNS 151
Table 4.7 continued ANOVA scores Sum of Mean Squares df Square F Sig. Between groups 1788.667 2 894.333 16.494 .004 Within Groups 325.333 6 54.222 Total 2114.000 8
Post Hoc Tests Multiple Comparisons11 Dependent Variable:scores Bonferroni 95% Confidence Interval Mean (I) (J) Difference Std. Lower Upper category category (I-J) Error Sig. Bound Bound BBSPS OM 26.667* 6.012 .013 6.90 46.43 CROSS 32.333* 6.012 005 12.57 52.10 OM BBSPS –26.667* 6.012 .013 –46.43 –6.90 CROSS 5.667 6.012 .000 –14.10 25.43 CROSS BBSPS –32.333* 6.012 .005 –52.10 –12.57 OM –5.667 6.012 .000 –25.43 14.10
Pearson’s Chi Square test (with = 0.05) was used to evaluate whether a change in strategy existed between goals scored in the three categories in the European
Championships over three tournaments. The Chi Square test was not statistically significant (4,7.202, p = .126); see Table 4.8.
11 BBSPS- BB&S and BBP&S combined, BB&S- Ball behind and strike, BBP&S- Ball behind pass and strike OM-Other Methods
GOAL SCORING PATTERNS 152
Table 4.8 Chi Square test results for categories of goals in European Championships Asymptotic Significance Value df (2-sided) Pearson Chi Square 7.202 4 .126 Likelihood Ratio 7.594 4 .108 Linear-by-Linear Association .064 1 .800 N of Valid Cases 170 a. Tournament = EURO b. 0 cells (0.0%) have expected count less than 5. The minimum expected count is 10.38.
Example 2 – Data not normally distributed
The data for the World Cups between 2002 and 2010 (Table 4.1) were normalized and
the assumption of normality between the scores was assessed through the Shapiro-Wilk
Test (Statistic = BBSPS, .980, df = 3, p-value = .726, Statistic = OM, .949, df = 3, p-
value = .567, Statistic = Crosses, .750, df = 3, p-value = <.001). The assumption of
equal variance between the scores was assessed through Leven’s Test (Statistic = 2.056,
df1 = 2, df2 = 6, p-value = 0.209). The data were not normally distributed so the
Kruskal-Wallis Test was applied and showed a significant difference existed in at least
one of the groups, ( = 7.261, p-value = .027). The one-way ANOVA result (p-value
= .004) indicated a statistically significant difference existed between the categories so
Bonferroni’s Test was applied and found the difference was between BBSPS and
Crosses, (Diff = 29.333, p-value = .004). See Table 4.9.
GOAL SCORING PATTERNS 153
Table 4.9 Statistical Test Results for categories of goals in World Cups Test of Normality Kolmogorov - Smirnova Shapiro-Wilk
category Statistic df Sig. Statistic df Sig. scored BBSPS .232 3 . .980 3 .726 OM .269 3 . .949 3 .567 CROSS .385 3 . .750 3 .000 a. Lilliefors Significance Correction Test of Homogeneity of Variances scores Levene Statistic df1 df2 Sig. 2.056 2 6 .209
ANOVA scores Sum of Mean Squares df Square F Sig. Between groups 1298.667 2 649.333 16.555 .004 Within Groups 235.333 6 39.222 Total 1534.000 8
Post Hoc Tests Multiple Comparisons Dependent Variable:scores Bonferroni 95% Confidence Interval Mean (I) (J) Difference Std. Lower Upper category category (I-J) Error Sig. Bound Bound BBSPS OM 12.667* 5.114 .114 –4.14 29.48 CROSS 29.333* 5.114 .004 12.52 46.14 OM BBSPS –12.667* 5.114 .114 –29.48 4.14 CROSS 16.667 5.114 .052 –.14 33.48 CROSS BBSPS –29.333* 5.114 .004 –46.14 –12.52 OM –16.667 5.114 .052 –33.48 .14 *. The mean difference is significant at the 0.05 level.
GOAL SCORING PATTERNS 154
Kruskal-Wallis Test Ranks category N Mean Rank Scores BBSPS 3 8.00 OM 3 5.00 CROSS 3 2.00 Total 9
Test Statisticsa,b scores Chi Square 7.261 df 2 Asymp.Sig. .027 a.Kruskal Wallis Test b. Grouping Variable: category
Pearson’s Chi Square test (with = 0.05) was used to evaluate whether a change in strategy existed between goals scored in the three categories of goals in the World cups over three tournaments. The Chi Square test was statistically significant (4,10.805, p
= .029); see Table 4.10.
GOAL SCORING PATTERNS 155
Table 4.10 Chi Square Test Results for categories of goals in World Cups TYPE * YEAR Crosstabulation YEAR 2002 2006 2010 Total TYPE BBSPS Count 49 45 54 148 Expected Count 54.2 43.8 50.0 148.0 % within TYPE 33.1% 30.4% 36.5% 100.0% % within YEAR 42.6% 48.4% 50.9% 47.1% Adjusted Residual –1.2 .3 1.0 CROSS Count 32 12 14 58 Expected Count 21.2 17.2 19.6 58.0 % within TYPE 55.2% 20.7% 24.1% 100.0% % within YEAR 27.8% 12.9% 13.2% 18.5% Adjusted Residual 3.2 –1.6 –1.7 OM Count 34 36 38 108 Expected Count 39.6 32.0 36.5 108.0 % within TYPE 31.5% 33.3% 35.2% 100.0% % within YEAR 29.6% 38.7% 35.8% 34.4% Adjusted Residual –1.4 1.0 .4 Total Count 115 93 106 314 Expected Count 115.0 93.0 106.0 314.0 % within TYPE 36.6% 29.6% 33.8% 100.0% % within YEAR 100.0% 100.0% 100.0% 100.0% a. Tournament = WC
Chi Square Testa Value df Asymptotic Significance (2-sided) Pearson Chi Square 10.805 b 4 .029 Likelihood Ratio 10.454 4 .033 Linear-by-Linear Association .026 1 .872 N of Valid Cases 314 a. Tournament = WC b. 0 cells (0.0%) have expected count less than 5. The minimum expected count is 17.18.
GOAL SCORING PATTERNS 156
4.4.2 Discussion – Research question 1
The combined total for goals from passing the ball behind opponents or to a player level with the last defender (BBSPS) is higher than the figure for the other two categories,
Crosses and Other Methods in each World Cup and European Championship.
The results show that passing the ball behind opponents or to a player level with the last defender (BBSPS) was the most dominant method of scoring goals in six international tournaments and in four out of six seasons in two professional leagues.
The percentage difference between goals from BBSPS and Other Methods (OM) was greater in international tournaments than in professional leagues (Table 4.7). The range in international tournaments was between 9% and 42%, compared with a range between 1% and 7% in the professional leagues. The only tournament where a statistically significant difference was found between BBSPS and Other Methods was in the UEFA
European Championships. Bonferroni’s test showed a statistically significant difference (p- value = .013). If the one ‘outlier’, the 42%, is ignored the results in each competition do not vary much from one end of the spectrum to the other, which suggests the goal scoring patterns have not changed a great deal over a period of ten years.
The only statistically significant Chi Square test result (Table 4.10) in all competitions was in the World Cups, which showed a change in the strategy of scoring goals in the category of ‘Crosses’ in the 2002 tournament compared with 2006 and 2010; adjusted residual (3.2).
In the two instances when more goals were scored from Other Methods than from
BBSPS (EPL 2011–12 and ‘A’ League 2009–10) the range was between –1% and –5%.
The greatest range within a competition was between 18% and 42% in the UEFA
European Championships (EURO 2008). In that tournament there was an unusually high number of goals from Crosses (28%, see Table 4.4), which subsequently increased the GOAL SCORING PATTERNS 157
difference between goals from BBSPS and Other Methods. The World Cup results show a range between 9% and 15%, which was considerably higher than the results in the two professional leagues where the range was between 1% and 7% for four of the six events.
Table 4.11 Percentage difference between goals from BBSPS and Other Methods in all competitions12 World Cups 2002 2006 2010 BBSPS 43% 48% 51% Other Methods 29% 39% 36% % Difference 14% 9% 15% European Championships 2004 2008 2012 BBSPS 53% 57% 49% Other Methods 33% 15% 31% % Difference 20% 42% 18% English Premier League 2001–02 2005–06 2011–12 BBSPS 42% 44% 39% Other Methods 41% 40% 44% % Difference 1% 4% –5% Australian ‘A’ League 2008–09 2009–10 2011–12 BBSPS 43% 39% 42% Other Methods 36% 40% 37% % Difference 7% –1% 5%
It is important to note that the results represent a time frame between five and ten years and two distinctly different types of competition; international tournaments are completed in a matter of weeks but professional leagues are played over eight to nine months and involve many more matches. International tournaments start with group matches, which teams must win to progress to the knock-out stages and from then onwards teams face elimination in a subsequent round or reach the final of the tournament with a maximum of seven games. International tournaments therefore encourage risk taking and a
‘must win’ mind set in many games, whereas league football is played over months and
12 BBSPS- BB&S and BBP&S combined, BB&S- Ball behind and strike, BBP&S- Ball behind pass and strike
GOAL SCORING PATTERNS 158
involves a much higher number of matches, for example, 38 in the EPL and 27 in the ‘A’
League followed by a Final-Series.
League football involves promotion and relegation and matches are played at home and away against stronger teams or weaker teams. These variables tend to encourage teams to take a more attacking or defensive approach and their strategies might change from week to week depending on the league position of each team and the current form. Another factor to be considered is the quality of players and the level of skill required to create and score goals by passing the ball behind opponents into areas where space is usually restricted by opponents. One might expect a higher standard of player at the international level than at club level. These factors combined might explain the difference in the results between international and club football for goals scored from BBSPS and Other Methods.
4.4.3 The top three teams compared with the bottom three teams and the average score for BBSPS in the league competitions
It is interesting to compare the results of the teams that finish in the top three positions of a league with those at the bottom and how they compare with the league average. In tournaments such as the World Cup the teams that make the quarterfinals and beyond are often compared with the teams that fail to qualify from the group stage. The outcomes of this type of comparison have more meaning in league football than international football because the data are from a larger sample of matches. For example, a team can win the World Cup and play seven matches in less than four weeks but to win the
English Premier League (EPL) a team has to play thirty-eight matches over a period of ten months. Table 4.12 and Table 4.13 show a comparison of the top three teams in the ‘A’
League and the EPL with the bottom three teams and the average league percentage for goals scored in each category in Open Play. The number of goals scored from Set Plays
(SP) is shown to provide the context of total goals scored in the 2011–12 season. The GOAL SCORING PATTERNS 159
lower section of Tables 4.12 and 4.13 show the breakdown of goals scored from BBSPS in the areas of the field where the pass was made, which will be referred to when discussing
Research Question 2.
In the two examples from the professional leagues two of the top three teams scored the majority of their goals from BBSPS compared with Other Methods and Crosses.
In the ‘A’ League, Perth Glory in 3rd position reversed the trend by scoring 40% of their goals from Other Methods, as did Manchester United in 2nd position in the EPL by scoring
49%. In both examples the Central Coast Mariners (CCM) and Brisbane (BRS) achieved a higher percentage from BBSPS, 52% and 51% respectively than the League average of
42%. In the EPL example Manchester City (MCT) and Arsenal (ARS) scored 49% and
55% respectively from BBSPS, which was higher than the League average of 39%.
Table 4.12 Comparison of top and bottom three teams in the ‘A’ League in 2011–1213 1st CCM 2nd BRS 3rd PER 8th MVT 9th ADL 10th Lge. Ave GCU BBSPS 16 52% 20 51% 6 24% 8 32% 7 47% 7 33% 42% OM 10 32% 14 36% 10 40% 13 52% 6 40% 10 48% 37% CROSS 5 16% 5 13% 9 36% 4 16% 2 13% 4 19% 21% Set Plays 9 11 15 10 11 9 Total 40 50 40 35 26 30 OH 1 1 0 2 1 0 WR 5 0 2 1 0 1 Z14+ 8 50% 17 85% 4 67% 4 50% 5 71% 5 71% 65% WL 0 0 0 0 1 0 IPA 2 2 0 1 0 0 TOTAL 16 20 6 8 7 7
13 BB&S- Ball behind and strike, BBP&S- Ball behind pass and strike, BBSPS- BB&S and BBP&S combined, OM-Other Methods, WR-Wide Right, WL-Wide Left, Z14+-Zone 14+
GOAL SCORING PATTERNS 160
Table 4.13 Comparison of top and bottom three teams in the EPL in 2011–12 1st MCT 2nd MUT 3rd ARS 18th BOL 19th 20th Lge. BLK WOL Ave BBSPS 34 49% 22 32% 36 55% 7 22% 7 32% 11 39% 39% OM 30 44% 34 49% 20 31% 13 41% 12 55% 13 47% 44% CROSS 5 7% 13 19% 9 14% 12 37% 3 13% 4 14% 7% SP 23 20 9 14 25 11 Total 92 88 74 46 47 39 OH 1 1 1 0 0 3 WR 2 2 5 0 0 3 Z14+ 23 68% 10 45% 25 69% 5 71% 6 86% 3 27% 62% WL 4 1 2 1 1 3 IPA 4 8 3 1 0 2 Total 34 22 36 7 7 11
Two of the top three teams, Manchester United and Liverpool scored more than the league average in the 2005–06 season but only Arsenal of the top three teams scored above the league average in 2001–02 season. In the ‘A’ League over the three seasons, two teams scored more than the league average in 2008–09 and 2011–12 seasons and all the top three teams scored higher than the league average in 2009–10 in the BBSPS category. In the
2009–10 ’A’ League season the highest percentage of goals was scored in the Other
Methods category (40%) followed by BBSPS (39%).
In the 2011–12 season, Table 4.12, Adelaide (ADL) were the only team in the bottom three clubs to score more than the ‘A’ League average with 47%, while in seasons
2008–09 one team scored above the average and in 2009–10 two of the bottom three teams scored above the league average in the BBSPS category. In the EPL example, Table 4.13,
Wolverhampton Wanderers were the only team in the bottom three teams to equal or better the EPL average of 39%, while Ipswich were the only other team to better the average score in the 2001–02 season. Appendix 3 shows all the results for the top and bottom three teams for each category of goals for each season in the EPL and ‘A’ League in this study.
Overall 8 of the 9 teams that reached the top three positions during the three seasons of the ‘A’ league and 5 of the 9 teams in the top three positions in the EPL scored GOAL SCORING PATTERNS 161
the highest number of goals in the BBSPS category, compared with Other Methods or
Crosses.
Overall 7 of the 9 A League teams and 5 of the 9 teams in the EPL in the top three positions scored a higher percentage of goals in the BBSPS category than the average for the league.
Of the teams that finished in the bottom three positions in both leagues, 3 of the 9 teams in the ‘A’ League scored the highest number of goals in the BBSPS category and one team equaled or scored a higher percentage than the league average. In the EPL 2 of the 9 teams in the bottom three positions equaled or had the highest number of goals in the
BBSPS category and one team had a higher score than the league average.
These results show that the majority of teams, 13 out of 18 (72%), that finished in the top three positions in both professional leagues scored more goals in the BBSPS category than Other Methods or Crosses.
The majority of teams, 13 out of 18 (72%), that finished in the bottom three positions scored more goals through Other Methods or Crosses than in the BBSPS category.
The average tenure for a manager in the EPL is less than two years, so it is rare to find a club that had the same manager for more than ten years but Alex Ferguson and
Arsene Wenger were the managers of Manchester United (MU) and Arsenal (ARS), respectively during the period of this investigation into goal scoring patterns. Table 4.14 shows the goal scoring patterns of both teams during the three seasons in question, however in season 2005–06 Arsenal were not in the top three teams.
GOAL SCORING PATTERNS 162
Table 4.14 Goals in categories for Manchester United and Arsenal for three seasons over ten years14 Teams & Goals in OP BBSPS Other Crosses Lge. Ave. BBSPS Methods Arsenal 2001–02 (59) 33 56% 18 30% 8 14% 42%
Arsenal 2005–06 (49) 30 61% 17 35% 2 4% 44% Arsenal 2011–12 (65) 36 55% 20 31% 9 14% 39%
Man Utd. 2001–02 20 36% 25 45% 8 14% 42% (53) Man Utd. 2005–06 32 56% 19 33% 6 11% 44% (57) Man Utd. 2011–12 22 32% 34 49% 13 19% 39% (68)
The results show clearly that ARS consistently scored more goals from BBSPS
than Other Methods and by a considerable margin of between 24–26%, whereas MU
scored more goals from BBSPS, on one occasion 2005–06, by 23%. In the other two
seasons MU scored more goals from Other Methods than BBSPS by a margin of 9% –17%.
It is unlikely that Alex Ferguson changed his attacking strategies after continual success in
winning league titles so the results may reflect a more flexible approach in MU’s attacking
style. The contrasting results for MU also may be a consequence of their success in the
EPL because opposing teams would often defend deep in their own half making it very
difficult for MU to get behind them. With so many talented players in the team they would
score goals through Other Methods and between 10–20% of goals from Crosses. Arsenal’s
results may be explained by the fact they are and have been recognized as a team that
attacks with speed and finesse and have been criticized by commentators for trying to pass
their way through the center of the field too often and for not having a more flexible
approach to scoring goals. They have scored goals more consistently from BBSPS than
14 BBSPS- BB&S and BBP&S combined, BB&S- Ball behind and strike, BBP&S- Ball behind pass and strike, OP-Open Play
GOAL SCORING PATTERNS 163
Other Methods with a lower number from Crosses (Appendix 3). The margin between winning the EPL and finishing in the top three is fine and it may be that a less rigid approach to scoring goals is what makes the difference along with a reasonably high number of goals from Set Plays because they often decide the outcome of very close matches.
4.4.4 The top four teams in international tournaments compared with the average score for the BBSPS category
In the two examples, Table 4.15 and Table 4.16, from international tournaments the percentage of goals scored by the top four teams by BBSPS, Other Methods and Crosses are compared with the average score for each category for the tournament. In the 2010
World Cup, Germany was the only team in the top four to score more than the tournament average for BBSPS.
Table 4.15 Comparison of top 4 teams with the average in the 2010 World Cup15 1st SPA 2nd HOL 3rd GER 4th URU Average
BBSPS 2 29% 4 36% 10 77% 2 29% 53% OM 5 71% 5 46% 3 23% 2 29% 33% CROSS 0 2 18% 0 3 42% 14% Set Plays 1 1 2 4 Total 8 12 15 11 OH 0 0 2 0 WR 0 1 4 0 Z14+ 1 50% 2 50% 3 33% 1 50% 59% WL 0 1 1 1 IPA 1 0 0 0 Total 2 4 10 2
Appendix 4 shows the results for the top four teams in each World Cup and
European Championships for goals in each category and the average for the tournament.
15 BBSPS- BB&S and BBP&S combined BB&S- Ball behind and strike, BBP&S- Ball behind pass and strike, OM-Other Methods, WR-Wide Right, WL-Wide Left, Z14+-Zone 14+, IPA-Inside Penalty Area
GOAL SCORING PATTERNS 164
The results of top four teams are compared with the average for the tournament rather than the results of the bottom teams that failed to progress beyond the group stage. Five teams scored two goals in Open Play, three teams failed to score and the other eight teams only scored one goal. In two of the three World Cups only one of the top four teams scored higher than the tournament average for goals in the BBSPS category; Brazil (54% over
43%) in 2002 scored 7 out of 13 goals and Germany (77% over 51%) in 2010 scored 10 out of 13 goals. In 2006 two of the top four teams scored higher than the average in the
BBSPS category; Italy and France (60% over 48%) both scored 3 out of 5 goals in Open
Play to reach the final and semi-final respectively.
Table 4.16 Comparison of top 4 teams with the average in the 2008 European Championships16 1st SPA 2nd GER 3rdTUR 4th RUS Average
BBSPS 9 82% 4 57% 4 50% 1 20% 57% OM 0 1 14% 2 25% 1 20% 15% CROSS 2 18% 2 29% 2 25% 3 50% 28% Set Plays 1 3 0 2 Total 12 10 8 7 OH 2 0 0 0 WR 0 0 2 0 Z14+ 6 67% 3 75% 1 25% 1 100% 53% WL 0 1 1 0 IPA 1 0 0 0 Total 9 4 4 1
The results in the European Championships are slightly better than the World Cup, with seven teams scoring equal with or higher than the tournament average in BBSPS;
Republic of Czechoslovakia (75% over 53%) scored 6 out of 8 goals in 2004, Spain (82% over 57%) scored 9 out of 11 goals in 2008 and the top four teams in 2012, Spain, Italy,
Germany and Portugal, while Germany were equal with 57% in 2008. See Appendix 4.
16 BBSPS- BB&S and BBP&S combined BB&S- Ball behind and strike, BBP&S- Ball behind pass and strike, OM-Other Methods, WR-Wide Right, WL-Wide Left, Z14+-Zone 14+, IPA-Inside Penalty Area
GOAL SCORING PATTERNS 165
In the three World Cups 7 of the 12 teams in the top four places and 9 of the 12
teams in the top four places in the European Championships equaled or scored more goals
in the BBSPS category compared with Other Methods or Crosses. The results also show
that 4 of the 7 teams in World Cups and 6 of the 9 teams in the European Championships
equaled or scored more goals than the tournament average in the BBSPS category. These
results show a trend in goal scoring patterns for the top teams that is the same as the
overall trend for all international tournaments in this study. There are examples where a
team has won by scoring the majority of goals by Other Methods; Spain won the 2010
World Cup and Greece won the 2004 European Championship, but these are exceptions.
Analysis of the teams in the international tournaments within this study, shows that
Germany reached the top four in the three World Cups between 2002 and 2010 and two of
the three European Championships since 2004, while Spain reached the final in World Cup
2010 and in Euro 2004 and 2008. Table 4.17 shows how both teams scored their goals.
Table 4.17 Goals and categories for Germany and Spain17 BBSPS Other Cross Set Plays BBSPS Methods Average
SPAIN EURO 12 8 80% 2 20% 0 2 49% EURO 08 9 82% 0 2 18% 1 57% WC 2010 2 29% 5 71% 0 1 51%
GERMANY EURO 12 5 63% 2 25% 1 12% 2 49% EURO 08 4 57% 1 14% 2 29% 3 57% WC 2010 10 77% 3 23% 0 2 51% WC 2006 4 33.3% 4 33.3% 4 33.3% 2 48% WC 2002 4 40% 0 6 60% 3 43%
17 BBSPS- BB&S and BBP&S combined BB&S- Ball behind and strike, BBP&S- Ball behind pass and strike, OM-Other Methods
GOAL SCORING PATTERNS 166
Since 2002, Germany has scored an equal number or more goals from the BBSPS category than Other Methods in World Cups and European Championships but has not won a tournament. On three out of four occasions they scored a higher percentage than the tournament average in 2012, 2010 and 2008. Spain had a very successful period between
2008 and 2012 winning the European Championships in 2008 and 2012 and the World
Cup in 2010. In the two European Championships, Spain scored more than 80% of their goals from BBSPS, which was much higher than the tournament average with 11 goals in
Open Play in 2008 and 10 goals in 2012 in six matches. The odd thing is that in between the European successes Spain won the World Cup in 2010 but only scored 7 goals in 7 matches with the majority of them through Other Methods, not BBSPS. So what happened? The fact is that when a team is successful as Spain were in Euro 2008 and after winning every qualification match for the 2010 World Cup, opponents accept that Spain will be the favorite to win most matches. So when teams played Spain they may have prepared to defend for most of the match and play for a draw, with the possibility that they might score with a counter attack or from a Set Play and win the game against all odds.
This mentality is encouraged in international tournaments and particularly in the World
Cup because in the event of a drawn game in the Round of 16, the quarterfinals and semi finals a penalty shoot-out will determine which team progresses to the next round or the final. The net effect is that a team such as Spain, who scored the majority of their goals in the BBSPS category, may be prevented from doing so by the opposition’s defensive strategy. The net effect was Spain scored fewer goals in total and in this example scored more through Other Methods. It is worth noting that in Euro 2008 Spain beat Italy in the quarter final by winning the penalty-shoot out after the teams were tied 0–0 after extra- time and in Euro 2012 Spain did it again against Portugal in the semi-final after a 0–0 draw. GOAL SCORING PATTERNS 167
The reality is that in both tournaments Spain actually scored their high number of goals in
five matches not six.
4.5 Research question 2
The second question is:
In which area of the field does the majority of passes originate that lead to goals from
passing the ball behind the opposing defence?
Tables 4.18 to 4.21 show the number and percentage of goals scored in each
competition from passing the ball behind opponents or to a player level with the last
defender from each of the five zones in category (a) BBSPS.
Table 4.18 Goals in the World Cup from a pass in each of the five Zones World Cup 2002 World Cup 2006 World Cup 2010
Categories BB&S BBP&S BB&S BBP&S BB&S BBP&S
Own Half - OH 3 (8.5%) 1 4 (11%) 1 3 (9%) 2 Wide Right - WR 3 (12%) 3 3 (7%) 0 5 (15%) 3 Zone 14+ – Z14+ 27 (61%) 3 26 (69%) 5 25 (59%) 7 Wide Left - WL 4 (10%) 1 0 (4%) 2 3 (11%) 3 Inside Pen. Area - IPA 4 (8.5%) 0 3 (9%) 1 3 (6%) 0
Sub Total 41 8 36 9 39 15 Total % in Open Play 49/115 = 43% 45/93 = 48% 54/106 = 51%
GOAL SCORING PATTERNS 168
Table 4.19 Goals in European Championships from a pass in each of the five Zones EURO 2004 EURO 2008 EURO 2012
Categories BB&S BBP&S BB&S BBP&S BB&S BBP&S
Own Half - OH 1 (12%) 2 1 (11%) 3 1 (10%) 2 Wide Right - WR 1 (7%) 1 2 (9%) 1 2 (7%) 0 Zone 14+ – Z14+ 10 (57%) 5 14 (53%) 4 16 (73%) 6 Wide Left - WL 2 (12%) 1 4 (21%) 3 2 (10%) 1 Inside Pen. Area - 3 (12%) 0 2 (6%) 0 0 (0%) 0 IPA Sub Total 17 9 23 11 21 9
Total % in Open Play 26/49 = 53% 34/60 = 57% 30/61 = 49%
Table 4.20 Goals in the English Premier League from a pass in each of the five Zones EPL 2001–02 EPL 2005–06 EPL 2011–12 Categories BB&S BBP&S BB&S BBP&S BB&S BBP&S
Own Half - OH 28 (13%) 6 28 (16%) 13 20 (9%) 8
Wide Right - WR 22 (9%) 3 14 (8%) 6 13 (7%) 9
Zone 14+ – Z14+ 148 (67%) 33 133 (60%) 22 142 (62%) 47 Wide Left - WL 16 (8%) 5 17 (8%) 3 14 (8%) 10
Inside Pen. Area - 7 (3%) 2 17 (8%) 3 37 (14%) 7
IPA
Sub Total 221 50 209 47 226 81
Total % in Open Play 271/641 = 42% 256/582 = 44% 307/792 = 39%
GOAL SCORING PATTERNS 169
Table 4.21 Goals in the Australian ‘A’ League from a pass in each of the five Zones A League 2008–09 A League 2009–10 A League 2011– 12 Categories BB&S BBP&S BB&S BBP&S BB&S BBP&S Own Half - OH 10 (12%) 4 10 (11%) 4 8 (10%) 3
Wide Right - WR 5 (7%) 5 6 (9%) 1 9 (7%) 4 Zone 14+ – Z14+ 32 (57%) 5 46 (53%) 16 50 (73%) 22 Wide Left - WL 3 (12%) 3 1 (21%) 2 6 (10%) 1 Inside Pen. Area - IPA 5 (12%) 0 10 (6%) 0 6 (0%) 1 Sub Total 55 17 73 23 79 31 Total % in Open Play 72/166 = 43% 96/247 = 39% 110/263 = 42%
4.5.1 Test results for goals from five zones in BBSPS category
In all competitions there was a statistically significant difference in a one way
ANOVA test for goals from the five zones in category (a) BBSPS, (p = <.001). The
Bonferroni test showed the difference was between goals from passes in Zone 14+ and the other four areas (p = <.001). An example is provided from the World Cups.
The one-way ANOVA test on the goals in the World Cups from passes in each of the five zones showed a significant difference existed between the passing zones, F (4,10)
= 138.006, p = <.001. The test results for normality and variance are shown in Table 4.22.
The Bonferroni test results showed the differences existed between Zone 14+ and each of the other four zones, p = <.001.
GOAL SCORING PATTERNS 170
Table 4.22 Test results for goals in World Cups from the five zones in the BBSPS category18
Tests of Normality Kolmogorov-Smirnova Shapiro-Wilk category Statistic df Sig. Statistic df Sig. scores OH .314 3 . .893 3 .363 WR .232 3 . .980 3 .726 Z14+ .314 3 . .893 3 .363 WL .337 3 . .855 3 .253 IPA .328 3 . .871 3 .298 a. Lilliefors Significance Correction
Test of Homogeneity of Variances scores Levene Statistic df1 df2 Sig. 2.342 4 10 .125
ANOVA scores Sum of Squares df Mean Square F Sig. Between Groups 6955.500 4 1738.875 138.006 .000 Within Groups 126.000 10 12.600 Total 7081.500 14
18 OH-Own Half, WR-Wide Right, WL-Wide Left, Z14+-Zone 14+, IPA-Inside Penalty Area
GOAL SCORING PATTERNS 171
Table 4.22 continued
Post Hoc Tests Multiple Comparisons Dependent Variable: scores Bonferroni 95% Confidence Interval (I) (J) Mean Difference Lower category category (I-J) Std. Error Sig. Bound Upper Bound OH WR –1.8333 2.8983 1.000 –12.213 8.547 Z14+ –53.5000* 2.8983 .000 –63.880 –43.120 WL 1.1667 2.8983 1.000 –9.213 11.547 IPA 1.6667 2.8983 1.000 –8.713 12.047 WR OH 1.8333 2.8983 1.000 –8.547 12.213 Z14+ –51.6667* 2.8983 .000 –62.047 –41.287 WL 3.0000 2.8983 1.000 –7.380 13.380 IPA 3.5000 2.8983 1.000 –6.880 13.880 Z14+ OH 53.5000* 2.8983 .000 43.120 63.880 WR 51.6667* 2.8983 .000 41.287 62.047 WL 54.6667* 2.8983 .000 44.287 65.047 IPA 55.1667* 2.8983 .000 44.787 65.547 WL OH –1.1667 2.8983 1.000 –11.547 9.213 WR –3.0000 2.8983 1.000 –13.380 7.380 Z14+ –54.6667* 2.8983 .000 –65.047 –44.287 IPA .5000 2.8983 1.000 –9.880 10.880 IPA OH –1.6667 2.8983 1.000 –12.047 8.713 WR –3.5000 2.8983 1.000 –13.880 6.880 Z14+ –55.1667* 2.8983 .000 –65.547 –44.787 WL –.5000 2.8983 1.000 –10.880 9.880 *. The mean difference is significant at the 0.05 level.
A Pearson’s Chi Square test (with α = 0.05) was used to evaluate whether a change in strategy existed between goals scored from the five zones on the field in the BBSPS category in all competitions. An example of the test results for the EPL is shown; see Table
4.23.
GOAL SCORING PATTERNS 172
Table 4.23 Test results for goals from five zones in the EPL
Chi-Square Testsa Asymptotic Significance (2- Value df sided)
Pearson Chi-Square 27.447b 8 .001
Likelihood Ratio 28.534 8 .000
Linear-by-Linear Association 15.498 1 .000
N of Valid Cases 833 a. Tournament = EPL b. 0 cells (.0%) have expected count less than 5. The minimum expected count is 19.98.
GOAL SCORING PATTERNS 173
Table 4.23 continued
TYPE * YEAR Crosstabulationa19 YEAR Total 2001 2005 2011 TYPE Count 34.00 41.00 28.00 103.00 OH Expected Count 33.39 31.65 37.96 103.00 % within TYPE 33.01 39.81 27.18 100.00 % within YEAR 12.59 16.02 9.12 12.36 Adjusted Residual .14 2.13 –2.17 WCount 25.00 20.00 22.00 67.00 WR Expected Count 21.72 20.59 24.69 67.00 % within TYPE 37.31 29.85 32.84 100.00 % within YEAR 9.26 7.81 7.17 8.04 Adjusted Residual .89 –.16 –.71 ZCount 181.00 155.00 189.00 525.00 Z14+ Expected Count 170.17 161.34 193.49 525.00 % within TYPE 34.48 29.52 36.00 100.00 % within YEAR 67.04 60.55 61.56 63.03 Adjusted Residual 1.66 –.99 –.67 WCount 21.00 20.00 24.00 65.00 WL Expected Count 21.07 19.98 23.96 65.00 % within TYPE 32.31 30.77 36.92 100.00 % within YEAR 7.78 7.81 7.82 7.80 L Adjusted Residual –.02 .01 .01 IPA Count 9.00 20.00 44.00 73.00 Expected Count 23.66 22.43 26.90 73.00 % within TYPE 12.33 27.40 60.27 100.00 % within YEAR 3.33 7.81 14.33 8.76 Adjusted Residual –3.84 –.65 4.34 Total Count 270.00 256.00 307.00 833.00 Expected Count 270.00 256.00 307.00 833.00 % within TYPE 32.41 30.73 36.85 100.00 % within YEAR 100.00 100.00 100.00 100.00 a. Tournament = EPL
19 OH-Own Half, WR-Wide Right, WL-Wide Left, Z14+-Zone 14+, IPA-Inside Penalty Area
GOAL SCORING PATTERNS 174
4.5.2 Discussion – Research question 2
The results in Tables 4.18 to 4.21 show that in six international tournaments and six seasons in professional leagues the majority of goals from playing the ball behind opponents or to a player level with the last defender (BBSPS) were scored from passes made in Zone 14+ (Z14+).
A one way ANOVA showed a statistically significant difference existed (p =
<.001) between the five zones in the BBSPS category in all competitions. Bonferroni’s test showed the difference was between Zone 14+ and the other four zones, Own Half (OH),
Wide Right (WR), Wide Left (WL) and Inside the Penalty Area (IPA) with p-values of
<.001.
Pearson’s Chi Square test was applied to the results in each competition. The EPL was the only competition where the result was statistically significant (8, 27.447, p
= .001) indicating a change in strategy of scoring goals in the five zones in the BBSPS category.
The adjusted residual scores (2.13, –2.17); see Table 4.23, show there was a change in the strategy of scoring goals from passes in the own half (OH) between seasons 2005–06 and 2011–12 and from passes inside the penalty area (IPA) between seasons 2001–02 and
2011–12 (–3.84, 4.34). The results of Chi Square tests for the other competitions are in
Appendix 26.
Table 4.24 shows the mean figures for each area of the field, expressed as percentages, for the goals scored in the BBSPS category for each competition. The similarities in the distribution of goals from this category are quite remarkable. The difference in each area of the field ranges between 2.3% (IPA) and 8.7% (WL).
GOAL SCORING PATTERNS 175
Table 4.24 Mean scores for goals from (BBSPS) from each area of the field OH WR Zone 14+ WL IPA
3 World Cups 10% 11% 63% 8% 8% 2002–2010 3 Euros 11% 8% 61% 14% 6% 2004–2012 3 EPL seasons 12% 8% 63% 8% 9% 2001/02 – 11/12 3 A Lge. seasons 15% 11% 60% 6% 8% 2008/09 – 11/12 Mean Difference 5% 3% 3% 8% 3%
The ability to attack centrally has been identified as a characteristic of successful teams whereas unsuccessful teams tended to attack more from the flanks (Carling et al.,
2005). The results in Table 4.24 support that observation across all levels of football examined in this study in terms of the BBSPS category, which produced the majority of goals in ten of the twelve competitions that were analyzed. The difference between goals from passes made in Z14+ and the next area with the highest score is: World Cups = 52%
(63% –11% in WR), EPL = 51% (63% – 12% in OH), Euros = 47% (61% – 14% in WL) and ‘A’ League = 45% (60% – 15% in OH).
The range of the difference between goals from Z14+ and the next area with the highest number of goals is from 45% to 52%, which demonstrates how dominant Z14+ is compared to the other four areas of the field in every competition. The second important observation is the consistency of results between the competitions. The mean difference of the goals scored from passes made in each of the five areas on the field ranges between 3% and 8%, which reflects remarkable similarities in four competitions played over a period of five to ten years. It would be reasonable to assume that there had not been any collusion among the teams with regard to attacking strategies during that period so this discovery GOAL SCORING PATTERNS 176
may reflect something that has been happening in professional football all over the world and only more analysis in the future will determine if that is actually the case. However, based on the evidence from the analysis in this study the results support strongly a strategy of passing the ball behind opponents or to a player who may be level with an opponent in the last line of defense and able to take the ball forwards and or shoot at goal. This strategy is particularly effective from the central area, Zone 14+, when possession is either regained or retained in the opponent’s half of the field.
4.5.3 The top and bottom three teams for passes in Zone 14+ in league competitions
Tables 4.12 to 4.15 provide examples of the goal scoring categories for each competition in this investigation. In the division of where the final pass was made in the
BBSPS category, Zone 14+ is clearly the most effective of the five areas on the field of play. Appendices 3 and 4 show the analysis for each year of competition in the EPL and the ‘A’ League and in the World Cups and European Championships. In the EPL and ‘A’
League every team in the top three had the highest percentage in Zone 14+ or equal highest percentage to another area of the field but less than 50% of them had a higher percentage than the average for the league. In the EPL and ‘A’ League 7 of the 9 and 6 of the 9 teams respectively, in the bottom three positions, had the highest percentage score of goals from passes in Zone 14+. These results show that regardless of where teams finished in the league they scored the vast majority of goals in the BBSPS category from passes made in
Zone 14+. Similar results were found in each of the competitions, the results of which are included in Appendix 2. GOAL SCORING PATTERNS 177
4.5.4 The top 4 teams for passes in Zone 14+ in International competitions
In the World Cups and European Championships 21 of the 24 teams that finished in the top four places had the highest or equal highest percentage in Zone 14+ but less than
50% of the teams had a higher score than the average for the competition.
It is important to recognize that when converting small numbers of incidents into percentage scores the outcomes can sometimes be slightly misleading. For example 2 out of 2 in Z14+ equates to 100%. For this reason the top teams were not compared with the bottom teams in the international tournaments.
4.5.5 Passes along the ground or in the air from Zone 14+
The evidence indicates that in every competition the vast majority of category (a) goals, BBSPS, came from passes in Zone 14+. A sample of each competition was analyzed to find the percentage of passes that were along the ground or in the air from each area on the field. A pass or header that went above head height was deemed to be in the air and a pass or header that went down below head height was deemed to be along the ground.
Table 4.25 shows the percentage of goals from passes along the ground in Zone 14+ for each competition and the total percentage for goals from passes in the other areas due to the small number of goals. A detailed description of goals from each area of the field is provided in Appendix 23.
GOAL SCORING PATTERNS 178
Table 4.25 Goals from passes along the ground in category (a) BBSPS20 OH WR Z14+ WL IPA
WC 2010 0 2 26 81% 2 2 EURO 2012 0 0 20 91% 2 0
EPL 2011=12 12 13 134 71% 13 34
A LGE 2011– 5 4 49 68% 2 4 12
Between 68% and 91% of goals were from passes along the ground from Zone 14+, which is a minimum of two thirds across all competitions and passes.
Passes along the ground from Inside the Penalty Area (IPA) accounted for the almost three quarters of all goals in this area. In the other three areas, Own Half (OH),
Wide Right (WR) and Wide Left (WL) the majority of goals were a result of passes made in the air.
4.6 Confirmed and Unconfirmed Coding
Accuracy in coding is extremely important and assiduous care was taken to ensure the highest standard was maintained. However, in two of the three EPL seasons, 2001–02 and 2005–06, a large number of goals were coded from TV Highlights Programs, which did not always show exactly where on the field possession was regained. Possession was coded from where the team had the ball, for example in the Back Third and the number of passes in the sequence were counted from that point onwards; the CONHL (confirmed) or
UNCON (unconfirmed) buttons were activated according to the situation observed by the operator. In some cases, a pass would obviously have started in the Back Third, from a goal kick for example, but the action was not visible on TV so the incident would have
20 OH-Own Half, WR-Wide Right, Z14+-Zone 14+, WL-Wide Left, IPA-Inside Penalty Area
GOAL SCORING PATTERNS 179
been coded as Regained in the Back Third but unconfirmed. Similarly, the number of passes in the passage of play would be counted from the first visible pass in the sequence and coded as unconfirmed. In some instances the number of passes went beyond six, which would have accurately placed the sequence in the 6+ passes category but the incident would still have been coded as unconfirmed. The two seasons in the EPL were the only competitions where there was a potential for inaccuracy in the number of goals from a
‘zero’ pass, 1–5 passes or 6+ passes and for the exact area where possession was regained.
Therefore, it should not be assumed that coding the number of passes, or where possession was regained, as ‘unconfirmed’ is equivalent to being incorrect. It means either the number of passes or where possession was regained cannot be confirmed. A comparison was made between all coded incidents and incidents that were coded as ‘confirmed’. Table 4.26 shows the percentage scores for goals in the three passing categories for all goals and for those with ‘confirmed’ status.
Table 4.26 Comparison of ‘Confirmed’ passes with totals in each category English Premier 2001–02 All Confirmed 2005–06 All Confirmed League 0 Pass 58 9% 50 18% 56 10% 54 14% 1–5 Passes 536 84% 299 79% 495 85% 321 82% 6+ Passes 47 7% 30 8% 31 5% 15 4% Total 641 100% 379 59% 582 100% 390 67%
The percentage scores for the ‘Confirmed’ incidents are within 5% of the overall figures with the exception of ‘zero’ Pass goals in 2001–02. The difference is 9% because there was a disproportionate number (50/58) of confirmed incidents. For example, if only
40 of the 58 incidents had been confirmed the percentage would have been 10% not 18%.
The sample of ‘confirmed’ incidents reflects the overall trend with a difference of 5% or less in each category for both seasons in the EPL. GOAL SCORING PATTERNS 180
When regained possessions in each third of the field are compared using
‘confirmed’ incidents and the totals in each category a similar pattern can be seen. Table
4.27 shows the percentage scores for the total goals and for those with ‘Confirmed’ status.
In 2001–02 there is a difference of 7% in the percentage scores for regained possessions in the Back Third and a 6% difference in regained possessions in the attacking half of midfield (TH-MF), which is 2% and 1% outside a 5% margin of error. In 2005–06 the difference between regained possessions in TH-MF is 6%, the rest are within 5%. When the figures for regained possessions in each half of the field are compared the greatest difference is 3% in 2001–02. This acknowledged inaccuracy in this aspect of the study, which in most cases is less than 5% will not change the statistically significant difference between goals from regained possessions in the Back Third and the Middle Third, or between the Middle Third and the Final Third, but the scores provided in seasons 2001–02 and 2005–06 should be viewed as having a standard deviation of 5–10%.
Table 4.27 Comparison of ‘Confirmed regained possessions with total in each area of the field21 EPL 2001–02 All Confirmed 2005–06 All Confirmed Reg. in B3 106 16% 89 23% 109 19% 95 24% Reg. in OH-MF 153 24% 74 20% 148 25% 80 21% Reg. in TH-MF 159 25% 73 19% 156 27% 79 21% Reg. in F3 223 35% 143 38% 169 29% 136 34% Total 641 100% 379 59% 582 100% 390 67% Reg. Own Half 259 40% 163 43% 257 44% 175 45% Reg. Their Half 382 60% 216 57% 325 56% 215 55%
4.7 Goals from regained possession in each third and each half of the field
The next stage of the investigation into goal scoring patterns required (a) analysis of the areas of the field where possession was regained to identify trends that might apply to all of the competitions in this study or between international and league competitions;
21 Reg.- Regained, B3- Back Third, OH-MF - Own Half of Midfield, TH-MF – Their Half of Midfield, F3 – Final Third GOAL SCORING PATTERNS 181
(b) analysis of the categories of goals scored in each area; and (c) analysis of the number of passes for the categories of goals in each area of the field. This information would develop a better understanding of the relationship between ‘possession based’ football and ‘direct play’.
In the EPL and ‘A’ League there were two clear trends in Open Play in relation to where possession was regained on the field. In each of the three seasons in both leagues, the majority of goals came from regained possession in the Middle Third (M3) of the field, compared with the Back Third (B3) and the Attacking or Final Third (F3); see Table 4.28.
Table 4.28 Regained possession in each third of the field in the EPL and ‘A’ League EPL Regained B3 Regained M3 Regained F3 Total 2001–02 106 16% 312 49% 223 35% 641 2005–06 109 19% 304 52% 169 29% 582 2011–12 202 26% 328 41% 261 33% 792 A LGE Regained B3 Regained M3 Regained F3 Total 2008–09 31 19% 87 52% 48 29% 166 2009–10 72 29% 100 41% 75 30% 247 2011–12 66 25% 126 48% 71 27% 263
A one-way ANOVA test on each competition showed there was a statistically significant difference between goals from regained possessions in each third of the field.
Bonferroni’s test showed the differences were between goals from regained possessions in the Middle Third and the Back Third and between the Middle Third and the Final Third.
Test results from the EPL are provided as an example; see Table 4.29.
A one-way ANOVA test on the goals in the EPL showed a significant difference existed, F (2,6) = 24.221, p = .001. The test results for normality and variance are shown and Bonferroni test results show the differences existed between the Back Third and the
Middle Third, p = <.001 and between the Middle Third and the Final Third, p = .025. The results of the statistical tests for each competition are provided in Appendix 1, Sheets 3 and
4. GOAL SCORING PATTERNS 182
Table 4.29 Test results for EPL goals from regained possession in each ‘Third’ Tests of Normality Kolmogorov-Smirnova Shapiro-Wilk areas Statistic df Sig. Statistic df Sig. scores B3rd .269 3 . .949 3 .567 M3rd .282 3 . .936 3 .510 F3rd .253 3 . .964 3 .637 a. Lilliefors Significance Correction
Test of Homogeneity of Variances scores
Levene Statistic df1 df2 Sig. .798 2 6 .493
ANOVA scores
Sum of Squares df Mean Square F Sig. Between Groups 1098.000 2 549.000 24.221 .001 Within Groups 136.000 6 22.667 Total 1234.000 8
GOAL SCORING PATTERNS 183
Table 4.29 continued Post Hoc Tests Multiple Comparisons Dependent Variable:scores Bonferroni 95% Mean Confidence Interval (I) (J) Difference (I- areas areas J) Std. Error Sig. Lower Bound Upper Bound Back3rd M3rd –27.000* 3.887 .001 –39.78 –14.22 F3rd –12.000 3.887 .064 –24.78 .78 Mid. 3rd B3rd 27.000* 3.887 .001 14.22 39.78 F3rd 15.000* 3.887 .025 2.22 27.78 Final 3rd B3rd 12.000 3.887 .064 –.78 24.78 M3rd –15.000* 3.887 .025 –27.78 –2.22 *. The mean difference is significant at the 0.05 level.
A Pearson’s Chi Square test (with α = 0.05) was used to evaluate whether a change in strategy existed between goals scored from regained possessions in the Back Third,
Middle Third and Final Third in each season of the competitions under investigation. The
Chi Square test was statistically significant in the English Premier League 2 (4, 27.587, p=<.001); see Table 4.30. The Chi Square test results for the other three competitions were not statistically significant indicating consistency in the proportion of goals scored over the three seasons or tournaments within each competition.
GOAL SCORING PATTERNS 184
Table 4.30 Chi Square test results (1) for EPL goals from regained possessions in each ‘Third’ Chi-Square Testsa Asymptotic Value df Significance (2-sided)
Pearson Chi-Square 27.587b 4 .000
Likelihood Ratio 27.519 4 .000
Linear-by-Linear Association 7.257 1 .007
N of Valid Cases 2014 a. Tournament = EPL b 0 cells (.0%) have expected count less than 5. The minimum expected count is 120.50.
The strong relationships (adjusted residuals, –3.2, 4.3) between the goals from regained possessions in the Back Third in seasons 2001–02 and 2011–12 and regained possessions in the Middle Third in seasons 2005–6 and 2011–12 (adjusted residuals, 3.1, –
3.9) showed a change in strategy in scoring goals from regained possessions in each third of the field; see Table 4.31. The Chi Square test results for each competition are in
Appendix 26.
GOAL SCORING PATTERNS 185
Table 4.31 Pearson’s Chi Square Test Results (2) for regained possessions in the EPL22 TYPE * YEAR Crosstabulationa YEAR 2001–02 2005–06 2011–12 Total TYPE RegB3 Count 106 109 202 417 Expected Count 132.7 120.5 163.8 417.0 % within TYPE 25.4% 26.1% 48.4% 100.0% % within YEAR 16.5% 18.7% 25.5% 20.7% % of Total 5.3% 5.4% 10.0% 20.7% Adjusted Residual –3.2 –1.4 4.3 RegM3 Count 312 304 328 944 Expected Count 300.4 272.8 370.8 944.0 % within TYPE 33.1% 32.2% 34.7% 100.0% % within YEAR 48.7% 52.2% 41.5% 46.9% % of Total 15.5% 15.1% 16.3% 46.9% Adjusted Residual 1.1 3.1 –3.9 RegF3 Count 223 169 261 653 Expected Count 207.8 188.7 256.5 653.0 % within TYPE 34.2% 25.9% 40.0% 100.0% % within YEAR 34.8% 29.0% 33.0% 32.4% % of Total 11.1% 8.4% 13.0% 32.4% Adjusted Residual 1.6 –2.1 .4 Total Count 641 582 791 2014 Expected Count 641.0 582.0 791.0 2014.0 % within TYPE 31.8% 28.9% 39.3% 100.0% % within YEAR 100.0% 100.0% 100.0% 100.0% % of Total 31.8% 28.9% 39.3% 100.0% a. Tournament = EPL
When the two halves of the field were compared the majority of goals were scored from regained possession in the opponents half in every case, see Figure 4.32. The details of all the competitions in this study are provided in Appendix 5 for the EPL and ‘A’
League and Appendix 6 for the international tournaments.
22 Reg.- Regained, B3- Back Third, M3 – Middle Third, F3 – Final Third GOAL SCORING PATTERNS 186
Table 4.32 Regained possession in Own Half (OH) and Their Half (TH) in the EPL and ‘A’ League EPL Regained Regained Total Regained Regained Total in B3rd Own Half Regained Their Half In F3rd Regained of MF In OH of MF In TH 2001–02 106 16% 153 24% 259 40% 159 25% 223 35% 382 60% 2005–06 109 19% 148 25% 257 44% 156 27% 169 29% 325 56% 2011–12 202 26% 151 19% 353 45% 178 22% 261 33% 439 55% A LGE 2008–09 31 19% 48 29% 79 48% 39 23% 48 29% 87 52% 2009–10 72 29% 52 21% 124 50% 48 20% 75 30% 123 50% 2011–12 66 25% 55 21% 121 46% 71 27% 71 27% 142 54%
In each of the three tournaments in the World Cups and European Championships, the majority of goals came from regained possession in the Middle Third (M3) of the field, compared with the Back Third (B3) and the Attacking or Final Third (F3); see Table 4.33.
When the two halves of the field were compared the majority of goals were scored from regained possession in the teams’ own half in four of the six tournaments; see Table
4.34.
GOAL SCORING PATTERNS 187
Table 4.33 Regained possession in each third of the field in the World Cups and European Championships World Cups Regained in Regained in Regained in Total Back 3rd Middle 3rd Final 3rd 2002 34 30% 45 39% 36 31% 115 2006 21 23% 47 50% 25 27% 93 2010 26 24% 60 57% 20 19% 106 EUROS 2004 16 33% 25 51% 8 16% 49 2008 19 32% 31 52% 10 16% 60 2012 18 30% 27 44% 16 26% 61
Table 4.34 Regained possession in Own Half (OH) and Their Half (TH) in the World Cups and European Championships World Regained Regained Total Regained Regained Total Cups In B3 Own half Regained Their half In F3 Regained of MF In OH of MF In TH 2002 34 30% 22 19% 56 49% 23 20% 36 31% 59 51% 2006 21 23% 32 34% 53 57% 15 16% 25 27% 40 43% 2010 26 24% 28 27% 54 51% 32 30% 20 19% 52 49% EUROS 2004 16 33% 10 20% 26 53% 15 31% 8 16% 23 47% 2008 19 32% 23 38% 42 70% 8 14% 10 16% 18 30% 2012 18 30% 9 14% 27 44% 18 30% 16 26% 34 56%
One difference between the league competitions, Table 4.32, and the international tournaments, Table 4.34, is the number of goals from regained possessions in the opposition’s half of the field compared with the number of regained possessions in the teams own half.
The difference between each half of the field being the most dominant ranges between 10%–20% in the EPL and 4%–8% in the ‘A’ League with the opposition’s half the most dominant in every one. However in the International tournaments the balance GOAL SCORING PATTERNS 188
swings in favor of regained possessions in the teams own half in 4 out of 6 cases with a range between 2% and 14% in World Cups and 6% and 40% in European Championships.
There is also a considerable difference in the number of goals from regained possessions in the Back Third in international tournaments compared with the league competitions. For example, in the six league competitions there were three with regained possessions in the Back Third under 20% and three with 20% – 29%; see Table 4.32, whereas in the international tournaments the lowest figure was 23% and four were 30% or higher; see Table 4.34.
Since the Middle Third accounted for the highest number of regained possessions in every competition in this study the Back Third was the area of the field that made the difference in the number of goals from regained possessions in each half of the field in international football. The opposite effect was observed in league football, where more goals came from regained possessions in the Attacking Third than the Back Third.
As an illustration, the lowest figure for regained possessions in the Final Third in league football was 27% and the highest in international football was 28%. Explanations for this difference could be the attacking strategies of international teams compared with club teams, in that they might be more inclined to keep possession of the ball and be more successful at building up from the Back Third to the Final Third due to the better quality of players available to national teams. Another strategic explanation could be the tendency to retreat towards the half way line when possession has been lost rather than pressing the opposing players quickly and closer to their goal, because pressing requires a lot of hard work and against very good players the success rate of winning the ball will be lower than against poorer quality players. This tactic, to save energy, is employed frequently by teams when playing in hot conditions, which is usually the case in international tournaments. GOAL SCORING PATTERNS 189
4.7.1 Regained possessions by the top three and bottom three teams of the EPL and ‘A’ League
When the top and bottom three teams in the EPL were compared, the trend was still consistent but tended to weaken over time. In 2001 all of the six teams scored more goals from regained possession in the middle third, in 2005 it was the top three teams and two of the three bottom teams, which is five out of six and in 2011 it was two of the top three teams and one of the bottom three teams, which is three out of six.
These figures indicate a change of 50% in 2011–12 compared with 2001–02 when
MU in the top three and Bolton in the bottom three teams scored more goals from regained possessions in the final third and Blackburn scored more goals from regained possessions in the Back Third; see Table 4.35. Details of regained possessions are reported in
Appendix 5.
Table 4.35 The top and bottom three teams in the EPL in 2011–12 EPL Regained in Regained in Regained in Open Play Set Plays Back 3rd Middle 3rd Final 3rd 1 MAN C. 22 32% 27 39% 20 29% 69 23 2 MUTD 15 22% 23 33% 31 45% 69 20 3 ARS 16 24% 27 42% 22 34% 65 9 18 BOL 6 19% 12 37% 14 44% 32 14 19 BLA 10 46% 8 36% 4 18% 22 25 20 WOL 7 25% 13 46% 8 29% 28 11
It would be misguided to assume that a higher number of goals from regained possessions in the Back Third would automatically result in a higher number of goals with passing sequences of 6 or more. This topic will be covered in more detail later in the text.
In this example, Table 4.35, Blackburn scored 22 goals in Open Play, which was the lowest number in the league. By cross referencing in the code matrix it was possible to determine how many were from regained possession in the Back Third and how many GOAL SCORING PATTERNS 190
passes were made for each of the goals. Ten of the goals were from regained possession in the Back Third and six were scored from 5 passes or less, the remaining four goals were from 6 passes or more. Without the detailed information many would have predicted all ten goals from 5 passes or less because at that time Blackburn were renowned for playing long balls from the Back Third and for scoring a lot of goals from Set Plays, which is why interpretation of results should preferably be done with the playing style in mind and in the overall context of goals scored in Open Play and from Set Plays. In 2011–12 Blackburn scored 25 goals from Set Plays, which was 53% of their total and two more than
Manchester City who won the league. In contrast Manchester City’s 23 goals represented
25% of their total of 92. Appendix 5 provides detailed analysis of each team in the top three and bottom three positions of each league table.
In the ‘A’ League the trend was more consistent than in the EPL but the time span between the first and last results was over four years not ten.
In the ‘A’ League eight of the nine teams that finished in the top three of the league and six of the nine teams that finished in the bottom three, scored more goals from regained possession in the Middle Third, which is 78% with the overall trend. Six out of the nine top three teams and seven of the nine bottom three teams also scored more from regained possessions in the opponent’s half of the field, which is 72% with the trend.
Regardless of whether a team finished in the top or bottom three positions in the
EPL or the ‘A’ league the pattern of scoring more goals from regained possessions in the
Middle Third of the field and from regained possessions in the opponents half of the field was evident.
GOAL SCORING PATTERNS 191
4.7.2 Regained possessions by the top four teams in the World Cups and European Championships
The number of goals scored by the top four teams from regained possessions in each third of the field were combined due to the low number of goals scored by each team and compared with the average for each tournament. The majority of goals were scored from regained possessions in the Middle Third of the field by the top four teams, which was the same outcome overall for each World Cup and European tournament. Full details are included in Appendix 6. Table 4.36 provides an example of the top four teams in Euro
2012 and the overall figures.
Table 4.36 Regained possessions overall and by the top four teams in Euro 2012 Euro 2012 Overall Regained Regained Regained Back 3rd 30% Middle 3rd 44% Final 3rd 26% Spain 2 6 2 Italy 2 1 0 Germany 3 3 2 Portugal 2 1 1 Total 9 36% 11 44% 5 20%
The majority of goals were scored from regained possessions in the Middle Third of the field, which was the same as the overall trend for the tournament.
The top four teams in the European Championships scored more from regained possessions in the opponents half of the field in two of the three tournaments, which was the opposite to the overall performance.
In the World Cups the top four teams combined scored more goals from regained possessions in their own half of the field in every tournament, which was in line with the overall result of two out of the three tournaments; see Table 4.37. Data are reported in
Appendix 6.
GOAL SCORING PATTERNS 192
Table 4.37 Regained possessions in Middle Third (M3) overall and by the top four teams in the World Cups and European Championships Top 4 Teams Regained Regained Regained Middle 3rd Regained Back 3rd Middle 3rd Overall Ave. Final 3rd Euro 04 15% 65% 51% 20% Euro 08 29% 55% 52% 16% Euro 12 36% 44% 44% 20% World Cup 38% 41% 39% 21% 2002 World Cup 18% 52% 50% 30% 2006 World Cup 32% 55% 58% 13% 2010
The results of the top four teams were consistent with the tournament average in that more goals were scored from regained possessions in the Middle Third, unlike the top three teams in the EPL where there was a 50% change in which third of the field had the most goals from regained possessions in Open Play between 2001 and 2012.
Overall there was not a statistically significant difference between the regained possessions in the Back Third compared with the Final Third in any of the competitions but there were differences in the number of goals from regained possessions in the Back
Third in international football compared with league football.
4.8 Regained possession in the Defending or Back Third
The figures for regained possessions in the Back Third in international tournaments are similar with a range between 23% and 30% in World Cups and between 30% and 33% in the European Championships.
In the EPL the overall percentage of goals from regained possessions in the Back
Third rose from 16% in 2001–02 to 26% in 2011–12 while in the ‘A’ League there was an GOAL SCORING PATTERNS 193
increase from 19% in 2008–09 to 29% in 2009-10 with a decrease to 25% in 2011–12.
Appendices 5 and 6 report the data to support these analyses.
The number of goals from regained possessions in the Back Third were generally higher in the international competitions than in the league competitions, which was the opposite to the number of goals from regained possessions in the Final Third, see Table
4.38.
Table 4.38 Regained possessions in the Back Third and Final Third Regained in Back 3rd Regained in Final 3rd World Cup 2002 34 30% 36 31% World Cup 2006 21 23% 25 27% World Cup 2010 26 24% 20 19% Euro 2004 16 33% 8 16% Euro 2008 19 32% 10 16% Euro 2012 18 30% 16 26% EPL 2001–02 106 16% 223 35% EPL 2005–06 109 19% 169 29% EPL 2011–12 202 26% 261 33%
A League 2008–09 31 19% 48 29% A League 2009–10 72 29% 75 30% A League 2011–12 66 25% 71 27%
Overall more goals were scored from regained possessions in the Back Third in international tournaments, between 23% and 33%, compared with the league competitions where the range was between 16% and 29%.
The difference in the range of goals may be a result of the type of competition in that international football is essentially knock-out football played over several weeks and usually in hot conditions with the best players from each country, whereas league football is played over nine months of the year in a mixture of climatic conditions with the majority GOAL SCORING PATTERNS 194
of players who do not play international football. However, the interest in what teams do with the possession once it has been regained has fuelled the debate over ‘possession based’ football and ‘direct play’. It is logical to expect a higher number of goals from regained possessions in the Back Third to have been scored from longer passing sequences due to the distance the ball has to travel before entering the goal. With more goals overall from regained possessions in the Back Third in international competitions it would be logical to expect a higher percentage of goals to be scored from six passes or more in international tournaments compared with league competitions. The details for the goals and number of passes in each third of the field and in each competition are provided later in the text in
Tables 4.42 to 4.45, at this point the goals and number of passes from regained possession in the Back Third are provided as an example in Table 4.30 to show the pattern over time.
GOAL SCORING PATTERNS 195
4.8.1 The number of passes for goals from regained possessions in the Back Third of the field
Table 4.39 Goals from regained possession in the Back Third and the number of passes Regained in < 5 Passes 6 Passes + Back 3rd World Cup 2002 34 30% 22 65% 12 35% World Cup 2006 21 23% 11 52% 10 48% World Cup 2010 26 24% 19 73% 7 27% EURO 2004 16 33% 9 56% 7 44% EURO 2008 19 32% 14 74% 5 26% EURO 2012 18 29% 10 56% 8 44% EPL 2001–02 106 16% 95 90% 11 10% EPL 2005–06 109 19% 100 92% 9 8% EPL 2011–12 202 * 26% 132 65% 69 35% A LGE 2008–9 31 19% 23 74% 8 26% A LGE 2009–10 72 29% 59 82% 13 18% A LGE 2011–12 66 25% 32 48% 34 52%
Note: EPL 11–12 * denotes one goal without a pass, scored directly by the GK
In the league competitions and European Championships there was an increase in the number of goals scored from six passes or more compared with the previous competition, shown in Table 4.39. The World Cup was the only competition to record a decrease in the number of goals from six passes or more between 2006 and 2010. Table
4.39 does show the trend of a higher percentage of goals from sequences of 6 passes or more in international tournaments compared with league competitions over time until the
2011–12 competitions.
In the EPL and the ‘A’ League in 2011–12 there was a substantial increase in the number of goals from 6 passes or more. The EPL figure rose from 8% to 35% between GOAL SCORING PATTERNS 196
2005–06 and 2011–12 and the ‘A’ League figure rose from 18% to 52% between 2008–09 and 2011–12. The increases of 27% and 34% respectively were not a result of a comparable increase in the number of regained possessions in the Back Third. In the EPL there was an increase of 7% in the number of regained possessions and in the ‘A’ League there was actually a decrease of 3.5% in the number of regained possessions in the Back
Third.
The ‘A’ League was the only competition that had a majority of goals from the
Back Third from 6 passes or more, compared with 5 passes or less and despite a significant increase in the EPL from 8% to 35% the majority of goals (65%) in the EPL were in fact scored from sequences of 5 passes or less.
To put the results in perspective, more goals from 5 passes or less, were recorded in the World Cups, the European Championships, the EPL and in the previous two seasons of the “A’ League. The question is why did this happen to such an extent in the ‘A’ League?
There are several possibilities; the first is that there was a change of approach by the coaches to attempt to play from the Back Third in preference to making long forward passes, either by choice or when under pressure from opponents. The second possibility is the teams had more success by passing the ball over shorter distances than long distances.
A video review of the 66 goals from regained possessions in the Back Third in the 2011–
12 ‘A’ League season showed that two of the 34 goals from 6 passes or more involved a goal kick or a drop kick by the goalkeeper and two more from a long forward pass. The remaining 32 goals from regained possessions in the Back Third that were scored from sequences of 5 passes or less included three that involved a goal-kick or drop-kick by the goalkeeper and four from a long forward pass, three of which lead to goals in the BBSPS category. GOAL SCORING PATTERNS 197
It would be a mistake to assume that goals with 5 passes or less from regained possession in the Back Third of the field were a result of making long forward passes. The evidence does not support that assumption.
In total there were 11 goals (17%) in the 66 from regained possession in the Back
Third that involved a long forward pass or a long kick by the goalkeeper, 7 (22%) were in the 32 goals of 5 passes or less and 4 were in the 34 goals of 6 passes or more. The remaining 25 of the 32 goals (78%) with 5 passes or less demonstrate that the ball can be moved quickly from one end of the field to the other by passing and running with it as an alternative to long forward passes.
One recommendation of the attacking strategy of ‘Direct Play’ was based on,
“playing the ball forward whenever possible with the aim of achieving a shooting opportunity within five consecutive passes” (Hughes, 1990:173). There were 66 goals from regained possessions in the Back Third, of which 29 were in the BBSPS category. Table
4.40 shows how the goals were scored from regained possessions in the Back Third and it is relevant to point out that 16 of the 32 goals from 5 passes or less were in the BBSPS category. Four of the 16 goals were from passes inside the team’s own half and seven were from Zone 14+. When the goals from 5 passes or less and 6 passes or more were combined the results showed that passing the ball behind opponents or the BBSPS category, was the most effective method of scoring goals with 29 out of the 66 and the central area, Zone
14+ was the most effective area to pass from accounting for 15 of the 29 goals.
GOAL SCORING PATTERNS 198
Table 4.40 Goals and categories from regained possession in the Back Third in the ‘A’ League 2011–12 23 Categories BBSPS OM Cross BBSPS Areas of the field OH WR Z14+ WL IPA Passes 1–5 32 16 10 4 3 7 0 2 6 1 2 8 1 1 Passes 6+ 34 13 11 10 5 5 15 1 3 Total 66 29 21 16 Passes 6–9 19 8 4 7 1 2 3 1 1 Passes 10–15 13 4 6 3 0 0 4 0 0 Passes 16+ 2 1 0 1 0 0 1 0 0 Total 6+ Passes 34 13 10 11
The ‘A’ League data for goals scored in each category with the number of passes from regained possessions in each third of the field are in Appendix 12, the World Cup,
European Championships and EPL data are in Appendices 9, 10 and 11 respectively. The
Final Third of the field is an area of importance because Hughes (1990) stated that 52% of goals came from regained possessions in that area but he did not state how the possessions were regained. Based on his evidence Hughes advocated teams should pressurize opponents as quickly as possible after losing the ball and as close as possible to the opponent’s goal. The next section compares the number of regained possessions in the final third in the different competitions and how possession was regained.
4.9 Goals from regained possession in the Final Third
In the league competitions, more goals were scored from regained possessions in the Final Third compared with the Back Third in every season. In the EPL the percentage of goals ranged from 29% – 35% and from 27% – 30% in the ‘A’ League, which is quite similar.
23 BBSPS- BB&S and BBP&S combined BB&S- Ball behind and strike, BBP&S- Ball behind pass and strike, OM-Other Methods, OH – Own Half, WR – Wide Right, Z14+ - Zone 14+, WL – Wide Left, IPA – Inside Penalty Area GOAL SCORING PATTERNS 199
In the international tournaments, more goals were scored from regained possessions in the Final Third in two of the World Cups and in the other four tournaments the figure was lower compared with the goals from regained possessions in the Back Third.
The biggest fluctuation was in the regained possessions in the Final Third in the
World Cups where the figure dropped from 27% in 2006 after 31% in 2002, to a low of
19% in 2010. In the European Championships, the figure jumped from 16% in 2004 and
2008 to 26% in 2012. It is difficult to explain why there should be an opposite effect in two different international tournaments but a strong reason for the increased percentage in the
2012 European Championships may have been a result of the tactical approach employed by Barcelona at club level and Spain at the international level where the large number of
Barcelona players in the national team probably had a major influence on the national team’s strategy. Both teams employed a tactic of trying to regain possession within five seconds and both teams were successful, so this may have initiated a tactical trend in 2012, which more teams may have used to increase their chances of success. The interesting thing is whether the tactical theory of regaining the ball quickly and presumably in the
Final Third, actually worked in producing more goals from that area. Spain was the most successful team between 2008 and 2012 but the evidence of their regained possessions does not support their tactical theory in terms of scoring goals. In Euro 2008, Spain scored
11 goals in Open Play and did not regain possession once in the Final Third. In 2012, they won the trophy again scoring 10 goals in Open Play and regained possession of the ball twice in the Final Third. Spain won the World Cup in 2010 and scored 7 goals without regaining possession once in the Final Third. The evidence is that Spain scored 28 goals to win three tournaments and regained possession of the ball twice in the Final Third of the field, despite employing a tactic of trying to press their opponents quickly to regain possession within five seconds. GOAL SCORING PATTERNS 200
The fact that Spain only scored twice from regained possession of the ball in the
Final Third is not really surprising because players are taught not to get caught in possession in their Back Third, so if and when players are pressed by the opposition with few available options they will pass the ball forwards even if it means giving the ball away.
In three European Championships between 2004 and 2012, the top four teams managed a high of 20% for regained possessions in the Final Third so it would be useful to know how possession was regained and particularly in 2012 when the average was 26%, up from 16% in 2008. By cross-referencing in the code matrix data the answer was found. In Euro 2012, there were 16 goals scored in Open Play from regained possessions in the Final Third: 5 came from clearances after corners or crosses; 4 were from rebounds off defenders after a shot or cross from outside the penalty area was blocked; 4 more were from defenders simply giving the ball away and one was under pressure from an opponent; 2 came from winning the ball off an opponent; and 1 was from a throw in. Thus, only three of the sixteen goals from regained possession in the Final Third were a result of direct pressure on the ball. Furthermore, in the absence of detailed information about events prior to regaining possession it would be easy to make an assumption that possession of the ball in the Final Third is regained by either winning it from an opponent or as a result of pressurizing an opponent in possession of the ball, thereby forcing a mistake. This would be an easy assumption to make if a mental image of either situation came to mind because the brain responds to suggestions that are logical and make sense with the information it has to process (Kahneman, 2011).
Hughes (1990) based his strategy of pressurizing opponents as quickly as possible and as far up the field as possible on the evidence from his analysis, which showed that
52% of all goals came from regained possession in the Final Third. The figures for regained possessions that Hughes stated were true but he did not make it clear at the time GOAL SCORING PATTERNS 201
that his figures included goals scored from Set Plays, including penalties. Without thinking
‘slowly’ about this statement it would most likely lead to a mental image of players regaining possession in Open Play and the assumption that the 52% of goals from regained possession in the Final Third occurred in Open Play. When the goals from Set Plays were deducted from the total number of goals scored only 13% of the goals in Open Play were a result of regained possessions in the Final Third.
One might ask if the results of the regained possessions in Euro 2012, which were used to illustrate the different ways possession might be regained, were typical of other competitions. In Euro 2008, there were 10 goals in Open Play from regained possession in the Final Third. Similar to Euro 2012, 2 goals 20% (2/10) came from winning the ball off an opponent, 30% were from throw ins (3/10), while the remaining five goals, 50%, were from intercepting passes (2) when the player was not under pressure and (3) after a tackle or a clearance.
In the 2011–12 ‘A’ League season, there were 70 goals from regained possessions in the final third, 18 of which were from throw-ins. It could be argued that regained possession from throw-ins are not really in Open Play and should maybe not be included.
The point of analyzing regained possessions in fine detail is to discover if pressurizing opponents results in winning the ball in the Final Third. Regained possessions from throw- ins are not a result of applying pressure to opponents in possession of the ball, but it is acknowledged that some throw-ins are given away as a result of pressure being applied by opponents. In this example the regained possessions from throw-ins account for 25%
(18/71), clearances account for 28% (20/71), giving the ball away, intercepting a pass or after a tackle by a defender account for 32% (23/71), 6% were from winning the ball off an opponent (4/71) and the remaining 9% (6/71) were from rebounds after shots or challenges. GOAL SCORING PATTERNS 202
The majority of regained possessions in the Final Third in Open Play come from clearances, after tackles by defenders, rebounds and the lowest number from dispossessing opponents. These results demonstrate the importance of having a large number of players in the opponent’s half of the field to increase the likelihood of regaining possession when an attacking move breaks down.
In the 2011–12 EPL season, regained possessions in the Final Third were as follows: throw-ins and quick free-kicks 18% (47/261); clearances 38% (98/261); giving the ball away, interceptions and after tackles by defenders 23% (60/261); winning the ball from an opponent 12% (31/261); and the remaining 9% (25/261) were from rebounds and blocked shots off defenders, not the goalkeeper. One might argue that clearances are made under pressure from an attacking player, which is usually the case but there is a difference between a clearance and pressurizing an opponent who has possession of the ball. When a player clears the ball in a dangerous situation the defender does not have possession of it, but manages to get to the ball before an attacking player and wins the contest. It is for this reason that clearances are viewed differently compared with winning the ball from an opponent when the player is in possession of the ball. The most common outcome from pressurizing opponents in the Final Third is a long forward pass, which usually results in a regained possession in the Middle or the Back Third.
The evidence from the examples provided shows the majority of goals from regained possessions in the Final Third came from events other than winning the ball from an opponent or pressurizing a player in possession of the ball, with throw-ins to be considered as a special case.
This statement is not suggesting that teams should not pressurize opponents in the
Final Third because there are obvious benefits of doing so, but regaining possession of the ball in the Final Third does not happen as often as one might expect. The data for the GOAL SCORING PATTERNS 203
number of passes preceding goals from regained possession in the Final Third of the field in each competition and in each category of goals are provided in Appendices 9 to 12. The following sections of this chapter provide an overview of the number of passes preceding goals in each competition from different areas of the field.
4.10 The number of passes preceding goals in each competition
An integral part of performance analysis of football is the identification of the number of passes made prior to goals scored. In previous research, goals were grouped according to the number of passes made and some included the number of goals from regained possessions in each third of the field (Breen et al., 2006). However, it was identified in the literature review in Chapter 2 that previous studies did not report the number of passes made for the goals from regained possession in each third of the field.
Table 4.41 shows the results for the number of passes for goals in each competition in this study. ‘Zero’ pass goals are shown separately with the total number of goals from 1–5 passes, the two figures are added together to present a combined total of goals from 0–5 passes.
GOAL SCORING PATTERNS 204
Table 4.41 The number of goals and passes in each competition 0 Pass 1–5 Passes Total 6+ Passes Total 0–5 Passes World Cup 15 13% 77 67% 92 80% 23 20% 115 2002 World Cup 7 8% 65 70% 72 78% 21 22% 93 2006 World Cup 9 8% 69 65% 78 73% 28 27% 106 2010 EURO 2004 3 6% 38 78% 41 84% 8 16% 49 EURO 2008 3 5% 44 73% 47 78% 13 22% 60 EURO 2012 7 11% 31 51% 38 62% 23 38% 61
EPL 2001–02 58 9% 536 84% 594 93% 47 7% 641 EPL 2005–06 56 10% 495 85% 551 95% 31 5% 582 EPL 2011–12 81 10% 525 66% 606 76% 186 24% 792
A Lge.2008–09 12 7% 130 78% 142 85% 24 15% 166 A Lge. 2009–10 21 9% 196 79% 217 88% 30 12% 247 A Lge. 2011–12 23 9% 175 67% 198 76% 65 24% 263
The data in Table 4.41 show how little has changed in terms of the higher
percentage of goals from 0–5 passes compared with goals from 6 or more passes.
In every competition, the majority of goals were scored with 5 passes or less
compared with 6 passes or more. However, there is a trend in every competition from the
first recorded year of the competition to the last recorded year for the number of goals in
the 0–5 category to decrease, with a subsequent increase over time in the number of goals
from 6 passes or more.
To illustrate this point, there was an increase from 20% in the 2002 World Cup to
27% in the 2010 World Cup, an increase from 16% in Euro 2004 to 38% in 2012, with
similar gains in the EPL where the figure was 7% in 2002 and 24% in 2012. Even over a
shorter time span the ‘A’ League showed an increase from 15% in 2008–09 to 24% in GOAL SCORING PATTERNS 205
2011–12. This trend is consistent across different competitions and probably reflects a universal approach to playing possession football.
Historically, the majority of goals are scored with 5 passes or less and the evidence suggests it is highly unlikely that will change in the future. Appendices 9–12 provide details of the number of goals in each category, where possession was regained in each third of the field and the number of goals from 0 passes, 1–5 passes or 6+ passes. The
Middle Third was divided into two areas, ‘Own Half of Midfield’ and ‘Their Half of
Midfield’ due to the large number of goals from regained possession in the Middle Third, to create four areas for comparative analysis and to make the distinction between goals from regained possession in each half of the field.
4.10.1 The number of goals and passes from regained possessions in four areas of the field
It was established earlier in the text (Tables 4.28 and 4.33) that the majority of goals in every competition came from regained possessions in the Middle Third of the field.
Data on regained possessions in each half of the field were reported (Breen et al., 2006;
Taylor et al., 2002) with particular reference to the Back Third. It was suggested earlier in the text (Section 4.8) that it is logical to expect more goals to be scored from sequences of six passes or more from regained possession in the Back Third because of the distance the ball has to travel before a goal can be scored. Tables 4.42 to 4.45 inclusive show the distribution of goals from 1–5 passes and 6+ passes in the Back Third (B3), each half of the Middle Third, Own Half (OH-M3) and Their Half (TH-M3) and the Final Third (F3) for each competition in this study and the total for goals in each half of the field. Zero pass goals are shown as a percentage of the total, for example 8/46=17% in Euro 2004, but are not included in the other calculations, which are expressed as percentages of the total GOAL SCORING PATTERNS 206
number of goals from passing sequences of 1–5 and 6 or more (6+), for example 9 (20%) goals from 1–5 passes from the B3rd is 9/46.
Table 4.42 Regained possessions and passes in European Championships24 EURO B3 OH-MF TH-MF F3 OH TH 2004 0 0 1 2 0 3 0 Passes 9 (20%) 9 14 6 18 20 1–5 Passes 7 (15%) 1 0 0 8 (100%) 0 6+ Passes 0 Passes 3/49 6% 8/46 17% EURO 2008 0 Passes 0 0 0 3 0 3 1–5 Passes 14 (25%) 17 7 6 31 13 6+ Passes 5 (9%) 6 1 1 11 (85%) 2 0 Passes 3/60 5% 13/57 23% EURO 2012 0 Passes 0 0 0 7 0 7 1–5 Passes 10 (19%) 4 10 7 14 17 6+ Passes 8 (15%) 5 8 2 13 (57%) 10 0 Passes 7/61 11% 23 of 54 43%
Table 4.43 Regained possessions and passes in the World Cup WC 2002 B3 OH-MF TH-MF F3 OH TH 0 Passes 0 0 0 15 0 15 1–5 Passes 22 (22%) 17 17 21 39 38 6+ Passes 12 (12%) 5 6 0 17 (74%) 6 0 Passes 15/115 13% 23/100 23%
WC 2006 0 Passes 0 0 0 7 0 7 1–5 Passes 11 (13%) 26 12 16 37 28 6+ Passes 10 (12%) 6 3 2 16 (76%) 5 0 Passes 7/93 8% 21/86 24%
WC 2010 0 Passes 0 0 6 3 0 9 1–5 Passes 19 (20%) 17 16 17 36 33 6+ Passes 7 (7%) 11 10 0 18 (57%) 10 0 Passes 9/106 8% 28/97 29%
24 B3 – Back Third, OH-MF – Own Half of Midfield, TH-MF – Their Half of Midfield, F3 – Final Third GOAL SCORING PATTERNS 207
Table 4.44 Regained possessions and passes in the EPL EPL 2001–02 B3 OH-MF TH-MF F3 OH TH 0 Passes 0 0 4 53 0 57 1–5 Passes 95 134 143 160 229 303 6+ Passes 11 19 12 10 30 (58%) 22 0 Passes 57/641 9% 52/584 9%
EPL 2005–06 0 Passes 0 1 3 52 1 55 1–5 Passes 100 136 145 114 236 259 6+ Passes 9 11 8 3 20 (65%) 11 0 Passes 56/582 9% 31/526 6%
EPL 2011–12 0 Passes 1 1 6 73 2 79 1–5 Passes 132 101 126 166 233 292 6+ Passes 69 49 46 22 118 (63%) 68 0 Passes 81/792 10% 186/711 26%
Table 4.45 Regained possessions and passes in the A League25 A Lge. 08–09 B3 OH-MF TH-MF F3 OH TH 0 Passes 0 0 0 12 0 12 1–5 Passes 23 42 31 34 65 65 6+ Passes 8 6 8 2 14 (58%) 10 0 Passes 12/166 24/154 16% 7% A Lge. 09–10 0 Passes 0 0 2 19 0 21 1–5 Passes 59 42 40 55 101 95 6+ Passes 13 10 6 1 23 (77%) 7 0 Passes 21/247 30/226 13% 9% A Lge. 2011– 12 0 1 4 18 1 22 0 Passes 32 39 55 49 71 104 1–5 Passes 34 16 12 3 50 (77%) 15 6+ Passes 0 Passes 23/263 65/240 27% 9%
25 B3 – Back Third, OH-MF – Own Half of Midfield, TH-MF – Their Half of Midfield, F3 – Final Third GOAL SCORING PATTERNS 208
4.10.2 Goals from 1–5 and 6+ passes from regained possessions in the Back Third of the field
The majority of goals from 6+ passes originated in the Back Third of the field 8 out
12 times. In the other four examples the majority originated in the own half of midfield
(OH-MF).
Earlier in the text (4.8.1) Table 4.39 showed how many goals were scored from regained possessions in the Back Third for all competitions and with the number of passes.
Table 4.40 provides an example of ‘A’ League goals from regained possessions in the
Back Third from sequences of 1–5 passes and 6+ passes. The results show that goals can be scored from a low number of passes from regained possession in the Back Third without kicking the ball in the air over a long distance from the Back Third, an assumption that can be made when a goal is scored from two or three passes and when detailed information to the contrary is unavailable. Table 4.46 shows the goals in the EPL and the
‘A’ League from regained possessions in the Back Third, the number of goals from 1–5 passes and 6+ passes and the number of goals from 1–5 passes after long passes, goal kicks or drop kicks by the goalkeeper have been taken out. Full details are presented in
Appendix 8.
Table 4.46 EPL and A League goals from regained possession in the Back Third Regained B3rd 1–5 passes 1–5 passes minus long 6+ passes forward passes EPL 2001–02 95 (15%) 79 (12%) 11 (1%)
EPL 2005–06 100 (17%) 78 (13%) 9 (2%)
EPL 2011–12 132 (17%) 110 (14%) 69 (9%)
A LGE. 2008–09 23 (14%) 14 (9%) 8 (5%)
A LGE.2009–10 59 (24%) 50 (20%) 13 (5%)
A LGE. 2011–12 32 (12%) 27 (10%) 34 (13%) GOAL SCORING PATTERNS 209
In the EPL and ‘A’ League, there was a considerable increase in the percentage of
goals from 6+ passes from regained possessions in the Back Third, up from 2% in 2005–06
to 9% in 2011–12 and from 5% to 13% in the ‘A’ League between 2009 and 2012.
However, with the exception of the ‘A’ League in 2011–12 there were more goals from 1–
5 passes than 6+ passes in each season of the EPL and the ‘A’ League.
When the goals that included a long forward pass or a kick from the goalkeeper,
were taken out of the calculations the percentages of goals from 1–5 passes were still
higher than from 6+ passes with the one exception in the ‘A’ League. Similar results were
found in the World Cups and European Championships in that more goals were scored
from 1–5 passes than 6+ passes but the gradual increase in the percentage of goals from 6+
passes in the league competitions was not replicated in the international tournaments. The
reason is most likely because international teams have played a possession-based style of
football in the tournaments and invariably in hot conditions so there has not been a change
in the approach to playing the game from an attacking perspective, more so in World Cups
than in the European Championships (see Table 4.47).
Table 4.47 Euro and World Cup goals from regained possession in the Back Third and the number of passes
Regained B3rd 1–5 passes 1–5 passes minus long 6+ passes forward passes
World Cup 2002 22 (19%) 18 (16%) 12 (11%)
World Cup 2006 11 (12%) 10 (11%) 10 (11%)
World Cup 2010 19 (18%) 12 (11%) 7 (6%)
EURO 2004 9 (19%) 8 (16%) 7 (14%)
EURO 2008 14 (23%) 13 (22%) 5 (9%)
EURO 2012 10 (16%) 0 8 (14%)
GOAL SCORING PATTERNS 210
Possession based football does not lead to more goals from sequences of 6+ passes than 1–5 passes, even though the percentage of goals from 6+ passes from regained possessions in the Back Third has increased over time in the league competitions in this study.
4.10.3 The percentage of goals in each category from regained possessions in the Back Third, Own Half of Midfield and Own Half of the field
Appendices 9 to 12 show the number of goals in each category, and how many were scored from 1–5 and 6+ passes in each area of the field and for each competition. A sample is provided in Table 4.48 to show the scoring pattern in the World Cups for the number of goals in each category from regained possession in the Back Third and own half of midfield.
Table 4.48 Goals in each category from regained possessions in the Back Third, OH of MF and Own Half and number of passes26 World Cup 2002 BBSPS 49 Other Methods 34 Crosses 32 Pass Category 1–5 6+ 1–5 6+ 1–5 6+ Back 3rd 19 11 8 7 7 0 8 4 4 Own Half of MF 10 9 1 5 3 2 7 5 2 Total in Own Half 29/49 59% 12/34 35% 15/32 47%
World Cup 2006 BBSPS 45 Other Methods 36 Crosses 12 Back 3rd 12 5 7 4 3 1 5 3 2 Own Half of MF 14 12 2 13 11 2 5 3 2 Total in Own Half 26/45 58% 17/36 47% 10/12 83%
World Cup 2010 BBSPS 54 Other Methods 38 Crosses 14
Back 3rd 14 12 2 9 5 4 3 2 1 Own Half of MF 16 11 5 6 4 2 6 2 4
Total in Own Half 30/54 56% 15/38 39% 9/14 64%
26 BBSPS- BB&S and BBP&S combined, BB&S- Ball behind and strike, BBP&S- Ball behind pass and strike, MF - Midfield
GOAL SCORING PATTERNS 211
The majority of goals in the World Cups from BBSPS were scored from regained possessions in the team’s own half of the field, 59% in 2002, 58% in 2006 and 56% in
2010, (Table 4.48) while a minority of goals from Other Methods were from regained possessions in the team’s own half, 35% in 2002, 47% in 2006 and 39% in 2010.
Table 4.49 shows the totals for goals in each category for all competitions to illustrate the consistency with which the majority of goals from BBSPS were scored from regained possessions in the teams own half regardless of whether the majority of goals overall were from regained possession in the teams own half or the oppositions half of the field.
For example in the six seasons in the EPL and ‘A’ League the majority of goals overall were from regained possessions in the opponent’s half of the field but in every case the majority of the goals in the BBSPS category were from regained possessions in the team’s own half of the field; ranging between 54% in the EPL in 2005–06 to 63% in the A
League in 2008–09. The trend applied in the international tournaments with the exception of Euro 2012 when the figure was 47% of goals in the BBSPS category, slightly less than
50% from Crosses.
GOAL SCORING PATTERNS 212
Table 4.49 Percentage of goals from regained possession in the teams own half of the field in each category BBSPS Other Methods Crosses
World Cup 2002 29/49 59% 12/34 35% 15/32 47%
World Cup 2006 26/45 58% 17/36 47% 10/12 83%
World Cup 2010 30/54 56% 15/38 39% 9/14 64%
EURO 2004 17/26 65% 7/16 44% 2/7 29%
EURO 2008 28/34 82% 3/9 33% 11/17 65%
EURO 2012 14/30 47% 7/19 37% 6/12 50%
EPL 2001–02 149/271 55% 78/264 30% 22/106 21%
EPL 2005–06 137/256 54% 82/230 36% 38/96 40%
EPL 2011–12 168/307 55% 125/350 36% 60/135 44%
A LGE 2008–09 45/72 63% 22/59 37% 12/35 34%
A LGE 2009–10 58/96 60% 40/98 41% 26/53 49%
A LGE 2011–12 60/110 55% 35/98 36% 26/55 47%
The figures show that fewer goals from Other Methods were from regained possession in the team’s own half of the field in every competition even when the majority of goals overall were scored from Other Methods, which happened in the EPL in 2011–12 and the A League in 2009–10. The percentage of goals from Crosses from regained possession is slightly in favor of the opponent’s half of midfield with only three competitions having more than 50% in the teams own half of midfield with an even split in
Euro 2012.
The most obvious explanation why there was a higher percentage of goals in the
BBSPS category from regained possessions in the own half of the field is that opposing teams were more vulnerable to a pass behind the last line of defence due to the available GOAL SCORING PATTERNS 213
space in their own half, when most of their players were attacking and in the opponents half of the field. Therefore it is somewhat surprising that the number of goals from playing the ball behind opponents from inside the Own Half (OH in the BBSPS category) was lower than from Zone 14+ in every competition. The majority of goals from BBSPS were from regained possessions in the Own Half of the field so one might expect a high number if not the majority of goals to be from playing the ball behind opponents from inside the
Own Half of the field, but that was not the case.
In every competition the percentage of goals from playing the ball behind opponents (BBSPS) from inside the Own Half of the field was below 30% and in 8 of the
12 competitions the figure was below 20%.
Table 4.50 shows the number of goals from regained possessions in the Own Half of the field and the percentage from BBSPS category that were played from inside the
Own Half (OH) and from Zone 14+.
GOAL SCORING PATTERNS 214
Table 4.50 Percentage of goals from regained possession in the teams own half of the field in each category27 Regained in Own Half BBSPS - OH BBSPS - Z14+
World Cup 2002 29/49 59% 4/29 14% 19/29 66% World Cup 2006 26/45 58% 5/26 19% 18/26 69% World Cup 2010 30/54 56% 5/30 17% 18/30 60%
EURO 2004 17/26 65% 3/17 18% 9/17 53% EURO 2008 28/34 82% 0/28 0% 14/28 50% EURO 2012 14/30 47% 2/14 14% 10/14 71%
EPL 2001–02 149/271 55% 33/149 22% 101/149 68% EPL 2005–06 137/256 54% 39/137 28% 70/137 51% EPL 2011–12 168/307 55% 28/168 17% 96/168 57%
A LGE. 2008– 45/72 63% 13/45 29% 20/45 44% 09 A LGE. 2009– 58/96 60% 12/58 21% 35/58 60% 10 A LGE. 2011– 60/110 55% 10/60 17% 36/60 60% 12
There are several explanations why there were more goals from Zone 14+ than
inside the Own Half (OH) despite more goals from regained possessions in the Own Half
of the field. Firstly there were fewer goals in any category from sequences that involved
long forward passes, which means the forward movement of the ball was by a player
running with the ball or passing the ball forwards, both of which would have transferred
the ball over the halfway line while opposing defenders were running back towards their
own half of the field. At the same time other defenders would have been trying to delay the
forward progress of the ball, recover the ball or challenge to stop the forward progress.
When a team is defending in its own half of the field, the majority of, if not all, the players
27 BBSPS- BB&S and BBP&S combined, BB&S- Ball behind and strike, BBP&S- Ball behind pass and strike, OH – Own Half
GOAL SCORING PATTERNS 215
will be in that half so when they regain possession of the ball a number of them will make forward runs immediately to get in the opposition’s half of the field to enable forward progress of the ball by passing. When goals are scored from five passes or less, from regained possession in the Own Half of the field, without making long forward passes there have to be forward runs by players off the ball. Forward runs by the player with the ball, from inside his/her own half, may explain why the majority of the goals from the
BBSPS category were scored from passes in Zone 14+, which is from inside the opponent’s half. Another reason why the ball would be transferred into the Zone 14+ before any attempt to play behind the opposition is because when defenders sense danger or if they are outnumbered by the opposition they will retreat towards their own goal to regroup before trying to win the ball. When defenders do this they make it very difficult for the opposition to pass the ball behind them because they will be favorites to win the ball. It is at times like this when attackers are making forward runs and get close to the defenders that defenders stop retreating in the hope that when the ball is passed the attackers will have run into an off-side position. This explains the high percentage of goals in the BBSPS category from passes in Zone 14+ compared with the goals from passes in the Own Half from regained possessions in the Own Half of the field. The details of regained possessions for each competition are in Appendices 9–12.
4.10.4 Goals from 1–5 and 6+ passes from regained possessions in the opponents’ half of midfield, ‘Their Half’ of midfield
The number of goals from 6+ passes was lower from regained possession in the opponents’ half of the field than regained possessions in the ‘own half’ in every competition. However, from the second last competition to the last competition the percentage of goals from 6+ passes doubled in the ‘A’ League from 13% to 27%, almost doubled in Euro 2012 going from 23% to 43%, quadrupled in the EPL from 6% to 26% but GOAL SCORING PATTERNS 216
in the World Cup the increase was much smaller, up from 24% to 29%. The results are shown in Tables 4.42 to 4.45 inclusive. It is difficult to explain why there would be a 43% increase in the number of goals from 6+ passes from Euro 2008 to 2012 and only a 14% increase in the World Cup from 2006 to 2010. One possible explanation for the difference could be the attitude of European national teams to keeping possession of the ball. There was a gradual increase in goals from 6+ passes from 17% in Euro 2004 to 23% in 2008 and an even bigger increase to 43% in 2012. Another reason for such a large increase in goals from 6+ passes could be a by product of teams dropping closer to the half way line to set up defensively rather than pressing the opposing teams as soon as possession was lost. If the defending team applied that strategy it would allow the opposing team more time on the ball and with less pressure from opponents, making it much easier for players to pass the ball from side to side to keep possession and make slow progress towards the Middle
Third of the field, which would increase the number of passes. The evidence indicates that the majority of goals from 6+ passes originate in the defending half the field, which one might expect, but that has been the situation in every European Championship and World
Cup. Closer inspection of the data in Tables 4.42 and 4.43 shows the biggest difference in
2012 compared with 2008 and 2004 was in the number of goals from 6+ passes from regained possession in the opponents half of the field.
In Euro 2004, there were 8 goals from 6+ passes from regained possessions in the opponents half, 13 goals in 2008 and 23 in 2012. These figures show an increase of approximately 50% every four years and in 2012 there were 10 goals from 6+ passes from regained possessions in the opponent’s half only three less than the 13 goals from regained possessions in the teams own half. In Euro 2004 and 2008 the figures were 8 to 0 and 11 to
2 for goals with 6+ passes from regained possessions in the teams own half compared with the opponents half. GOAL SCORING PATTERNS 217
In the 2010 World Cup, there was a threefold increase compared with the 2006
World Cup in the number of goals from 6+ passes in the opponents half of midfield, which reflected the trend in the European Championships and the domestic leagues, but not as strongly. This may have been because the World Cup includes teams from every
Confederation in the world whereas the European Championship is for the teams in the
UEFA Confederation. The EPL is in Europe and the Australian League is greatly influenced by European football on TV and in the media, through club and player connections to European clubs and culturally through European immigration.
When emphasis is placed on keeping possession of the ball it is possible teams might do that in preference to attacking quickly from a regained possession in the opponents’ half of the field and especially if the team is in the lead at the time.
Examination of the 8 goals from 6+ passes in Euro 2012 from regained possessions in the attacking half of midfield showed that the teams scoring the goals were only in the lead on
3 occasions. When the same criterion was applied to the 10 goals from 1–5 passes the teams scoring the goals were in the lead on 4 occasions, so the game score may not have had a major influence on how quickly the teams attacked or made sure they kept possession, in this example. Another situation worth considering is where the bulk of the players were on the field when possession was regained in the opponents’ half of midfield.
If the majority of the players were in the Final Third one might expect the player regaining the ball in midfield to be able to pass forward rather than sideways or backwards, which might happen more often if the bulk of players were in the Middle Third when possession was regained. Further examination of the 8 goals from 6+ passes in Euro 2012 from regained possessions in the attacking half of midfield showed that 6 of the goals were scored when the bulk of players were in the Middle Third and 2 goals when the bulk of players were in the Final Third. In 5 of the 6 cases when the bulk of players were in the GOAL SCORING PATTERNS 218
Middle Third a pass was made back into the team’s own half of the field; this happened once when the bulk of players were in the Final Third. When the same criterion was applied to the 10 goals from 1–5 passes there were 7 cases when the bulk of players were in the Middle Third and 3 when the bulk of players were in the Final Third; but there was only one occasion when the ball was played back into the team’s own half and that was from a throw-in. This evidence suggests that the location of the bulk of players when possession is regained in the opponents half may not be a good indicator of how many passes a team will make prior to scoring.
The direction of the first pass may be a better indicator of how many passes a team will make prior to scoring. In 8 of the 10 goals from 1–5 passes from regained possession in the opponents’ half of midfield the first pass was forwards.
When the first pass was played backwards in the other 2 goals, one was from a throw-in after which the ball was played forwards for a goal to be scored after 3 passes; it was Spain’s fourth goal against Italy. Interestingly, Spain the team renowned for keeping possession of the ball and the European Champions scored 6 goals from regained possession in the opponents’ half of midfield; 5 from 1–5 passes and 1 from 6+ passes. A noticeable difference between the goals from 1–5 passes and 6+ passes was how quickly other players made forward runs once possession had been regained, which is often the catalyst for forward passes behind the opposing defense. The same pattern was recorded in goals from 6+ passes but with less urgency. Tables 4.51 to 4.53 inclusive are examples to show the goals in Euro 2012, the 2010 World Cup and the ‘A’ League in 2011–12, from 1–
5 and 6+ passes after regained possession in the opponents half of midfield and the category of goals. The details for every competition are provided in Appendices 9–12.
GOAL SCORING PATTERNS 219
Table 4.51 Regained possession in the opponents half of midfield Euro 2012 1–5 Passes 6+ Passes Total
BBSPS 8 (62%) 5 (38%) 13 Other Methods 1 1 2 Crosses 1 2 3 Total 10 (8 Forwards) 8 (4 Forwards) 18
The higher number of goals from 1–5 passes (62%) in the BBSPS category was the
outcome of the player in possession passing the ball forwards and other players making
forward runs. The 8 goals in the BBSPS category were all from passes in Zone 14+, which
emphasizes the importance of central attacks once possession is regained. Three of the five
goals from 6+ passes were from Zone 14+; the other two were from Wide Left (WL) and
inside the team’s own half (OH). The difference between goals from 1–5 passes (10) and
6+ passes (8) is not huge but if players pass the ball forwards, when they can, rather than
backwards when they regain possession in the opponents half of midfield it might increase
their tally of goals. Similar results were recorded in the 2010 World Cup, in Table 4.52.
Table 4.52 Regained possession in the opponents half of midfield WC 201028 1–5 Passes 6+ Passes Total
BBSPS 11 (73%) 4 (27%) 15 Other Methods 5 5 10 Crosses 0 1 1 Total 16 (10 Forwards) 10 (5 Forwards) 26
In 10 of the 16 goals from 1–5 passes the first pass went forwards with 11 of the 16
goals in the BBSPS category; 6 from passes in Zone 14+.
28 BBSPS- BB&S and BBP&S combined, BB&S- Ball behind and strike, BBP&S- Ball behind pass and strike
GOAL SCORING PATTERNS 220
In 8 of the 10 goals when the first pass went forwards the bulk of the players were in the Final Third of the field. There were two goals from 1–5 passes when possession went back into the team’s own half and both were from regained possessions from a throw- in under pressure from opponents. There were other regained possessions from throw-ins but with little pressure from opponents so the ball could be thrown forwards or passed forwards easily by the player receiving the ball from the thrower. In the 10 goals from 6+ passes there were five occasions when the first pass went backwards and possession went back into the team’s own half each time. There were more goals from Other Methods than in the BBSPS category, which was the opposite in Euro 2012 but the direction of the first pass, either forwards or backwards was a clear indicator for the number of passes that would be made prior to goals scored in both international tournaments. It is understood that it is not always possible to pass the ball forwards after regaining possession but the recommendation would be to do so if possible and for other players to make forward runs when it is possible.
Table 4.53 shows the goals in the 2011–12 ‘A’ League season, from regained possessions in the opponent’s half of midfield when the number of goals was much higher with 1–5 passes (55) compared to 6+ passes (12). The first 12 goals in the timeline with 1–
5 passes were compared with the 12 goals with 6+ passes and the results were very similar to the international tournaments.
Table 4.53 Regained possession in the opponents half of midfield in the ‘A’ League 2011–12 and goals from 5 passes or less and 6+ passes
1–5 Passes (55) 6+ Passes (12) Total BBSPS 8 5 13 Other Methods 1 5 6 Crosses 3 2 5 Total 12 12 24
GOAL SCORING PATTERNS 221
In 11 of the 12 goals from 1–5 passes the first pass was forwards and the only time play went back into the team’s own half of midfield was when the first pass went backwards. Of those 12 goals 8 were scored in the BBSPS category, with 6 from Zone 14+.
The bulk of players in 9 of the 12 goals were in the Middle Third and 3 when the bulk were in the Final Third, which reinforces that the location of the bulk of players is not an indicator of the number of passes likely to be made preceding goals.
The 12 goals from 6+ passes produced a mixture of passes forward (6) and backwards (6) with a total of 7 occasions when play went back into the teams own half, which included 4 goals when the first pass went forwards.
This highlights the importance of continual forward passing or more passes forwards than backwards, which happens more often when goals are scored with 1–5 passes. The opportunity to pass forwards is greatly affected by the amount of pressure opponents apply to the player in possession of the ball and the willingness of other attackers to make forward runs when the ball can be passed forwards. In the three examples from international tournaments and league competition there were a higher number of goals in the BBSPS category from 1–5 passes than from 6+ passes with the majority in each competition from a pass in Zone 14+. Overall there were more goals in the BBSPS category than from Other Methods and Crosses, which highlights the importance of attacking quickly by passing forwards and making forward runs when possession is regained in the opponents half of midfield. The increase in the number of goals from 6+ passes from regained possessions in the opponents half of midfield (TH of
MF) appears to be influenced by the direction of the first pass away from the opponents’ goal rather than towards it. This is likely to happen more often when the emphasis of the team is on keeping possession of the ball rather than attacking quickly when the opportunity arises. GOAL SCORING PATTERNS 222
4.10.5 The percentage of goals in each category from regained possession in the opponents’ half of the field and the Final Third
The results in Tables 4.54 and 4.55 show that opportunities to play the ball behind the opponents’ last line of defence diminish as play gets closer to the opponents’ goal. The figures in each category show the total from regained possessions in the opponents’, or their half of the field (TH), with the total for regained possessions in the Final Third (F3).
The percentage of goals from ‘their half’ in the BBSPS category was less than 50% in 11 of the 12 competitions with the exception in Euro 2012 when 53% was recorded. The percentage of goals in the BBSPS category from regained possessions in in the Final Third was less than 50% of the total for that half of the field in 10 of the 12 competitions and by considerably less than 50% in some. For example, in Euro 2004 none of the goals were scored from regained possession in the Final Third. In Euro 2008, only 2 of the 6 goals from regained possession in their half (TH) were in the Final Third and in Euro 2012 only
3 of the 16 goals from their half were from the Final Third, which shows that this method of scoring goals becomes more difficult when possession is regained closer to the opponents’ goal. Scoring becomes more difficult because the space behind the defenders is smaller than when defenders are further away from their own goal. The goalkeeper is positioned better to intercept passes behind the defence and defenders are usually facing the attackers and not running back towards their goal, which allows them to see more of the players and move in any direction with greater ease. All of these factors impact negatively on the success rate of passing the ball behind the oppositions’ last line of defence to score goals.
The majority of goals from Other Methods were from regained possessions in their half (TH) of the field, ranging between 53% and 70% for all competitions.
In contrast to goals in the BBSPS category, the percentage of goals from Other
Methods from regained possessions in the Final third was more than 50% of the total for GOAL SCORING PATTERNS 223
that half of the field in 11 of the 12 competitions, with World Cup 2010 the exception. It has been established that scoring goals in the BBSPS category is harder the closer possession is regained to the opponents’ goal so it is to be expected that more goals from regained possessions in ‘their half’ would be scored through Other Methods.
Table 4.54 The percentage of goals from regained possession in the opponents Half, ‘their half’ (TH) and Final Third (F3) in World Cups and European Championships in each category29 BBSPS Other Methods Crosses WC 2002 TH 59 20/49 41% 22/34 65% 17/32 53% WC 2002 F3 36 8/49 16% 20/34 59% 8/32 25% WC 2006 TH 40 19/45 42% 19/36 53% 2/12 17% WC 2006 F3 25 12/45 27% 12/36 33% 1/12 8% WC 2010 TH 52 24/54 44% 23/38 61% 5/14 36% WC 2010 F3 20 9/54 17% 7/38 18% 4/14 29%
EURO 2004 TH 23 9/26 35% 9/16 56% 5/7 71% EURO 2004 F3 8 0/26 0% 5/16 31% 3/7 43% EURO 2008 TH 18 6/34 18% 6/9 67% 6/17 35% EURO 2008 F3 10 2/34 6% 5/9 56% 3/17 18% EURO 2012 TH 34 16/30 53% 12/19 63% 6/12 50% EURO 2012 F3 16 3/30 10% 10/19 53% 3/12 25%
There is another reason why the percentage of goals from Other Methods tends to be the highest in the Final Third. It is because ‘zero’ pass goals, which mainly occur in the
Final Third are a specific type of goals that can be included only in Other Methods because they do not involve passing the ball.
29 BBSPS- BB&S and BBP&S combined, BB&S- Ball behind and strike, BBP&S- Ball behind pass and strike, WC – World Cup, TH – Their Half, F3 – Final Third
GOAL SCORING PATTERNS 224
Table 4.55 The percentage of goals in each category from regained possession in the opponent’s half (TH) and Final Third (F3) in EPL and A League BBSPS Other Methods Crosses
EPL 2001–02 TH 392 122/271 45% 186/264 70% 84/106 79% EPL 2001–02 F3 223 59/271 22% 118/264 45% 46/106 43%
EPL 2005–06 TH 325 119/256 46% 148/230 64% 58/96 60% EPL 2005–06 F3 169 44/256 17% 97/230 42% 28/96 29%
EPL 2011–12 TH 439 139/307 45% 225/350 64% 75/135 56% EPL 2011–12 F3 261 65/307 21% 154/350 44% 42/135 31%
A LGE 2008–09 TH 87 27/72 37% 37/59 63% 23/35 66% A LGE 2008–09 F3 48 13/72 18% 22/59 37% 13/35 37%
A LGE 2009–10 TH 123 38/96 40% 58/98 59% 27/53 51% A LGE 2009–10 F3 75 20/96 21% 40/98 41% 15/53 28%
A LGE 2011–12 TH 142 50/110 45% 63/98 64% 29/55 53% A LGE 2011–12 F3 71 17/110 15% 41/98 42% 13/55 24%
Table 4.56 ‘Zero’ pass goals from 4 areas of the field and percentage of total in Open Play B3 OH of MF TH of MF F3 Total in Open Play WC 2002 0 0 0 15/20 75% 15/1 13% WC 2006 0 0 0 7/12 58% 7/93 8% WC 2010 0 0 6/16 3/7 43% 9/106 8% EURO 04 0 0 1/4 2/5 40% 3/49 6% EURO 08 0 0 0 3/5 60% 3/60 5% EURO 12 0 0 0 7/10 70% 7/61 11% EPL 01–02 0 0 4/68 53/118 45% 57/641 9% EPL 05–06 0 1/54 3/51 51/97 53% 55/582 9% EPL 11–12 1/71 1/54 6/71 72/154 47% 80/792 10% A League 0 0 0 12/22 55% 12/166 7% 2008–09 A League 0 0 2/18 19/40 48% 21/247 9% 2009–10 A League 0 1/15 4/22 18/41 44% 23/263 9% 2011–12
GOAL SCORING PATTERNS 225
4.10.6 Zero pass goals in Other Methods
The results in Table 4.56 represent the number of ‘zero’ pass goals from the total of goals from regained possessions in the four areas of the field. For example in the World
Cup 2002, 15 of the 20 goals (F3) were ‘zero’ pass goals from regained possession in the
Final Third and 15 of the 115 in Open Play represents 13% of the total.
It is very unusual for a goal to be scored from a regained possession outside the
Final Third and rare to find a goal in this category coming from the Back Third. The goal from a regained possession in the Back Third in the EPL in 2011–12 was scored by the goalkeeper when he drop kicked the ball and scored when the ball bounced over the opposing goalkeeper, something that does not happen very often. Similarly there were three goals from regained possessions in the ‘Own Half’ (OH) of midfield and quite a few more from ‘Their half’ (TH) of midfield, which usually happen when a player wins the ball and runs with it before scoring. The figures show that the majority of goals from this type of activity occur in the Final Third of field, closer to the opponents’ goal.
When the number of ‘zero’ pass goals for each competition are added together and shown as a percentage of the total goals in Open Play the range in the EPL and ‘A’ League is between 7% and 10%, but in the international tournaments the range is slightly wider, between 5% and 13%.
The number of ‘zero’ pass goals in this study is lower than it would have been if conventional criteria had been followed. For example, in this study, goals following a save by the goalkeeper were not included, nor were goals from rebounds off the goal posts, nor were deflections or rebounds off players inside the penalty area, including headers.
The number of ‘zero’ pass goals reported in other studies is often much higher. For example, 47 goals from a total of 169 were classified as ‘No Feed’ goals, which is 28% of the total including Set Plays (Wright et al., 2011). GOAL SCORING PATTERNS 226
4.10.7 Goals from Crosses and where possession was regained
In the three categories of goal scoring (Section 4.4) the lowest number of goals was
scored from Crosses in 11 of the 12 competitions analyzed, with the exception in Euro
2008; see Tables 4.3 to 4.6.
The majority of Crosses were scored from regained possessions in the opponents’
half of the field in 8 of the 12 competitions including the six league competitions, which
shows a level of consistency in the league competitions but not in international
competitions.
Tables 4.57 and 4.58 show where Crosses were regained and the percentage for
each half of the field. Full details are in Appendices 9–12.
Table 4.57 Crosses and the areas where possession was regained in World Cups and European Championships30 Back OH-MF Own Half TH-MF Final Their Half 3rd 3rd WC 2002 8 7 15 47% 9 8 17 53% WC 2006 5 5 10 83% 1 1 2 17% WC 2010 3 6 9 64% 1 4 5 36%
EURO 2004 1 1 2 29% 2 3 5 71% EURO 2008 4 7 11 65% 3 3 5 35% EURO 2012 2 4 6 50% 3 3 6 50%
30 WC – World Cup, OH-MF, Own Half of Midfield, TH-MF, Their Half of Midfield
GOAL SCORING PATTERNS 227
Table 4.58 Crosses and the areas where possession was regained in the EPL and ‘A’ League31 Back OH-MF Own Half TH-MF Final Their Half 3rd 3rd EPL 01–02 10 22 32 21% 28 46 74 79% EPL 05–06 15 23 38 40% 30 28 58 60%
EPL 11–12 34 26 60 44% 33 42 75 56%
A LGE 08– 3 9 12 34% 10 13 23 66% 09 A LGE 09– 14 12 16 49% 12 15 27 51% 10 A LGE 11– 17 9 26 47% 16 13 29 53% 12
There was not a trend for regained possessions in any of the four areas of the field
but other aspects of ‘Crosses’ warrant mention. For example, in 9 of the 12 competitions
there were more goals from Crosses on the right side of the field than the left side with an
equal number on both sides in the World Cup 2010. The percentage of Crosses referred to
as ‘Out-swingers’, those that curve away from the goalkeeper was considerably higher than
‘In-swingers’, the ones that curve towards the goalkeeper in every competition. The range
was between 62% and 92% with 8 of the competitions between 62% and 77%. The success
rate of ‘Out-swingers’ may be due to the fact that the ball curves away from the goalkeeper
and other defenders rather than towards them. The number of headers from Crosses was
calculated on the total of ‘In-swingers’ and ‘Out-swingers’ because it would be unrealistic
to expect a header from a Cross, which was passed into the penalty area. The details are in
Appendix 13 and the figure in brackets represents the number of headers. The range of
headed goals from Crosses in all competitions was between 29% and 55% and the number
of headers from ‘Out-swingers’ was less than 50% in 11 of the 12 competitions. The
number of goals from ‘In-swingers’ was much lower than ‘Out-swingers’ with all
31 OH-MF, Own Half of Midfield, TH-MF, Their Half of Midfield GOAL SCORING PATTERNS 228
competitions under 30%, and ten of the twelve under 20%. There were fifty percent or more headed goals from ‘In-swinging’ crosses in ten of the twelve competitions. One explanation for the lower percentage of headers from ‘Out-swinging’ crosses is the ball can be played into the space between the defenders and the goalkeeper and it does not have to be in the air, which allows forward players more opportunities to score with their feet, whereas ‘In-swinging’ crosses are usually played when the defenders are already between the ball and the goal so the ball has to be played over the defenders in the air. This would explain why there were fewer goals from ‘In-swinging’ crosses and why most of them were scored with headers.
4.10.8 Goals from ‘zero’ passes, 1–5 and 6 or more passes from regained possessions in the Final Third of the field
Regained possessions in the Final Third of the field are often within striking range of the goal so the majority of goals are often scored with less than five passes or none at all.
Table 4.59 shows the number of goals from zero passes, 1–5 and 6 or more for each category.
The figures how that the percentage of goals from 6+ passes is very low with a range between 0% and 12% in all competitions with 8 of the 12 competitions below 5% of all goals from regained possession in the final third. The range of goals from ‘zero’ passes is between 15% and 44% with 9 competitions between 24% and 31%.
GOAL SCORING PATTERNS 229
Table 4.59 Goals and number of passes from regained possession in the Final Third Zero 1–5 6+ Passes Passes Passes World Cup 2002 15 42% 21 58% 0 World Cup 2006 7 28% 16 64% 2 8% World Cup 2010 3 15% 17 85% 0
EURO 2004 2 25% 6 75% 0 EURO 2008 3 30% 6 60% 1 10% EURO 2012 7 44% 7 44% 2 12%
EPL 2001–02 53 24% 160 72% 10 4% EPL 2005–06 52 31% 114 67% 3 2% EPL 2011–12 73 28% 166 64% 22 8%
A LGE 2008–09 12 25% 34 71% 2 4% A LGE 2009–10 19 25% 55 73% 1 2% A LGE 2011–12 18 25% 50 71% 3 4%
The vast majority of goals were scored with 1–5 passes with a range between 44%
and 85%. The main reasons are: the ball has less distance to travel; and defenders have less
time to react and organize themselves even when possession is regained from a clearance
because the tendency is for players to advance forwards before it is clear who will win the
ball. This situation can leave the defence vulnerable because of the temptation to focus on
the ball to the exclusion of marking opponents. Players tend to move forwards quickly
when the ball is cleared from the penalty area and are unable to defend properly when the
ball is played back in.
GOAL SCORING PATTERNS 230
4.10.9 Goals from ‘zero’, 1–5 and 6+ passes for top and bottom three teams in leagues and top four in international competitions
The top three teams in the EPL and A League have followed the overall trend in each competition and with a higher percentage of goals from 6+ passes. Table 4.60 shows the collective scores for the top three teams in the EPL and the overall percentage for each category of passes for three seasons. In 2001–02 the top three scored 14% from 6+ passes and the average was 7%, in 2005–06 the top three scored 6% and the average was 5% and in 2011–12 the top three scored 29% compared with 24% overall.
Between 2001 and 2005 the average or overall percentages were similar but from
2005 to 2011 there was a big increase from 5% to 24% in the goals from 6+ passes with an even bigger increase by the top three teams from 6% to 29%.
Table 4.60 The top three teams in the EPL and the overall scores for goals from ‘Zero’, 1-5 and 6+ passes EPL 2001–02 2005–06 2011–12
Top 3 Ave. Top 3 Ave. Top 3 Ave.
Zero Passes 9 6% 9% 12 8% 10% 17 8% 10% 1–5 Passes 123 80% 84% 124 86% 85% 128=63% 66%
6+ Passes 21 14% 7% 9 6% 5% 58 29% 24%
Even though there was an overall increase from 5% to 24% in goals from 6+ passes in the EPL in 2011–12 and an increase from 6% to 29% for the top three teams, there were still more than twice as many goals overall from 1–5 passes and by the top three teams. To put the increase in 6+ passes in perspective, in 2005–06 overall there were 17 times as many goals from 1–5 passes and 14 times as many by the top three teams.
Table 4.61 shows in the ‘A’ League in 2011–12 there were almost three times as many goals overall from 1–5 passes compared with 6+ and less than double the number by the top three teams. Those figures show how significant the increase was for goals from 6+ GOAL SCORING PATTERNS 231
passes because in 2009–10 there were nearly seven times as many goals overall from 1–5 passes and four times as many by the top three teams.
Table 4.61 The top three teams in the ‘A’ League and the overall scores for goals from ‘Zero’, 1–5 and 6+ passes ‘A’ League 2008–09 2009–10 2011–12
Top 3 Average Top 3 Average Top 3 Average
0 Passes 6 9% 7% 3 4% 9% 7 7% 9%
1–5 Passes 50 75% 78% 68 78% 79% 56 59% 67%
6+ Passes 11 16% 15% 16 18% 12% 32 34% 24%
The time between competitions in the ‘A’ League was shorter than in the EPL, which may explain why there was less of an increase in the number of goals from 6+ passes in the ‘A’ League compared with the previous competition. The evidence in both competitions shows there has been a shift towards more goals from 6+ passes and the top three clubs have scored more than the overall average. The details of goals scored from different numbers of passes and where possession was regained are provided for each year of competition in Appendices 15 to 18.
Table 4.62 shows the results for goals in the European Champions over an eight- year span and the percentage scored from 6+ passes. The overall percentage increased from 16% in 2004, to 22% in 2008 to 38% in 2012 reflecting the same trend as in the league competitions but with a higher total percentage, for example 38% in 2012, compared with 24% in both the EPL and the ‘A’ League in 2011–12 seasons.
The figures for the top four teams combined for the number of goals from 6+ passes have followed the overall trend and been higher than the tournament average, for example 34% in 2012 compared with 24% overall.
GOAL SCORING PATTERNS 232
Table 4.62 The top 4 teams in the European Championships and the overall percentages for goals from ‘Zero’, 1–5 and 6+ passes EUROS 2004 2008 2012
Top 4 Average Top 4 Average Top 4 Average
0 Passes 1 5% 6% 1 3% 5% 1 4% 11%
1–5 Passes 18 90% 78% 21 68% 73% 14 56% 51%
6+ Passes 1 5% 16% 9 29% 22% 10 40% 38%
Table 4.63 shows the results for goals in World Cups from 2002 to 2010. There was a higher figure for goals from 6+ passes in 2002, 20% compared with 7% in the EPL and slightly higher than the 16% recorded in Euro 2004. The trend was reflected in the results for the top four teams in the World Cup with 22% of goals from 6+ passes. More recently, the percentage of goals from 6+ passes has been closer. In the World Cup 2010 there were 27% overall and 32% by the top four teams compared with 24% and 34% respectively in the EPL 2011–12 season and 38% and 40% respectively in Euro 2012. The percentage of goals from 6+ passes in the World Cups increased by 30% over eight years compared with an increase of approximately 350% in the EPL over a period of ten years.
These figures show the percentage of goals from 6+ passes, has not changed much over three competitions in the World Cup but there has been a huge increase overall in the EPL and particularly by the top three teams.
Table 4.63 The top 4 teams in the World Cups and the overall percentages for goals from ‘Zero’, 1–5 and 6+ passes World Cups 2002 2006 2010
Top 4 Average Top 4 Average Top 4 Average
0 Passes 3 8% 13% 2 7% 10% 2 5% 8%
1–5 Passes 26 70% 67% 20 74% 68% 24 63% 65%
6+ Passes 8 22% 20% 5 19% 22% 12 32% 27% GOAL SCORING PATTERNS 233
The top four teams combined in international competitions followed the trend of league competitions in that the percentage of goals from 6+ passes increased over time and was higher than the tournament average, with two exceptions; in Euro 2004 and the World
Cup 2006. A different aspect of scoring goals that was consistent throughout all competitions was the number of goals scored within the penalty area.
4.11 Goals scored inside and outside the penalty area
In every competition in this study, more than 80% of goals in Open Play were scored from inside the penalty area. This is consistent with earlier studies (see for example,
Breen et al., 2006; Mitrotasios & Armatas, 2014). This evidence makes football highly predictable in terms of scoring goals. The results for all competitions are in Appendix 21, which also contains the percentages for goals scored between 18 yards (the edge of the penalty area) and within 23 yards and for goals scored from 23 yards or more. The distances from goal are measured according to the position of the ball relative to imaginary parallel lines with the goal line. The average figure for goals between 18–23 yards was
10% and for 23 yards or more the figure was 6%. An example of goals scored in the World
Cups is shown in Table 4.64.
Table 4.64 The distance from the goal line for goals in the World Cups 2002–2010 Inside Pen. Area 18–23 yards 23 yards +
World Cup 2002 100 87% 11 10% 4 3% World Cup 2006 75 81% 8 9% 10 11%
World Cup 2010 87 82% 9 8% 10 10%
Regardless of tactics, playing style or the speed of the game this aspect of scoring goals has not changed over time. The low success rate of scoring from distances outside 23 yards reflects how difficult it is to beat a goalkeeper with a shot from outside the penalty GOAL SCORING PATTERNS 234
area and explains why successful teams take more shots from inside the penalty area than unsuccessful teams (Hook & Hughes, 2001).
4.12 Goals from Set Plays and Open Play
This study is concerned primarily with goals in Open Play but goals from Set Plays are included to account for all goals scored and because they play an important part in the outcome of matches. Appendix 22 provides a breakdown of Set Plays in each competition into Corner Kicks, Free Kicks, Penalties and Long Throws, with a sub-total of goals from
Direct Free Kicks.
The results show that in at least two out of three years in each competition, Set
Plays accounted for 20%–30% of the total scored and in one year the figure was over 30%.
The average for Set Plays in all competitions was 28%.
Table 4.65 shows the percentage of goals from Set Plays and Open Play for all competitions and Table 4.66 shows the percentages for each category of Set Plays.
Table 4.65 Average percentage of goals from Set Plays and Open Play Set Plays Open Play Set Plays Open Play
World Cups 31% 69% EPL 26% 74% Euros 26% 74% A Lge. 30% 70%
Table 4.66 Percentage of goals in each category of Set Plays Corners Free Kicks Penalties Throw Ins
World Cups 31% 40% 27% 2%
Euros 35% 40% 23% 2%
EPL 38% 33% 24% 5%
A Lge. 29% 36% 34% 1%
GOAL SCORING PATTERNS 235
The most productive category for goals from Set Plays was from Free Kicks in three of the four competitions, with Corners the highest category in the EPL. The lowest percentage of goals was from Long Throw-Ins followed by Penalties and then Corner
Kicks.
4.13 Summary of Chapter
This chapter has produced the evidence to answer the primary research questions and the relevant components of each question.
For ease of reference the answers to the research questions are in the following order: