<<

2008 NHL Review

Alan Ryder HockeyAnalytics.com

Copyright Alan Ryder 2008

2008 NHL Review Page 2

Table of Contents

Introduction 3 Team Performances 5 Lucky and Unlucky Teams ...... 5 Offense ...... 7 Shots and Quality ...... 8 Defense ...... 10 Goaltending ...... 11 The Shootout ...... 14 Individual Performances 16 Top Forward ...... 16 Top Defensive Forward ...... 20 Top Defenseman ...... 22 Top Defensive Defenseman ...... 24 Top ...... 26 Clean Play Awards ...... 27 Top Rookie ...... 30 Most Improved ...... 31 Shootout Awards ...... 33 All Star Contributors 34 NHL ...... 34 East ...... 35 West ...... 36 Rookie ...... 36 Green ...... 37 Grey ...... 38 All Offense ...... 38 All Defense ...... 39 Even Handed ...... 40 Power Play ...... 40 Short Handed ...... 41 Most Valuable Performances ...... 42 All Cap Roster ...... 43 Watch 46

Copyright Alan Ryder, 2008 Hockey Analytics www.HockeyAnalytics.com 2008 NHL Review Page 3

Introduction

This paper provides a commentary on the most outstanding individual performances in the NHL during the 2007-08 (“2008”) . I will also be commenting on certain aspects of team performance since individual performances are difficult to assess without understanding the team context.

My tool for measuring individual player impact is Player Contribution1 (PC). The method is a system of credits and debits. Offensively, a player is credited for goals and assists but debited for ice time (greater ice time, especially for forwards and on the power play, means greater expectations). Defensively, players get credited for ice time but debited for goals scored while on ice. are measured based on their save percentage, adjusted for the quality of team defense, in excess of a threshold.

‘PCO’ is a measure of offense – it is based on goals created in excess of a threshold level of performance. ‘PCD’ and ‘PCG’ are, respectively, based on ‘goals prevented’ in excess of a threshold for defense and goaltending. PC is also determined for taking and drawing. In this case, however, the benchmark is essentially the average propensity to take/draw penalties (rather than the marginal propensity).

Threshold performance is a mathematical construct. It is determined by observing that (a) the marginal impact of more/fewer goals on wins is nearly linear over the normal performance range of teams and (b) the “slope” of that linear relationship is the average number of goals scored per game. Marginal or threshold performance is determined by extrapolating the linear relationship between goals and wins until it predicts zero wins.

Although not really true, you can think of a ‘marginal’ player as a borderline NHLer. It is difficult to be precise about where the borderline is, so the PC method draws a line in the sand somewhere near that line. Why subtract out marginal performance? Because, in theory, performance at that level is worth ‘nothing’. Marginal players sit on the end of the bench and / or spend a great deal of time in transit to / from the minors. Marginal performance is so far from average as to be zero valued.

Where we can, PC is measured separately for even handed (EH), short handed (SH), power play (PP) and shootout (SO) situations. This ensures that specialty team performance is assessed relative to marginal performance on specialty teams. In other words, a player who runs up big offensive numbers on the power play is judged against other power play performances while a penalty killing specialist has his offense judged against other penalty killers.

1 PC is described in http://www.HockeyAnalytics.com/Research_files/Player_Contribution.pdf

Copyright Alan Ryder, 2008 Hockey Analytics www.HockeyAnalytics.com 2008 NHL Review Page 4

On defense, short handed situations are further subdivided into penalty killing (PCDSHK) and penalty taking (PCDSHO). On offense, power play situations are further subdivided into power play production (PCOPPP) and penalty drawing (PCOPPO).

Finally, since advancing in the standings is the objective, PC is denominated in points in the standings (that is goals created/prevented are translated into points in the standings). A PC “” is scaled to be 1/10th of a standings point and the PC points allocated to a team are therefore 10 x Points. To get a lot of PC points one needs to both (a) play a lot and (b) play well. As (a) and (b) tend to be correlated, PC is also a measure of ‘talent’. However, coaches tend to over play top talent with a resulting distortion of apparent relative 2008 Red Wings value. Cost per As a rough rule of thumb it takes Player Cap Cost PC PC Point 100 or more PC points for a skater 6,700,000 132 50,786 to be an all-star candidate (the story Nicklas Lidstrom 7,600,000 121 62,663 with goaltenders is different). At 80 2,650,000 121 21,901 points you would consider a skater 6,000,000 105 57,116 to be a team star, 60 is a team 850,000 75 11,289 leader, 40 is a solid contributor and 3,000,000 74 40,386 20 is a weak link. With a 1,150,000 57 20,279 Johan Franzen 941,667 54 17,538 of $50 million (all figures U.S.) for 662,500 53 12,602 the 2008 season, a rough guide to 733,333 49 14,966 player value is $50,000 per annum Jiri Hudler 1,015,000 47 21,591 per PC point (or $1,000,000 for Brett Lebda 650,000 46 14,048 every 20 PC points). This is based 1,200,000 38 31,880 on a team spending the cap amount Dominik Hasek 4,100,000 37 109,963 and targeting a 100 point season (a Andreas Lilja 1,000,000 34 29,395 comfortable berth in the ). Tomas Holmstrom 2,250,000 29 77,590 A conference winner will need to Drake 1,100,000 17 64,945 get more value for players. And, of Kris Draper 2,128,000 14 152,415 course the market value of a player 483,333 14 34,848 may be different due to supply and Kirk Maltby 883,333 13 69,296 demand and other factors. 733,333 12 63,279 Tomas Kopecky 500,000 7 69,515 To put this in perspective I have 3,500,000 6 546,129 shown, to the right, cap costs, the Matt Ellis 475,000 3 145,745 PC scores (rounded to the nearest 495,000 3 156,041 integer) and cap costs per PC point 520,000 2 241,420 (I have shown negative cost per PC 511,667 1 642,473 points results as 999,999) for the 500,000 1 856,065 . Note that cap Darren McCarty 892,192 0 1,921,822 costs in this table are (a) annualized Mark Hartigan 550,000 0 999,999 (the actual “cap hit” depends on Justin Abdelkader 450,000 -1 999,999 days on the roster, which is a big Darren Helm 599,444 -2 999,999 factor for those at the bottom of the 525,000 -12 999,999

Copyright Alan Ryder, 2008 Hockey Analytics www.HockeyAnalytics.com 2008 NHL Review Page 5 table) and (b) not salary (cap costs are the average salary/bonus over the contract).

Detroit was the NHL’s top team in the regular season, finishing with 115 points. Note the inexpensive performances the Red Wings got from Zetterburg, Osgood, Chelios, Franzen, Cleary, Filppula, Hudler and Lebda. Datsyuk was right-priced at $6.7 million and the two expensive defensemen, Lidstrom and Rafalski, were not far above fair price.

The Red Wings were a max-cap team (in fact I estimate that they were slightly over cap). Let’s 2008 Maple Leafs compare them to another (slightly under) max-cap team that missed Cost per the play-offs. Player Cap Cost PC PC Point 1,375,000 109 12,590 The collected Tomas Kaberle 4,250,000 93 45,883 83 points on the basis of very few 5,500,000 80 68,353 cheap performances (Toskala, 2,050,000 60 34,090 Steen, White, Stajan) and a bunch 900,600 56 16,044 of pay-cheques that are difficult to Pavel Kubina 5,000,000 51 98,919 Jason Blake 4,000,000 48 82,541 justify (starting with defensemen 850,000 44 19,422 McCabe and Kubina). 2,075,000 43 48,106 Bryan McCabe 5,750,000 37 154,057 875,000 36 24,563 Team Performances 2,105,000 35 59,758 Darcy Tucker 3,000,000 34 87,879 900,000 24 37,850 Although I want to focus on 1,283,333 21 62,032 individual performance it sets the Kyle Wellwood 875,000 21 42,475 stage best if I first look at some 575,000 20 28,986 team metrics using a marginal Anton Stralman 731,667 18 40,101 goals analysis. Marginal goals, a Dominic Moore 700,000 18 38,628 building block for Player Jiri Tlusty 855,000 15 55,708 Contribution, are, for offense, Staffan Kronwall 487,500 5 101,086 simply goals scored in excess of a Andy Wozniewski 500,000 4 117,209 threshold or, for defense, goals Simon Gamache 625,000 3 230,899 allowed subtracted from a Jeremy Williams 487,500 1 879,940 threshold. Robbie Earl 600,000 0 1,241,117 Alex Foster 660,000 0 999,999 Lucky and Unlucky Teams Darryl Boyce 525,000 0 999,999 Ben Ondrus 475,000 0 999,999 Wins are about 94% predicted by Johnny Pohl 462,500 -1 999,999 goals for and against (or by 2,166,667 -1 999,999 marginal goals totals). When a Bates Battaglia 1,200,000 -2 999,999 team wins in spite of a low Kris Newbury 475,000 -3 999,999 marginal performance it is 625,000 -6 999,999 either very skilled at winning close Scott Clemmensen 500,000 -10 999,999 games or it is lucky. Historical 2,000,000 -24 999,999

Copyright Alan Ryder, 2008 Hockey Analytics www.HockeyAnalytics.com 2008 NHL Review Page 6

analysis suggests that this is mostly luck. I would not completely rule out some Marginal Goals per Point intangible, but nobody has found it yet. Lucky teams tend to regress the following In Skating Time In Shootout season (and vice versa). But these teams Team per Skating Point per Shootout Win may also be systemically able to win tight ANA 2.51 1.78 ATL 2.45 2.28 games. BOS 2.43 1.49 BUF 2.80 2.25 To the right is a table of the marginal goals CAL 2.47 2.11 scored per point during the conventional CAR 2.52 1.78 part of the game (‘skating time’) and during CBJ 2.64 2.31 the shootout. The five most efficient/lucky CHI 2.72 2.10 teams are highlighted in green and the five COL 2.63 2.52 least efficient/lucky teams are highlighted DAL 2.78 2.13 DET 2.69 1.94 in red. EDM 2.69 2.35 FLA 2.68 2.27 During skating time the Islanders were the LA 2.82 2.08 most ‘efficient’ team in the NHL, requiring MIN 2.45 1.27 only 2.32 marginal goals per point. MON 2.67 2.31 (2.43) and (2.45) repeated from NAS 2.57 1.45 2007 as lucky teams. The NJD 2.51 2.27 NYI 2.32 2.49 flipped from being one of the most unlucky NYR 2.67 2.03 teams in 2007 to being one of the luckiest OTT 2.63 2.70 ones (2.45). The Maple Leafs were also PHI 2.62 1.36 top five in skating time winning efficiency PHO 2.65 2.86 (2.44). PIT 2.64 2.62 SJ 2.47 2.14 Los Angeles was the unluckiest team in the STL 2.54 2.42 NHL, requiring 2.82 marginal goals to TB 2.58 1.86 TOR 2.44 2.27 fashion a point. Buffalo (2.80) repeated as VAN 2.73 2.55 an unlucky team. (2.73) went WAS 2.60 2.92 from lucky in 2007 to unlucky in 2008. AVG 2.60 2.21 (2.72) and Dallas (2.78) rounded out the bottom five in skating time winning efficiency.

Average marginal goals per point are determined by dividing goals scored by regulation time and points and were 2.60 (down from 2007).

The Minnesota Wild were double lucky, also ranking as the luckiest team in the shootout. They required just 1.27 marginal goals to collect a shootout win. The Bruins were also double lucky, needing only 1.49 marginal goals for a shootout point. Philadelphia (1.36) repeated from 2007 as one of the luckiest teams in the shootout. Carolina was awful in the 2007 shootout, scoring only once and failing to register a win. This season they were one of the luckiest teams (1.78). The other top-five team in shootout luck was Nashville (1.45).

Copyright Alan Ryder, 2008 Hockey Analytics www.HockeyAnalytics.com 2008 NHL Review Page 7

Washington went from the luckiest shootout team in 2007 to the unluckiest in 2008 (2.92 marginal goals per shootout point). Pittsburgh also went from being one of the luckiest shootout teams to being one of the least lucky (2.62). Vancouver (2.55) was double unlucky in 2008 and repeated, with (2.70), as one of the NHL’s unluckiest shootout teams. Phoenix (2.86) was the other bottom-five team in shootout efficiency.

Average marginal shootout goals per shootout win (shootout goals divided by number of shootouts) were 2.21 (down from 2007).

There is no correlation between marginal goal efficiency in 2007 and 2008. This is part of the evidence that (in)efficiency is just (bad) luck.

Note that PC attempts to allocate team performance, whether Marginal Goals - Offense lucky or skilled, to individuals. It does not Offense MGO MGOEH MGOPPP MGOPPO MGCSH set out to determine OTT 128 91 26 -2 13 whether a performance MON 127 70 51 2 3 is from luck or skill. It DET 122 77 41 4 0 translates a player’s BUF 121 87 28 2 4 marginal goals into PC CAR 120 78 36 7 -1 points using these PHI 115 59 44 3 8 marginal goal factors. PIT 110 68 38 3 1 The implicit assumption WAS 108 79 29 -1 0 here is that these DAL 107 71 27 0 8 observed team CHI 104 70 20 1 12 performances are a result TOR 98 74 25 -1 -1 of skill. This means that NAS 97 78 16 1 2 a goal scored (or CAL 96 69 23 0 4 prevented) by an LA 96 62 26 2 6 Islander is worth more COL 94 84 14 -2 -3 than a goal scored (or TB 91 68 26 -6 2 prevented) by a King. MIN 90 64 29 -1 -2 EDM 90 67 21 -1 2 Offense SJ 86 50 31 2 2 FLA 81 58 28 -4 -1 A marginal goals PHO 79 46 28 -1 6 analysis helps us to VAN 77 47 25 2 3 deconstruct offenses. To ATL 77 53 19 -4 8 the right is a summary of BOS 76 52 23 -3 4 marginal goals offense NYR 75 51 23 2 -1 (MGO) by situation – STL 72 63 12 -2 -2 even handed (EH), NJD 68 53 17 -3 1 power play production ANA 67 40 23 1 3 (PPP), power play CBJ 60 40 16 0 3 NYI 59 42 14 -2 5

Copyright Alan Ryder, 2008 Hockey Analytics www.HockeyAnalytics.com 2008 NHL Review Page 8

opportunities (PPO) and short handed (SH).

While even handed Ottawa, Buffalo and were the most dangerous teams in the NHL with 91, 87 and 84 marginal goals respectively. However each of these teams was much less effective on the power play.

Montreal won the East on the basis of a league leading power play (so much for the departure of ), nearly catching Ottawa for the league lead in goal scoring. In 2007 the Canadiens had the dubious distinction of being the only NHL team with more marginal goals on the power play than while even handed. They were better on the power play in 2008 but even better even handed.

The Senators’ prevailed over as the NHL’s leading offense on the basis of short handed scoring (13 marginal goals on the basis of a league leading 18 goals). Chicago also racked up impressive scoring totals while on the penalty kill.

Buffalo demonstrated that a good offense (ranked 4th overall) does not guarantee a playoff spot.

Weak offensive teams included Anaheim (67 MGO) and New Jersey (68), proving that defense can get a team to the playoffs. The Devils were especially weak on the power play. The Ducks tied the Blue Jacket for the worst even handed offense.

MGOPPO (Marginal Goals on Offense from Power Play Opportunities) measures a team’s relative ability to generate power play opportunities (draw penalties). Carolina won an extra game by their ability to draw penalties. This was the third year in a row for a strong performance by the Hurricanes in this metric. Tampa Bay lost an extra game by its inability to draw penalties.

Shots and Shot Quality

Marginal goals analysis is an even more helpful tool for assessing defenses. But, to get at MGD, it is necessary to separate goal prevention into defense and goaltending. To identify a team’s contribution from goaltending I compare its goals against, adjusted for shot quality, to a threshold level (based on the shots allowed).

A high (shot quality neutral) save percentage against high shot totals suggests a big goaltending contribution. A low (shot quality neutral) save percentage against low shot totals suggests a small goaltending contribution. And whatever is not goaltending must be attributable to defense.

If you do this math you are attributing to defense the responsibility for the number and quality of shots on goal. Below are the shots and shot quality leader boards for the 2008 season.

Average shots on goal per game is a familiar metric. In the table below we see Detroit, San Jose, NY Rangers and Dallas with strong performances again in 2008. Tampa, Anaheim and Toronto sagged from near the top of this list in 2007 to middling

Copyright Alan Ryder, 2008 Hockey Analytics www.HockeyAnalytics.com 2008 NHL Review Page 9 performances in 2008. The most shots allowed were by Defensive Measures Atlanta, Florida, Philadelphia, Montreal and Avg Shot Defensive . The Habs and Team Shots Team Quality Team Index Flyers are repeat offenders. DET 23.5 CBJ 0.888 SJ 22.9 Two years ago the Oilers SJ 24.2 MIN 0.902 DET 23.2 were near the top of this NYR 25.9 CAL 0.930 CBJ 24.3 list. Under Bruce DAL 26.1 WAS 0.945 WAS 26.0

Boudreau Washington COL 27.3 VAN 0.946 CAL 26.5 made the biggest move this CBJ 27.4 SJ 0.947 NJD 26.6 STL 27.4 PHO 0.958 NYR 26.6 year (from 29th to 9th) and NJD 27.5 BOS 0.960 MIN 27.2 made the playoffs. WAS 27.5 NJD 0.966 VAN 27.3 ANA 28.0 ATL 0.972 DAL 27.4 Shot quality is based on an TB 28.2 NYI 0.977 TB 27.6 assessment of the BUF 28.4 NAS 0.977 BUF 27.7 characteristics of each shot 2 CAL 28.5 BUF 0.978 COL 27.7 allowed . The ‘expected CHI 28.6 TB 0.979 CAR 28.9 goals’ from this VAN 28.9 PIT 0.979 NAS 29.1 assessment, normalized for CAR 28.9 DET 0.986 BOS 29.2 variations in shots on goal, TOR 29.3 MON 0.988 STL 29.4 can be compared to NAS 29.8 PHI 0.995 PHO 29.4 average. My shot quality OTT 30.0 CAR 1.002 NYI 29.6 factors are the ratio of these MIN 30.1 FLA 1.008 ANA 29.7 normalized expected goals NYI 30.3 LA 1.013 TOR 30.0 to average (a shot quality BOS 30.5 OTT 1.016 CHI 30.1 factor of 0.950 means that PHO 30.7 COL 1.017 PIT 30.1 shots taken are, on average, PIT 30.8 TOR 1.024 OTT 30.4 5% less likely to result in EDM 31.4 NYR 1.027 MON 31.2 goals). MON 31.6 EDM 1.038 PHI 31.7 PHI 31.8 CHI 1.053 LA 32.4 There is a clear recording LA 32.0 DAL 1.053 EDM 32.6 bias3 in certain arenas FLA 33.6 ANA 1.057 ATL 33.0 rendering home shot ATL 33.9 STL 1.071 FLA 33.9 quality factors meaningless. The SQ factors indicated in the table are road factors. The use of road factors eliminates most of this reporting bias.

The worst reporting bias is in . The data there wrongly suggests that the Rangers give up shots of high quality (danger). The use of road factors reveal the truth – the Rangers are, in fact, a below average (defensive) shot quality team. The true

2 My approach to Shot Quality is described in http://www.HockeyAnalytics.com/Research_files/Shot_Quality.pdf

3 See http://www.HockeyAnalytics.com/Research_files/Product_Recall_for_Shot_Quality.pdf

Copyright Alan Ryder, 2008 Hockey Analytics www.HockeyAnalytics.com 2008 NHL Review Page 10 laggards were St. Louis (repeat offender), Anaheim (the NHL’s most penalized team), Dallas and Chicago (another repeat offender). The Stars present an unusual profile. Across the NHL there is a positive correlation between (low) shot quality and (few) shots allowed. But not in Dallas.

The leaders in defensive shot quality were Columbus, Minnesota and . The latter two teams repeat near the top of the list (San Jose slipped a bit from their performance in 2007). Under the coaching of Columbus moved up from 6th to 1st overall in shot quality. Thanks to the Capitals moved up from 8th to 4th.

Defense

If you multiply shots allowed and shot quality you get Marginal Goals - Defense a “defensive index”. This is a Team MGD MGDEH MGDPP MGDSHK MGDSHO comprehensive, SJ 147 96 3 38 10 simple assessment DET 147 107 5 36 -1 of team defense (see CBJ 132 92 7 40 -7 table above). Note WAS 120 86 4 29 1 that, as the range of NYR 117 80 5 31 2 shot quality results CAL 117 89 4 34 -9 is much smaller than NJD 112 77 4 24 7 the range of shots MIN 111 71 2 34 5 allowed, the latter is VAN 110 80 4 30 -4 the more dominant DAL 107 70 3 35 -1 factor in my BUF 104 63 4 32 4 defensive index. COL 102 64 6 21 10 Using this TB 100 45 6 36 13 measurement the top CAR 95 74 0 22 -1 defensive teams in ANA 93 73 3 25 -9 the NHL in 2008 NAS 92 55 2 32 3 were San Jose, BOS 91 68 4 15 4 Detroit (repeating as STL 90 59 0 33 -2 the top two PHO 88 64 0 20 4 defensive teams) NYI 86 64 -2 28 -5 and Columbus. PIT 84 64 2 19 -1 TOR 84 63 2 19 0 The worst defensive CHI 83 58 4 27 -6 teams were Florida, OTT 81 56 5 25 -6 Atlanta (again), MON 75 48 5 20 1 Edmonton and Los PHI 72 49 4 25 -6 LA 61 35 6 13 8 Angeles. When the EDM 61 35 2 27 -2 Panthers latch on to ATL 59 41 2 17 -1 a very reliable FLA 54 37 4 16 -3 goaltender (first

Copyright Alan Ryder, 2008 Hockey Analytics www.HockeyAnalytics.com 2008 NHL Review Page 11

Luongo and now Vokoun) they have shown a clear tendency to abandon defensive play. In hockey there is a -off between offense and defense. The Florida situation tells us clearly that a top-notch netminder can indirectly improve offense.

Marginal goals provides a slightly more sophisticated look at this (the enhancement being largely a reflection of the on-ice situation) with nearly identical overall results. Above is a summary of marginal goals from defense (MGD) by situation – even handed (EH), power play (PP), short handed opportunities or penalty taking (SHO) and penalty killing (SHK).

Proving that playoff teams don’t need to play defense were Philadelphia (MGD of 72), Montreal (75), Ottawa (81) and Pittsburgh (84). On defense the most similar team to Pittsburgh was Toronto, a frightening thought for Penguin fans.

San Jose (MGD of 147), Detroit (147) and Columbus (132) led the way on defense. The Red Wings and Sharks ranked 1st and 2nd in shots allowed. Columbus was the leader in shot quality (Detroit was near average and San Jose ranked 6th)

As was the case in 2007 Detroit was the most effective defensive team while even handed (MGDEH of 107). The Wings and Sharks had pretty similar MGD totals on the penalty kill. San Jose made up the even handed defensive shortfall by taking few penalties (one of the league’s best in each of the last three seasons). Tampa Bay (MGDSHO of 13) had the third lowest total in short handed opportunities. They lead in MGDSHO because of context – this performance was more valuable as it was in from of very poor goaltending. This was worth about 5 points in the standings.

Calgary and Anaheim were the NHL’s least disciplined teams, giving up around four points by taking penalties. In Anaheim’s case the damage was mitigated by stellar goaltending.

The MGDSHK leaders were Columbus (40) and San Jose (38). Tampa Bay allowed the fourth fewest number of short handed opportunities and tied for third with Detroit with 36 MGDSHK. Their penalty avoidance and solid penalty killing was seen by my marginal goals analysis as awfully good news given their poor goaltending.

Generally a team ends up with low MGDSHK scores because of a poor penalty kill. That was the case in Los Angeles, Boston and Atlanta. The Kings’ league worst score was enhanced by taking few penalties. Florida’s weak penalty killing was hidden by great goaltending. This analysis sorts that out for you.

Goaltending

Isolating shots and shot quality lets one better assess goaltending. The impact of goaltending is highest when a strong goalie allows ‘few’ goals notwithstanding a high number of shots faced and / or shots of high quality.

Below is a table of the marginal goals from goaltending (MGG) by team (excluding the shootout which is discussed below).

Copyright Alan Ryder, 2008 Hockey Analytics www.HockeyAnalytics.com 2008 NHL Review Page 12

Although it is certainly possible, you won’t normally find impactful goaltending behind a great defense. It just does not get the opportunity to shine. So it is not so surprising to see the goaltending of San Jose, Detroit and Columbus in the bottom half of this list. This does not mean that goaltending for these teams is necessarily weak, just that it did not contribute much to overall team success.

In 2007 Nashville was on top by some distance. This season it is Florida (by a nose). The common thread was Tomas Vokoun. In Nashville he played behind a weak defense. In Florida he suffered through the NHL’s worst defense.

Anaheim was close behind Florida and improved MGG from 2007. There was no change in goal here, just improved play from J-S Giguerre.

A marginal goals analysis compares performance to a threshold. The threshold is based on my estimate of Marginal Goals - zero-value performance. You can see from the table Goaltending below that the threshold was cleared by every team, save one. Tampa’s goalies collectively did not make Team 2008 2007 the grade. Tampa Bay finished last in the East FLA 79 27 because of a complete lack of goaltending. The latest ANA 76 59 savior is Mike Smith, acquired at the trade deadline in MON 62 51 the deal. PIT 57 47 PHI 54 13 There was also an unimpressive goaltending NJD 48 60 contribution in Washington, Columbus, Carolina, BOS 47 31 Calgary, Toronto, Buffalo and San Jose. NYR 46 45 EDM 45 29 The Sharks traded Vesa Toskala to the Leafs and DAL 42 44 upped the workload of Evgeni Nabakov. The result PHO 40 -15 was a small improvement in Toronto and a small CHI 39 38 deterioration in San Jose. NAS 37 88 VAN 37 71 It slays me that Nabakov was a COL 36 32 finalist. What got him there was a league leading 46 MIN 32 65 STL 31 41 wins. Of course he got a lot of wins. He played a lot OTT 30 65 (77 games) for an outstanding team (second best in LA 29 15 the NHL). Goaltender wins is the most useless ATL 28 68 statistic in the NHL. DET 28 40 NYI 27 69 Calgary suffers from goaltending delusions. The SJ 19 33 Flames remember Kiprusoff as he was. His goals BUF 17 34 against average (the second most useless statistic in TOR 13 7 hockey) is trimmed by good defensive play. His play CAL 12 42 has been going downhill for a couple of years now. CAR 12 15 CBJ 11 10 Toronto has joined the delusional club. I said at the WAS 7 27 beginning of the season that Toskala was a small TB -13 7 upgrade in goal and he was. But the team seems to

Copyright Alan Ryder, 2008 Hockey Analytics www.HockeyAnalytics.com 2008 NHL Review Page 13

be giddy based on only the direction of the change. This team should have, could have obtained Vokoun. If they had done so there would be less unemployment in Toronto.

Let’s look at the biggest swings in goaltending:

• Phoenix (+65 MGG) had a revolving door of terrible goaltending in 2007 and then hit the jackpot with the early season acquisition of (.921 save percentage in 55 games).

• In Florida (+52) and Nashville (-51) it was the Vokoun story. Nashville, needing to dump salary, placed a bet on and lost (.898 save percentage). A total calamity was avoided by the surprising play of Dan Ellis (.924) and some improvement in defensive play.

• Philadelphia (+41) found a keeper in and made the playoffs. The Sabres (-17) saw him as their number two goalie, let him go and missed the playoffs. Biron (.918) played much better than (.906) in 2008.

• Boston (+16) benefited from the rise of unheralded (.921). There is a rich history in the NHL of late blooming goalies. But expect his performance to regress next season.

• The Oilers (+16) acquired (.913) as a backup for aging veteran (.901) and he slowly assumed the starter’s job. And what a contribution in the shootout (see below)!

• Just when you thought that the Lightning (-20) had no goaltending, they showed up with even less.

• In Calgary (-30) Kiprusoff has gone from a save percentage of .923 in 2006 to .917 in 2007 to .906 in 2008.

• The story in Minnesota (-33) was that Niklas Backstrom went from stunning (.929) in 2007 to simply stellar (.920) in 2008.

• Vancouver (-34) acquired for the 2007 season, the Canucks MGG improved by 31 goals and the team went from playoff spectators to division winners. In 2008 Vancouver’s MGG regressed by 34 goals. Luongo had an “off” season (save percentage of ‘only’ .917 versus .921 the prior year), but backup goaltending was weaker (MGG -5) and team defense also improved (better defense means less impact from goaltending).

• What a soap opera in Ottawa (-35). was acquired to succeed Domenic Hasek who was acquired to lead the Senators out of goaltending purgatory to the Promised Land. The problem with Gerber was that he got off to a rough start. surprised and won the starting job. That’s the history. This season opened with the Senators seeking to trade Gerber but an injury to

Copyright Alan Ryder, 2008 Hockey Analytics www.HockeyAnalytics.com 2008 NHL Review Page 14

Emery made Gerber the guy in the early season. He was white hot so Sugar Ray sat for a while and began to pout. He got his chance as Gerber cooled considerably, but his performance was not compelling and he returned to the bench (and to pouting). Overall Gerber’s performance was OK (.910 save percentage) but it was not up to the Emery standard from the previous season (.918).

• In Atlanta (-40) the number one guy, , played better in 2008 (save percentage of .916 versus .912 the prior year). But he played less – 48 games versus 68 games. Johan Hedburg’s performance (.892) as the backup was marginal.

• Rick Dipietro’s work for the NY Islanders (-42) was way off in 2008 (save percentage of .902 versus .919 the prior year) – not good enough for his over-the- top contract.

The Shootout

My method for assessing shootout performance for goaltenders is the same Shootout Statistics as for skating time. To get marginal shootout goals saved (MGGSO) I Statistic 2006 2007 2008 compare save percentages to a threshold Shootouts 145 164 156 and then multiply the difference by the Attempts 981 1215 1057 number of attempts faced. For skaters I Goals 330 398 344 use the same kind of logic to derive Attempts per Shootout 6.77 7.41 6.78 MGOSO4. Goals per Shootout 2.28 2.43 2.21 Shooting Percentage .336 .328 .325 Shooting Threshold .131 .141 .202 The history of the shootout is brief, but a Save Threshold .533 .531 .472 trend line is beginning to emerge – Goalie Attribution .390 .431 .621 goaltenders are getting smarter faster than shooters. To the right is some data from the shootout from its inception.

You can see that the number of shootouts has been relatively stable since 2006 (a function of what is going on during the previous 65 minutes of ‘skating time’).

The number of attempts and goals per shootout tends to reflect the number of shootout marathons, which was down in 2008. These tend to favour the goaltender as the rules prohibit shooters from multiple attempts.

There has been a downward trend in shooting percentage, indicating that goalies are winning the battle.

4 For a full description of my method see http://www.HockeyAnalytics.com/Research_files/Shootout_at_the_Oval_Corral.pdf

Copyright Alan Ryder, 2008 Hockey Analytics www.HockeyAnalytics.com 2008 NHL Review Page 15

The shooting and save thresholds I have used in my calculations are also shown. They were Marginal Goals - Shootout very stable from 2006 to 2007 but changed dramatically in 2008. The driver of this is Team SW MGOSO MGGSO the final statistic – ‘goalie attribution’. In ANA 8 6 8 2006 I attributed 39% of shootout results to ATL 9 7 14 goaltending. In 2007 I increased this slightly BOS 6 1 8 BUF 4 3 6 to 43%. But in 2008 I attributed 62% of the CAL 3 3 3 shootout to goalies. CAR 2 4 -1 CBJ 3 4 3 Goalie attribution is a measure of the relative CHI 5 5 6 team-to-team variation in shooting COL 7 8 10 percentages and save percentages. In the DAL 5 4 6 extreme, no team-to-team variation in save DET 5 6 4 EDM 15 11 24 percentages would imply that shootout FLA 5 1 11 success was determined 100% by the LA 5 6 4 shooters. Likewise no team-to-team MIN 3 3 0 variation in shooting percentages would MON 5 2 10 imply that shootout success was determined NAS 3 2 3 100% by the goaltenders. This year there NJD 8 7 12 was much less team-to-team variation in NYI 5 5 8 NYR 8 4 12 shooting percentages but about the same OTT 3 4 4 team-to-team variation in save percentages. PHI 3 5 -1 PHO 5 4 10 This means that goaltending became PIT 7 8 10 relatively more valuable in the shootout and SJ 6 10 3 my Player Contribution results reflect that. STL 3 0 8 TB 2 2 2 To the right is a summary, by team, of the TOR 3 3 4 VAN 6 2 14 shootout in 2008 – wins and marginal goals WAS 4 2 10 (offense/goaltending).

In 2008 nearly one Edmonton game in four was decided by a shootout. And this was good for the Oilers as they went 15-4. So far as I can tell the shootout is a lottery so this was one fortunate team. This success was largely (about 70%) attributed to goaltending (24 of 35 marginal goals). No other team cleared 10 shootout wins. Atlanta had the second best win total (9) based on 21 marginal goals.

At the other end of the spectrum was Tampa Bay with just 2 shootout wins (on 4 marginal goals). This is more proof of the lottery nature of the shootout – Tampa Bay had the best shootout performance in 2007 (10 wins on 29 marginal goals). Two other leaders in 2007 faired poorly – Minnesota went from 10 wins on 24 marginal goals to 3 lucky wins with only 3 marginal goals and Buffalo went from 10 wins on 23 marginal goals to 4 wins on 9 marginal goals.

Copyright Alan Ryder, 2008 Hockey Analytics www.HockeyAnalytics.com 2008 NHL Review Page 16

Individual Performances

Hart Trophy

The is awarded to the player judged to be “the most valuable to his team”. The Lester B Pearson Award is awarded to the player judged, by his peers, to be “the most outstanding player”. This year’s “finalists” (the top three vote getters) for both trophies were , and .

Although a literal read of this clearly means that a goaltender ought to win this prize each year, the award has typically (nearly 90% of the time) been presented to the NHL’s most impactful skater, as judged by the voters. In fact the Hart Trophy has usually gone to a forward (about 80% of the time). In their voting for the Pearson Award the players have shown an even greater bias towards forwards than do the hockey writers who choose the Hart winner. I will leave this debate alone and go straight to a discussion of my …

Top Forward

Alexander Ovechkin laid claim to an unprecedented and highly deserved Quad Award Crown – the Trophy, the Art Top Forward Ross Trophy, the Hart Trophy and the Pearson Award. He also gets my Wayne Gretzky Award Player Team PC for the top contribution by a forward to his Alexander Ovechkin WAS 162 team’s success. Jarome Iginla CAL 141 Pavel Datsyuk DET 132 Richard’s hardware is earned by the league’s top ATL 124 goal scorer. Ovechkin won this prize with 65 Henrik Zetterberg DET 121 goals (his second 50 goal season in his three SJ 115 years of professional hockey).

The Trophy is won by the player with the most scoring points. As an average of about 1.7 assists is awarded for each goal scored, it is more common for this trophy to be claimed by a top playmaker. Only about 40% of Art Ross Trophies have been claimed by the league’s top goal scorer.

It is not so uncommon for the Art Ross and Hart trophies to be captured by the same player. In fact, as 40 of 84 Hart Trophies have been won by the NHL’s leading collector of scoring points, this is the surest route to the MVP award. Demonstrating that assists are overrated, only 21 Hart Trophies have been won by the NHL’s leading goal scorer.

The short history of the Richard Trophy (first awarded in 1999) is what makes a Triple or Quad Crown unprecedented. In 18 seasons since 1923-24, when it was first presented, the Hart Trophy has gone to a player that was both the top goal and point scorer – Wayne Gretzky (5 times), (4 times), (2 times), ,

Copyright Alan Ryder, 2008 Hockey Analytics www.HockeyAnalytics.com 2008 NHL Review Page 17

Howie Morenz, Jean Beliveau, , , and now Ovechkin are the only players to perform this magic. However, in 16 other seasons the Double Crown winner has been denied (in 7 seasons by another forward).

Here is why Ovechkin’s name belonged on the Hart and Pearson awards (given the voters’ current preference to largely ignore defensemen and goalies).

First of all, did not simply win the Richard Trophy. He smoked the competition with a season for the ages.

Maurice Richard set the goal standard with 50 goals in 50 games in the 1944-45 season. The count of 50 goal seasons now stands at 183. To properly compare these results across over 60 years of the changing face of hockey it is necessary to adjust for (at least) two things. The first is the increasing length of the season, which has ranged from 50 games in Richard’s seminal season to as many as 84 games. A second factor is the ebb and flow of scoring in the NHL. Since the end of WW2 the average number of goals scored per game has been as high as 8.02 (in 1981-82) and as low as 4.79 (in 1952-53).

If we “normalize” the history of 50 goal seasons to today’s context (an 82 game, no injuries season and average scoring of 5.44 goals per game) we can make a better comparison of the performances. When you translate these performances in this fashion about two thirds of all 50 goal seasons go away. Those that failed to make the cut are all from the wide open era from the mid-1970s through the mid-1990s. The top 18 seasons (60 goals or more in today’s context) are shown in the table below:

Goals League Avg Normalized Player Team Season GP Scored Goals per Game Goals 1 STL 1990–91 78 86 6.82 72 2 Mario Lemieux PIT 1992–93 60 69 7.18 72 3 BOS 1993–94 49 50 6.40 71 4 Mario Lemieux PIT 1995–96 70 69 6.20 71 5 Phil Esposito BOS 1970–71 78 76 6.24 70 6 Mario Lemieux PIT 1988–89 76 85 7.39 68 7 Wayne Gretzky EDM 1983–84 74 87 7.80 68 8 FLO 1999–00 74 58 5.41 65 9 Alexander Ovechkin WAS 2007-08 82 65 5.44 65 10 Wayne Gretzky EDM 1981–82 80 92 8.02 64 11 Phil Esposito BOS 1971–72 76 66 6.13 63 12 Brett Hull STL 1991–92 73 70 6.87 63 13 Bobby Hull CHI 1968–69 74 58 5.69 62 14 BUF 1992–93 77 76 7.18 62 15 Teemu Selanne ANA 1997–98 73 52 5.20 61 16 Bobby Hull CHI 1965–66 65 54 6.08 61 17 Phil Esposito BOS 1973–74 78 68 6.39 61 18 Maurice Richard MON 1944–45 50 50 7.35 61

Copyright Alan Ryder, 2008 Hockey Analytics www.HockeyAnalytics.com 2008 NHL Review Page 18

And there is Ovechkin’s 2008 season ranked just ahead of Wayne Gretzky’s record rattling season of 92 goals (set in the highest scoring season since the Second World War) – rather elite company. This is a ‘wow’ performance however it is easily underrated because of the low scoring nature of today’s game.

But was he, as the Hart Trophy begs, “valuable to his team”?

The story here is actually more compelling. The scored 238 goals in 2008. Ovechkin accounted for about 27% of this total (and assisted on another 20% of Washington goals). Imagine this team without that contribution.

To put his participation in scoring in context, only two players have scored a greater percentage of their team’s goals since the 1930s. In 2001 Pavel Bure accounted for nearly 30% of Florida’s offense. Ten years earlier, Brett Hull accounted for 28% of the offense of the St. Louis Blues. This is a “most valuable player to his team” performance.

The lens of Player Contribution enables us to break down this performance and confirm that he was, indeed, the NHL’s most valuable skater. Shown on the next page are the details of the PC calculation for the top 30 forwards.

Ovechkin was quite solid defensively with a PCD score of 24 (with only 13 minutes of penalty killing) including a relatively impressive PCDEH of 13. He can be compared to Ilya Kovalchuk, another 50 goal scorer who does not play defense at all (PCDEH of 1).

Of course it is on offense that Ovechkin excelled. On the power play (PCOPP of 36) he ranked second to (43). But he was seriously dangerous while even handed – his PCOEH score of 95 was out of this world. Interestingly he did nothing in the shootout (PCOSO of -1).

New to PC this year is PCOPPO. Actually, it is not quite new. I have always had a way of attributing PC points to penalty drawing but it was lame because if a lack of data. This season the NHL finally started to report the player drawing a penalty and PCOPPO can now be calculated in a manner that is completely consistent with PCOSHO (for penalty taking). Ovechkin drew a lot of penalties and ended up with a PCOPPO score of 8.

Nobody was close to Ovechkin overall or on offense. Jarome Iginla had a season that would have earned the MVP in another year with 119 PCO points and solid defense (PCD 22). Kovalchuk, the NHL’s other 50 goal scorer, collect 114 PCO points and, other than avoiding penalties (PCOSHO of 7) played no defense. Evgeni Malkin would have matched Kovalchuk overall if not for taking more penalties (PCOSHO of -3). You can see that several players made a better case for the Hart Trophy than did Malkin.

Exceptional defensive performance propelled Pavel Datsyuk (PCD of 39), Joe Thornton (30), (32), Martin St. Louis (33) and (33) up the leaderboard. But these players did it in different ways. Thornton had a PCDEH score of 18 whereas St. Louis’ even handed defense was actually awful (PCDEH -1). He added value on the penalty kill (PCDSHK 13) and in penalty avoidance (PCDSHO 21).

Copyright Alan Ryder, 2008 Hockey Analytics www.HockeyAnalytics.com 2008 NHL Review Page 19

2008 Player Contribution – Forwards (items may not total due to rounding)

PCO PCD Player Team POS PC EH PPP PPO SH SO PCO EH PP SHK SHO PCD Alexander Ovechkin WAS LW 162 95 36 8 0 -1 138 13 1 1 9 24 Jarome Iginla CAL RW 141 78 29 3 0 9 119 16 0 1 5 22 Pavel Datsyuk DET C 132 48 28 5 1 10 93 16 0 8 15 39 Ilya Kovalchuk ATL LW 124 75 26 -1 5 9 114 1 1 2 7 10 Henrik Zetterberg DET LW 121 54 31 3 1 5 95 14 0 4 8 26 Joe Thornton SJ C 115 52 26 3 -1 4 85 18 0 5 7 30 Evgeni Malkin PIT C 113 68 34 9 -1 -4 106 8 0 3 -3 7 OTT RW 109 56 17 -2 18 2 91 3 1 7 6 17 Marian Gaborik MIN RW 106 70 21 5 1 0 98 10 0 2 -4 8 CAR C 102 42 31 9 -2 3 84 9 0 6 4 18 OTT C 99 57 25 -4 0 10 88 7 1 1 2 11 Jason Pominville BUF RW 94 45 11 -2 2 5 62 9 1 9 13 32 STL RW 94 56 13 -3 0 10 75 8 -1 1 10 18 TB C 93 61 19 -1 4 2 86 -3 1 13 -3 8 CBJ LW 92 43 16 6 8 9 81 12 0 13 -14 11 Derek Roy BUF C 92 46 12 7 7 0 73 3 1 12 3 19 NJD LW 90 40 18 2 2 8 69 10 0 1 9 21 PHI C 87 33 20 5 12 6 74 6 0 4 2 13 Alexei Kovalev MON RW 86 27 43 5 1 -2 75 8 0 6 -4 11 Sidney Crosby PIT C 85 45 18 12 1 2 78 5 0 1 0 7 Nikolai Zherdev CBJ RW 85 30 13 3 0 15 61 13 0 1 9 24 Martin St. Louis TB RW 85 32 21 -2 4 -3 52 -1 0 13 21 33 Daymond Langkow CAL C 83 32 19 -6 4 0 50 13 1 3 16 33 Anze Kopitar LA C 83 28 24 4 6 7 69 2 1 1 10 14 CAL LW 83 40 14 -4 0 14 64 12 1 3 4 19 SJ C 82 13 16 2 2 22 55 13 0 9 5 27 PHO RW 82 32 19 2 11 1 65 5 0 5 7 17 Mats Sundin TOR C 80 50 22 2 2 8 84 9 1 -6 -7 -3 Brenden Morrow DAL LW 80 34 20 7 3 4 68 11 0 5 -5 12 Dustin Brown LA RW 80 28 16 11 5 11 72 3 1 1 3 8

The shootout shuffled the order of this list. Joe Pavelski was a shootout machine, scoring 7 times in 11 attempts and collecting 22 PCOSO. Nikolai Zherdev and Kristian Huselius also made big moves up in the rankings because of their shootout performance. Dustin Brown made the leaderboard because of his performance in the shootout and in drawing penalties – stuff that gets ignored. Most people have no way of factoring these things into player evaluation. Now you do.

Copyright Alan Ryder, 2008 Hockey Analytics www.HockeyAnalytics.com 2008 NHL Review Page 20

Pavel Datsyuk might be the NHL’s most Top Forwards – 2003-08 under-rated player. Shown to the right are the top seven aggregate PC scores from Player Team PC forwards over the past five seasons … and Pavel Datsyuk DET 429 there is Datsyuk on top of the list. Each of Joe Thornton BOS/SJ 423 Thornton, Iginla, St. Louis and Ovechkin Jarome Iginla CAL 420 have Hart Trophies to brag about. Last Martin St. Louis TB 413 season Lecavalier should have had one. Vincent Lecavalier TB 398 Note that Ovechkin makes this list based Marian Hossa OT/AT/PIT 398 only three seasons in the NHL. Alexander Ovechkin WAS 398

Top Defensive Forward

The Frank Selke trophy goes to the best defensive forward in the NHL. Having no Bob Gainey Award metric for defense on which to rely, the Top Defensive Forward voters rarely get this right. Reputations tend to rule. The three finalists for this Player Team PCD trophy were Pavel Datsyuk (the winner), Pavel Datsyuk DET 39 and Henrik Zetterberg. SJ 37 Daymond Langkow CAL 33 My Bob Gainey Award winner, for the top Martin St. Louis TB 33 defensive contribution by a forward, was Jason Pominville BUF 32 also Pavel Datsyuk, who put up very Boyd Gordon WAS 31 impressive PC numbers for the second year in a row. Both Madden and Zetterberg had strong defensive seasons but many others deserved greater consideration in the Selke voting. On the following page are the details for the PCD calculation for the top 20 defensive forwards in 2008.

Let’s compare the performance of Datsyuk to his peers, focusing on Zetterberg and Madden. PC had his even handed defense ahead of teammate Zetterberg (PCDEH 14) and former Selke winner Madden (10). His PCDEH of 16 trailed only San Jose’s Joe Thornton (18) and Milan Michalek (17) and was tied with Jarome Iginla, Viktor Kozlov and teammates Valtteri Filppula and Mikael Samuelsson.

Even handed goals against averages tell part of the story – Thornton (1.88), Michalek (1.76), Datsyuk (1.88), Iginla (2.29), Kozlov (2.03), Filppula (1.60) Samuelsson (1.34), Zetterberg (1.98) and Madden (2.35). But two other factors need to be considered to calculate PCDEH – ice time and goaltending. Madden, in particular, played in front of the best goaltending of this group of players. Given that, his GAA is unimpressive. Datsyuk’s teammates obviously had the same goaltending and can be compared directly. Filppula and Samuelsson had better defensive numbers per minute on ice, but had fewer of those minutes. Datsyuk’s even handed performance can be compared directly to that of Zetterberg – more ice time with a slightly better GAAEH.

Copyright Alan Ryder, 2008 Hockey Analytics www.HockeyAnalytics.com 2008 NHL Review Page 21

The top penalty killing performance by a forward 2008 Player Contribution in 2008 was by the PCD for Forwards Senators’ SHK (PCD score of 16). He Defense (PCD) was on the ice for 23 goals Player Team POS EH PP SHK SHO PCD against in 284 minutes of Pavel Datsyuk DET C 16 0 8 15 39 penalty killing (a short Mike Grier SJ RW 10 0 14 12 37 handed GAA of 4.86) in Daymond Langkow CAL C 13 1 3 16 33 Martin St. Louis TB RW -1 0 13 21 33 front of average Jason Pominville BUF RW 9 1 9 13 32 goaltending. Other top Boyd Gordon WAS C 13 0 8 10 31 short handed defensive Joe Thornton SJ C 18 0 5 7 30 performances (both for 15 Ryan Johnson STL C 9 0 15 6 30 SHK PCD points) were from Viktor Kozlov WAS C 16 0 0 13 29 Ryan Johnson (28 goals Alexander Steen TOR LW 9 0 13 6 28 against, 300 minutes, P.J. Axelsson BOS LW 10 0 7 11 28 GAASHK 5.61), and Dave Chris Kelly OTT C 8 0 16 4 28 Steckel (24, 253, 5.69). Valtteri Filppula DET C 16 0 5 6 27 Joe Pavelski SJ C 13 0 9 5 27 From the leaderboard Patrick Rissmiller SJ LW 10 0 14 2 26 Dave Steckel WAS C 9 0 15 2 26 several players had notably Mathieu Darche TB LW 6 0 12 8 26 better goals against John Madden NJD C 10 0 9 8 26 averages. San Jose’s Joe NYR C 8 0 14 4 26 Pavelski and Joe Thornton Henrik Zetterberg DET LW 14 0 4 8 26 had short handed GAAs of 2.14 and 2.50 respectively. But these numbers were not statistically credible (3 goals in 84 minutes in the case of Pavelski and 2 goals in 48 minutes in the case of Thornton). In Tampa Bay Mathieu Darche had a performance much like that of Pavelski (3 goals, 86 minutes, 2.09). But he earned more PC points (12 versus 9) because that record came in front of very poor goaltending. Finally Daymond Langkow had a short handed GAA of 2.40 based on 1 goals in just 25 minutes of penalty killing.

Of the three Selke trophy finalists Datsyuk (14 goals, 148 minutes, GAA 5.69, 8 PCDSHK) had the best performance and Madden (33 goals, 287 minutes, GAA 6.89, 9 PCDSHK) the most impact because of more playing time. Madden played in front of better goaltending. Zetterberg (22 goals, 162 minutes, GAA 8.14, 4 PCDSHK) can be compared directly to Datsyuk.

To top this list you generally need have good penalty avoidance. And this brings me to Martin St. Louis. He has, for years, played very uninspired even-handed defense, killed penalties a great deal and with good effect and avoided the sin-bin like a choir boy. This rather odd profile has placed him high on my defensive play leader board for several years in a row.

But getting any defensive job done while avoiding penalties is much better than getting the job done while taking penalties. Vincent Lecavlier looked a fair bit like St. Louis

Copyright Alan Ryder, 2008 Hockey Analytics www.HockeyAnalytics.com 2008 NHL Review Page 22

(PCDEH of -3 and PCDSHK of 13) until you get to his penalty taking (32 minors versus 13). PC has this 19 penalty gap figured as a 2.4 point (24 PC points) difference, fueled partially by Tampa’s weak goaltending. Rick Nash looks like a defensive gem (PCDEH of 11 and PCDSHK of 13) until you look at his penalty taking (45 minors). His PCDSHO score of -15 (one of the worst scores in the NHL) says that he cost Columbus 1.5 points (when compared to an average penalty taker).

Datsyuk ranked fourth amongst forwards in PCDSHO with a score of 15. The leader was Brad Richards (TB/DAL) with a score of 24. Madden and Zetterberg each had a PCDSHO score of 8.

PCD tries to sort out the impact of goaltending on defense, but these stats are still materially influenced by teammates. So we see some pairings from certain teams (notably Datsyuk and Zetterberg). And sometimes the best defense is a good offense. Spending a great deal of time on the attack is a great way to prevent goals (note some famous offensive players on the leaderboard, like Datsyuk and Zetterberg).

Another caveat is that not all forwards receive uniformly difficult defensive assignments. Top offensive forwards frequently skate against checkers and vice versa. This tendency may cause the defense of top offensive forwards to be over rated (and the defense of checkers to be underrated).

Top Defenseman

Finalists for the Norris Trophy in 2008: Zdeno Chara, Nicklas Lidstrom and Dion Award Phaneuf. Lidstrom was the winner in a Top Defenseman landslide – he was the top choice on 127 of 134 ballots. Player Team PC Nicklas Lidstrom DET 121 To the right I show the top six PC Brent Burns MIN 108 performances in 2008 by defensemen. You BUF/SJ 107 can see that, like Ovechkin, Lidstrom put a Brian Rafalski DET 105 great deal of distance between himself and Zdeno Chara BOS 97 the pack. On the following page I show the Mike Green WAS 97 PC breakdown for the top 20 defensemen which demonstrates how Lidstrom accomplished that feat.

Having now won 6 of the last 7 Norris Trophies, Nicklas Lidstrom has now emerged as one of the greatest defensemen of all time. Since I started doing Player Contribution I had him ranked as the league’s most valuable skater and top defenseman in 2003 (PC 124) and fifth amongst defenders in 2004 (PC 93). In 2006, with his best PC score ever (143), I had him ranked as the NHL’s second most valuable skater but only because of ’s unusual shootout contribution. Last season I had Lidstrom ranked at the top of my list of defenders (PC 130) and, in 2008, I have him as the top defenseman in the NHL (PC 121) and winner of my Bobby Orr Award for the top contribution by a defenseman.

Copyright Alan Ryder, 2008 Hockey Analytics www.HockeyAnalytics.com 2008 NHL Review Page 23

This makes him, by far, the NHL’s most valuable skater over the past five seasons. Top Defenseman – 2003-08 To the right is a list of the top seven defensemen over that period. Notice the Player Team PC Detroit, New Jersey, Anaheim triangle in Nicklas Lidstrom DET 487 this list. Sergei Zubov DAL 412 SL/ED/AN 397 Offensively, in 2008, Lidstrom’s 43 PCO Tomas Kaberle TOR 382 points ranked him fifth among defensemen. Brian Rafalski NJ/DET 359 The PCO leaders were (60), DET/ANA 340 Zdeno Chara (55), Mike Green (53) and Scott Niedermeyer NJ/ANA 338 Brent Burns (51). As he is below average at drawing penalties (he was named the victim in a minor penalty call only 4 times), a realistic measure of PCOPPO hurt Lidstrom. The highest PCOPPO score (13) in the NHL (based on 53 minor penalty draws) belonged to Phaneuf. Brent Burns went 3 for 5 in the shootout, worth 16 PCOSO points, to move onto the list of the top five offensive defensemen.

2008 Player Contribution – Defensemen (items may not total due to rounding)

PCO PCD Player Team POS PC EH PPP PPO SH SO PCO EH PP SHK SHO PCD Nicklas Lidstrom DET D 121 25 23 -5 0 0 43 48 3 17 10 78 Brent Burns MIN D 108 20 17 -2 0 16 51 35 3 18 1 57 Brian Campbell BUF/SJ D 107 19 23 -2 -1 0 39 26 5 18 19 68 Brian Rafalski DET D 105 14 24 -2 0 0 36 45 6 9 9 69 Zdeno Chara BOS D 97 22 21 5 3 4 55 28 5 16 -6 42 Mike Green WAS D 97 31 17 7 0 -1 53 38 4 3 -1 44 Dion Phaneuf CAL D 93 21 26 13 2 -3 60 39 4 12 -22 33 Tomas Kaberle TOR D 93 12 19 -5 0 10 35 33 4 0 20 58 Adrian Aucoin CAL D 88 12 11 0 0 0 23 47 4 4 10 65 Andrei Markov MON D 83 17 23 2 4 3 49 14 5 10 5 34 BOS D 83 9 17 4 -1 -1 28 35 9 13 -2 55 CHI D 82 21 1 -3 4 0 23 37 3 13 7 59 PIT D 81 12 34 -4 -1 -1 41 27 2 12 0 41 OTT/CAR D 81 19 21 2 0 0 42 25 6 0 8 39 Ron Hainsey CBJ D 79 -2 17 -4 1 0 12 37 10 3 17 67 Marc-Edouard Vlasic SJ D 78 -2 3 -1 -1 0 -1 47 3 14 16 80 Daniel Girardi NYR D 78 11 7 2 0 0 19 26 3 16 14 59 Mathieu Schneider ANA D 77 23 10 -2 0 8 38 32 1 5 0 39 Paul Martin NJD D 76 10 5 -1 1 0 15 39 3 7 12 61 Rostislav Klesla CBJ D 76 3 2 -2 -1 0 2 49 6 16 3 73

Copyright Alan Ryder, 2008 Hockey Analytics www.HockeyAnalytics.com 2008 NHL Review Page 24

Defensively he ranked number two (PCD of 78) behind Marc-Edouard Vlasic. His 48 PCDEH points were third, trailing only Rotislav Klesla and . On the penalty kill his PCDSHK score (17) was third (behind Burns and Brian Campbell). And he did this without taking penalties (PCDSHO of 10).

Player Contribution can be reduced to ‘performance’ times ‘ice time’. Lidstrom logged the ice time – 2031 minutes on ice, eighth among defensemen Phaneuf (2167), Duncan Keith (2096), Campbell (2084), Chara (2067), Andrei Markov (2048), Kaberle (2039) and Denis Wideman (2038).

Three other defenders broke 100 PC points. Burns and Campbell led the league in PCDSHK. Burns had an unusual shootout contribution and Campbell was second only to Kaberle in penalty avoidance. Rafalski had a very balanced profile. Phaneuf would have broken 100 if he had taken an average number of penalties. Chara had a much improved season but his Norris Trophy voting was influenced again more by his stature than his play.

Top Defensive Defenseman

The list of top defensemen is usually highly influenced by offensive play. The NHL Rod Langway Award needs a defenseman’s version of the Selke Top Defensive Defenseman Trophy. Mine is the Rod Langway Award, for the top defensive contribution to team Player Team PCD success by a defenseman. Marc-Edouard Vlasic SJ 80 Nicklas Lidstrom DET 78 Who were the best defensive defensemen Jan Hejda CBJ 73 in the NHL? Below I present a list of the Rostislav Klesla CBJ 73 top 20 defensive performances by Brian Rafalski DET 69 defensemen in 2008. It is an interesting list Brian Campbell BUF/SJ 68 with some big names and some no-names.

San Jose tied the Wings for the NHL’s best defense and four Sharks made the list. Marc- Edouard Vlasic moved from third in 2007 to first in 2008. Not only did he turn in the top defensive performance by a defenseman but he beat out Rotislav Klesla for the stay-at- home-defenseman of the year (highest differential between PCD and PCO). Vlasic had a PCO score of -1 whereas Klesla’s PCO was +2.

Four Columbus defenders made the top 10. The Blue Jackets had the NHL’s third best defense on the basis of sixth fewest shots and the top shot quality ranking. Hainsey repeated on this list from last season. Foote amassed all but 6 of his PCD points in Columbus.

There are also four Minnesota Wild defensemen on this list. The Wild were ranked eighth in team defense. These calculations suggest that it was the defensemen who made that happen. Detroit placed only two players on this list suggesting that the forwards were helping out more.

Copyright Alan Ryder, 2008 Hockey Analytics www.HockeyAnalytics.com 2008 NHL Review Page 25

2008 Player Contribution Defense – Defensemen (items may not total due to rounding)

PCD Player Team POS EH PP SHK SHO PCD Marc-Edouard Vlasic SJ D 47 3 14 16 80 Nicklas Lidstrom DET D 48 3 17 10 78 Jan Hejda CBJ D 50 1 22 1 73 Rostislav Klesla CBJ D 49 6 16 3 73 Brian Rafalski DET D 45 6 9 9 69 Brian Campbell BUF/SJ D 26 5 18 19 68 Ron Hainsey CBJ D 37 10 3 17 67 Adrian Aucoin CAL D 47 4 4 10 65 CBJ/COL D 42 0 28 -6 64 Nick Schultz MIN D 34 0 19 8 62 Paul Martin NJD D 40 3 7 12 61 Kim Johnsson MIN D 33 3 17 8 61 SJ D 47 3 17 -7 60 Duncan Keith CHI D 37 3 13 7 59 Daniel Girardi NYR D 26 3 16 14 59 Tomas Kaberle TOR D 33 4 0 20 58 Brent Burns MIN D 35 3 18 1 57 SJ D 39 2 15 0 56 Martin Skoula MIN D 27 1 17 11 56 CAL D 37 3 18 -3 55

The top even handed defenders were Hejda (50 PCD points, 1.82 GAA), Klesla (49, 1.94), Lidstrom (48, 1.71), Vlasic (47, 2.00), Aucoin (47, 1.65) and Ehrhoff (47, 1.81). I thought Aucoin was washed up. I was wrong.

There is such a thing as power play defense and Ron Hainsey showed how this is done. He was on ice for just 1 goal against in 349 power play minutes.

As I consider penalty avoidance to be an element of defense, it is not surprising that most of these players were above average in taking penalties. Kaberle led the way again with 20 PCDSHO points while Ehrhoff hurt himself the most taking penalties (-7 PCDSHO).

PCD is highly influenced by teammates. A good defensive team has lots of PCD to go around. Notably, given the way PCD is determined, we are unable to separate the work of a defensive pair. If two defenders play together at all times they will have the same PCD score (save for penalty taking). If they play together a great deal, they will have similar scores. It is always possible that one player carried the other but it is probable that a material difference in performance would result in less playing time for the inferior player.

Copyright Alan Ryder, 2008 Hockey Analytics www.HockeyAnalytics.com 2008 NHL Review Page 26

Top Goaltender

The Vezina Trophy goes to the goalkeeper adjudged to be “the best at this position as voted by the general managers of all NHL clubs”. I always struggle with this definition. If a goalie has great stats and plays 70 games, should he lose this award to another with better numbers but in only 45 games? I don’t think so. I prefer a “most valuable” definition (PC is an impact measure) for my Award.

This year’s finalists for the Vezina were , Evgeni Nabakov and . Brodeur won the trophy over Nabakov in a tight, two horse race (113 voting points to 106). The NHL GMs choose the Vezina winner whereas the professional hockey scribes vote for the MVP and all-star teams. Something really weird happened with the writers’ vote – Brodeur beat out Nabakov in the Hart Trophy voting (239 – 109) but Nabakov was selected to the first all-star team (go figure!).

Sit down for this. PC says that Tomas Vokoun was the NHL’s most valuable Patrick Roy Trophy goalie, yet he was not named on a single Top Goaltender Vezina or Hart ballot. Here is how he went unnoticed – although he started (and Player Team PC completed) 69 games and played 4031 Tomas Vokoun FLA 267 minutes (both very impressive numbers), Martin Brodeur NJD 259 he played for a crummy team and ‘his’ Jean-Sebastien Giguere ANA 232 record was 30-29-10. Vezina voters (the Tim Thomas BOS 200 GMs) love wins so this record did not cut Roberto Luongo VAN 199 it. His goals against average was an Henrik Lundqvist NYR 189 unimpressive 2.68. Did I mention that he played for a crummy team? Florida allowed more shots on goal than any team not based in Georgia. And the Panthers shot quality was such that I ranked them as the NHL’s worst defensive team.

To properly evaluate any goaltender one needs to look at save percentages. Vokoun’s shot quality neutral save percentage was an impressive .919.

To be clear, Martin Brodeur was a very close second and hardly a bad choice for the Vezina. Playing more (77 games, 4635 minutes) and for a better team, he had more wins (44). His goals against average was better (2.17) but, of course, he played behind better defense. His contribution in the shootout (PCGSO of 51) was a hair better than that of Vokoun (47). But his shot quality neutral (.918) save percentage was a bit behind that of Vokoun and that was the difference in the PC calculations.

Brodeur is famous for being an exceptional puck handler. To Player Contribution that skill looks like defense rather than goaltending. So the 8 point PC gap between Vokoun and Brodeur might be a bit overstated. This was a very close race.

A couple of other goaltenders are worth mentioning in the discussion of “best” performance by a goaltender. J-S Giguerre earned 232 PC points (based on 58 games,

Copyright Alan Ryder, 2008 Hockey Analytics www.HockeyAnalytics.com 2008 NHL Review Page 27

3310 minutes and a very impressive .927 shot quality neutral save percentage). And Dan Ellis (44, 2229, .922) spent a considerable amount of time winning the starting role. As they had higher save percentages, each of these goalies could have won my Patrick Roy Award with more playing time.

Roberto Luongo had an “off” season with “only” 199 PC points based principally on a shot quality neutral save percentage of .912 behind a solid but much injured blueline. But, over the past five seasons, he has been the NHL’s most valuable player by some distance.

Clean Play Awards

The Lady Byng Memorial Trophy goes to the player “adjudged to have exhibited the best type of sportsmanship and gentlemanly conduct combined with a high standard of playing ability”. Voters tend to go down the list of top scorers until they find someone with low penalty totals. This season Pavel Datsyuk (20), Jason Pominville (20) and Martin St. Louis (26) had the lowest penalty minute totals among the top 20 point scorers. They were the Lady Byng Trophy finalists. Dead simple.

Clean and Impactful Play

In honour of , a seven time Lady Byng Trophy winner (in fact they gave him the original trophy and now award a second generation version), let me present the Frank Boucher Award to the player who best combines both clean and impactful play. This is close to the Lady Byng definition. The word “combines” is an “AND” condition. With “AND” conditions you multiply (with “OR” conditions you add). You need both factors to be strong to get a good “AND” rating.

My “Frank” points are therefore Player Contribution (the measure of impact) x (50 – PIM) (my measure of clean play). Fifty minutes is a pretty arbitrary part of this formula. In fact the formula is pretty arbitrary.

For the third year in a row the voters chose Datsyuk as the Lady Byng man (132 PC points, 20 penalty minutes) and my approach concurred. The voters had Pominville (94, 20) as a well deserved finalist. Martin St. Louis was also in the finals. His numbers (85, 26) were strong but I had a number of players ranked ahead. Frank Boucher Award Gentlemanly Play and Ability As usual, the Lady Byng voters ignore defensmen in their vote. Kaberle (93, 22) Player Team Franks Pavel Datsyuk DET 3959 is an annual role model in clean, effective th Jason Pominville BUF 2817 play (he finished 30 in Lady Byng Brad Boyes STL 2811 voting). Daniel Girardi’s performance, a Daniel Girardi NYR 2790 PC score of 78 and 14 penalty minutes, also Radim Vrbata PHO 2753 impressed my formula (he finished tied for Tomas Kaberle TOR 2591 83rd in Lady Byng voting).

Copyright Alan Ryder, 2008 Hockey Analytics www.HockeyAnalytics.com 2008 NHL Review Page 28

Penalty Avoidance

To the right is a list of the six highest penalty avoidance Penalty Avoidance – Best contributions (PCDSHO) in the NHL in 2008. The formula to Player Team Minors PCDSHO determine this is basically ice Brad Richards TB/DAL 5 24 time x (threshold minor Martin St. Louis TB 13 21 penalties – actual minor Tomas Kaberle TOR 11 20 penalties) with a few Brian Campbell BUF/SJ 10 19 adjustments along the way, Ron Hainsey CBJ 10 17 most notably for the impact of Daymond Langkow CAL 7 16 goaltending. I have already remarked on many of these Penalty Avoidance – Worst players before so I won’t spend Player Team Minors PCDSHO more on them other to point out OTT 57 -28 that three are defensemen. PHO 52 -26 John Erskine WAS 38 -24 Also shown are the worst Dion Phaneuf CAL 61 -22 penalty taking offenders. Each CAL 50 -22 of these players played a great Shane O'Brien TB 42 -18 deal, took a lot of minor penalties and hurt their team to the tune of about 2 points in the standings. Neil and Carcillo are forwards. The rest are defensemen. As Carcillo is a tough guy (19 fighting majors), his minors might have had a number of offsetting penalties (I did not check).

Penalty Drawing

I think penalty drawing is an indication of an ability to play an up-tempo game. It is probably the best statistical indication of aggressive play. Thank you NHL for finally giving us this data.

The list of the top PCOPPO scores (below) demonstrate that it is not just talented players (like Crosby) who draw penalties, but also muckers like Avery and Burrows.

I expected forwards to dominate this list and they do – only 2 of the top 30 PCOPPO scores belong to defensemen. Even though the expectations for penalty drawing that are built in to the PC calculations are lower for defensemen, drawing penalties is just not what they do. It takes an unusual skill set for a defender to draw a lot of penalties. And Dion Phaneuf did just that (53 minors).

Copyright Alan Ryder, 2008 Hockey Analytics www.HockeyAnalytics.com 2008 NHL Review Page 29

At the other end of the spectrum, eight forwards tied Penalty Drawing– Best for the worst penalty drawing performance in 2008. Others Player Team Minors PCOPPO were penalized less but this Dion Phaneuf CAL 53 13 group averaged 1460 minutes of Sidney Crosby PIT 55 12 playing time and, on average, Dustin Brown LA 62 11 drew a minor penalty once VAN 60 11 every 6 games. Erik Cole CAR 52 10 NYR 43 10 Net Penalty Impact Penalty Drawing – Worst The name that leaps off the page from the above analysis is Player Team Minors PCOPPO Dion Phaneuf. According to NYR 11 -6 Alexander Frolov LA 10 -6 my calculations he was the Daymond Langkow CAL 16 -6 NHL’s most valuable penalty PHI 16 -6 drawer. But, by taking a lot of MIN 17 -6 penalties (the fourth worst Petr Sykora PIT 11 -6 SHO PCD score in the NHL), he Eric Perrin ATL 13 -6 cost his team a more than he NAS 13 -6 contributed by drawing penalties. He could have been the NHL’s most valuable defenseman (in my books) but for his penalty taking record. The net impact of taking and drawing penalties cost him 9 PC points (and Calgary about one point in the standings).

This begs the following question – does Net Penalty Opportunities aggressive play, generally considered a Best good thing, have to result in penalties? I have been considering this question for Player Team PCNPO years with no clear answer. Martin St. Louis TB 20 Pavel Datsyuk DET 20 My best answer is in the tables to the right Brad Richards TB 17 where I show PC points for net penalty Alexander Ovechkin WAS 17 opportunities (PCNPO = PCOPPO – PCDSHO). Daniel Girardi NYR 16 While there is a pretty clear correlation Dustin Brown LA 15 between taking and drawing penalties, for Net Penalty Opportunities the most part, those taking a lot of penalties Worst do not make up for it by drawing penalties. And the impact of penalty avoidance is Player Team PCNPO generally not watered down by the failure John Erskine WAS -24 to draw penalties. Cory Sarich CAL -23 Chris Neil OTT -22 TB -18 Pavel Kubina TOR -18 Daniel Carcillo PHO -18

Copyright Alan Ryder, 2008 Hockey Analytics www.HockeyAnalytics.com 2008 NHL Review Page 30

Top Rookie

A lot of famous names have won the Calder Trophy as the NHL’s top rookie. But a very large number of very good hockey players never got their name on this trophy. Of the NHL’s top ten all time leading scorers only one, Mario Lemieux, was awarded the Calder (the others – Wayne Gretzky, , Gordie Howe, , , , , Jaromir Jagr and Phil Esposito). With this kind of history, I am thinking that Sidney Crosby is happy that he lost out to Alexander Ovechkin in the 2006 voting.

The top eight PC scores in 2008 from rookies are shown to the right. The reason Mark Messier Award I have eschewed my usual format (of six Top Rookie leaders) is the goaltender factor – three of the top ten were goalies. Player Team PC MON 157 Carey Price was a bit late to the party but ANA 95 played increasingly and increasingly well CHI 80 over the season to cause to Erik Ersberg LA 69 be dispatched to Washington at the trade Nicklas Backstrom WAS 64 deadline. Price’s numbers read like half a Matt Niskanen DAL 62 season (41 games, 2413 minutes, 1282 EDM 57 shots) but that was enough activity for the VAN 53 PC formula to award 157 PC points (4 on offense, 33 in the shootout and 121 for regulation/overtime goaltending). The main driver of this was a high shot quality neutral save percentage (.919). With a full season of that kind of activity Price might well have been my top goaltender and MVP. His consolation prize is my Mark Messier Award for the best contribution by a rookie.

Calder Trophy voters have not been shy about selecting a goaltender (Alexander Raycroft in 2004 and Evgeni Nabakov in 2001 are recent examples), but they have discriminated against goalies in the past. On the basis of not enough games (or wins!) their implicit rule book excluded Price, although he did finish fourth in the voting.

In second place was Hiller (23 games, 1223 minutes, 578 shots). His save percentage was .927 and I had his shot quality neutral save percentage at .931 (Anaheim had poor shot quality numbers). He made quite a contribution in limited playing time but got no votes for the Calder. The other goalie on this list is Ersberg (LA, 14, 799, .928). LA used four rookies in net behind the youthful LaBarbera.

The finalists for the Calder Trophy were Washington’s Nicklas Backstrom and the dynamic duo from a very promising young Chicago team, Patrick Kane and . Kane was the deserving winner. A look at the PC breakdown of the leading rookie performances by skaters (below) shows that there are many ways to contribute to the success of a team.

Copyright Alan Ryder, 2008 Hockey Analytics www.HockeyAnalytics.com 2008 NHL Review Page 31

PCO PCD Player Team POS PC EH PPP PPO SH SO PCO EH PP SHK SHO PCD Patrick Kane CHI RW 80 29 19 2 0 25 74 6 0 0 0 6 Nicklas Backstrom WAS C 64 28 8 0 0 2 39 11 0 2 12 26 Matt Niskanen DAL D 62 11 3 -1 0 0 12 28 4 13 5 49 Tom Gilbert EDM D 57 21 4 -2 1 -1 23 14 0 10 10 33 Alexander Edler VAN D 53 4 4 -2 0 2 8 30 2 8 4 44

In sorting out Kane versus Toews versus Backstrom the Calder voters would have gone straight to conventional scoring stats. Toews collected 24 goals and 30 assists in just 64 games, the other two played the full schedule with Kane scoring 21 goals and assisting on 51 and Backstrom scoring 14 times and contributing 55 helpers.

This analysis misses two important considerations – defense and the shootout. Kane collected a very large and league leading (see below) 25 PC points in the shootout, whereas Backstrom had a PCOSO score of 2 and Toews was at -2. Backstrom was pretty effective defensively (26 PCD versus 6 for Kane and 9 for Toews). PC shows Kane as clearly having the top rookie performance by a skater. Backstrom is revealed as a very effective two way player. Toews’ numbers (49 PC points) were hurt by injury time, but his numbers gross up to Backstrom and not to Kane.

Calder voters typically ignore defensemen. Three of my top five rookie skaters were blueliners. Niskanen put up some very useful defensive numbers (49 PCD). Edler and Enstrom had similar profiles. Gilbert got on this list on the strength of his offense (23 PCO).

Most Improved

We have already seen a number of young, no-name defensemen near the top of the PC rankings with breakthrough seasons. It turns out that Brent Burns, and Mike Green were two of the most improved players in the NHL. To the right are the top six most improved PC scores among skaters (as always, I am carrying decimal places in my calculations that mean that the differences reflect rounding). Each of these players had a full season in 2007 (Richards played the fewest games in the group – 59) so the improvement was due to performance and the additional ice time that typically goes with it. Gordie Howe Award Most Improved Skater If there was a vote on this I suspect that Alexei Kovalev Player Team 2007 2008 Change would have been the Mike Green WAS 19 97 79 winner. He went from Adrian Aucoin CAL 22 88 66 Montreal’s goat in 2007 to Alexander Ovechkin WAS 99 162 64 Nikolai Zherdev CBJ 23 85 61 the interim in 2008 Brent Burns MIN 48 108 59 with a much improved Mike Richards PHI 29 87 59 performance (+46 PC points) that got a lot of

Copyright Alan Ryder, 2008 Hockey Analytics www.HockeyAnalytics.com 2008 NHL Review Page 32

media attention. But my calculations say that others showed more improvement.

Mike Green came out of nowhere to become an elite defenseman in 2008 and win my Gordie Howe Award as the most improved player in the NHL. His playing time nearly doubled from 1084 minutes (895 even handed, 12 short handed, 177 power play) in 2007 to 1938 minutes (1525, 28, 385) in 2008 reflecting the significant uptick in his offensive performance (he was the NHL’s top offensive defenseman while even handed).

Mike Green PCO PCD Season PC EH PPP PPO SH SO PCO EH PP SHK SHO PCD 2007 19 0 -2 0 0 -1 -3 17 2 2 1 21 2008 97 31 17 7 0 -1 53 38 4 3 -1 44

Brent Burns more than doubled his impact year over year. He went from 1216 minutes in 2007 to 1894 minutes in 2008, picking up 359 minutes even handed, 158 minutes on the penalty kill and 161 minutes on the power play. And he showed considerable impact in the shootout.

Brent Burns PCO PCD Season PC EH PPP PPO SH SO PCO EH PP SHK SHO PCD 2007 48 9 6 0 1 0 16 25 1 1 5 33 2008 108 20 17 -2 0 16 51 35 3 18 1 57

If anyone had forecast at the beginning of the season that Ovechkin would have been one of the most improved players in the league there would have been no takers. His 2008 season was truly remarkable.

And I thought Adrian Aucoin was washed up.

When one looks for improved performance amongst goaltenders, changes in roles are more binary and get in the way. The biggest improvement in PC score came from Ilya Bryzgalov (+138) mainly because he went from the bench in Anaheim to the starter’s role in Phoenix. The second highest improvement, however, was more due to performance. (+127) has been a career reliever. He did have a material improvement in playing time in 2008 (due to injuries to Marc-Andre Fleury) but he also played well above his previous numbers. Both Jose Theodore (+104) and Martin Gerber (+77) were much improved in 2008, winning back starting jobs they had lost. But the most improved player without a role change was probably Tomas Vokoun (+113).

Shootout Awards

There were 156 points contested in the shootout in 2008. The rather limited number of contestants meant that certain players earned a very significant percentage of their PC points in the fifth period.

Copyright Alan Ryder, 2008 Hockey Analytics www.HockeyAnalytics.com 2008 NHL Review Page 33

The third annual Wyatt Earp Award, as the top shootout gunslinger, goes this season to Chicago rookie Patrick Kane, successful in 7 of his 9 attempts. His 25 PCOSO points represented nearly one third of his total (80 PC points). The leaderboard is shown below.

Tied with 7 shootout goals was Joe Pavelski, but he required 11 attempts. His 22 PCOSO points represented over one quarter of his total (82 PC points).

Ales Hemsky was third in goals (6) but needed 16 tries and his PCOSO score of 12 (PC credits goals but ‘taxes’ attempts) was not good enough for the leaderboard.

Here is the performance of the rest of the leaders – Elias went 5 of 6 attempts, Kessel was 5 of 13, Brent Burns was the top defenseman (3 of 5) and Dawes was 5 of 9. Burns limited exposure to the shootout was Shootout Awards grossed up by Minnesota’s very efficient use of shootout goals (3 wins based on just Wyatt Earp Award 3 marginal goals). PC gave him roughly Top Shooter half credit for Minnesota’s shootout success. Player Team PC Patrick Kane CHI 25 The worst shootout performances by a Joe Pavelski SJ 22 skater were from Jaromir Jagr (0 of 5), Patrik Elias NJD 17 BOS 16 Chris Drury (0 of 5), (0 of 6), Brent Burns MIN 16 (0 of 5) and Alexander Nigel Dawes NYR 16 Radulov (1 of 8), each with a PCOSO score of -1. Cork Award Top Stopper For goaltenders in the shootout I present the Cork Award for the best stopper. As Player Team PC with all of PC, this can be thought of as Mathieu Garon EDM 63 “performance” x “workload” and each of Roberto Luongo VAN 53 the leaders had a high exposure to the Martin Brodeur NJD 51 shootout. Mathieu Garon was acquired by Tomas Vokoun FLA 47 Edmonton as a backup for Roloson and Henrik Lundqvist NYR 45 ended up playing 47 games. He stopped 30 Johan Hedberg ATL 41 of 32 (save percentage .938!) shootout Dwayne Roloson EDM 41 attempts and PC credited him with 63 points. Roloson did just fine, saving 27 of 37 attempts (.730 save percentage) for a sixth place (41 PCGSO) ranking. Luongo faced more attempts (54) than any other goaltender, allowing 15 goals (save percentage of .722). Brodeur was his usual self as well (43 attempts, 32 saves, .744 save percentage). Vokoun had a similar record (41, 30, .732). Lundqvist was a little off the pace (42, 29, .690). Hedberg had less work but performed a bit better (31, 24, .774).

Copyright Alan Ryder, 2008 Hockey Analytics www.HockeyAnalytics.com 2008 NHL Review Page 34 All-Star Contributors

NHL

NHL First Team Second Team Position Name Team PC Name Team PC LW Alexander Ovechkin WAS 162 IIya Kovalchuk ATL 124 C Pavel Datsyuk DET 132 Joe Thornton SJ 115 RW Jarome Iginla CAL 141 Daniel Alfredsson OTT 109 D Nicklas Lidstrom DET 121 Brian Campbell BUF/SJ 107 D Brent Burns MIN 108 Brian Rafalski DET 105 G Tomas Vokoun FLA 267 Martin Brodeur NJD 259

The NHL All-Star Team is selected by the hockey writers. You would expect the voting to mirror theirs for the Hart Trophy. But a strange thing happened this season – Martin Brodeur, who was selected by the GMs as the NHL’s best goaltender, was the top goaltender in the Hart race but was selected only to the second all star team behind Evgeni Nabakov.

This is just plain strange. And wrong.

I had Nabakov (77 games, 4561 minutes, 2.14 GAA, .910 SV%, .904 SQNSV%) ranked 26th (!!!!!) in ‘contribution’ or ‘impact’ or ‘value’ (take your pick – I think these words are interchangeable here) amongst goaltenders with 92 PC points. Through the lens of Player Contribution his best comparator is Buffalo’s Ryan Miller (76 games, 4474 minutes, 2.64 GAA, .904 SV%, .906 SQNSV%, 98 PC points). I gave both players 15 PC points in the shootout. A superficial assessment would be that Nabakov played a bit better (GAA and save percentage) in similar playing time. But this ignores shots (Nabakov – 23.7 per game versus Miller – 28.2 per game) and shot quality. The best single metric for a goaltender is his shot quality neutral save percentage and Miller’s was also .904. Miller had more work (shots) and was therefore more impactful.

I am still shaking my head. Repeat after me “goaltender wins is the most useless statistic in hockey”. Nabakov had a league leading 46 wins based on (a) 77 games and (b) very good teammates. Miller had 36 wins but had the privilege of playing behind much weaker teammates.

We get a similar anomaly with the defensive all-stars. The top four in the Norris Trophy voting were, in order, Lidstrom, Phaneuf, Chara and Sergei Gonchar. Yet the voters managed to correctly place Brian Campbell on the second all-star team over Gonchar (who I had ranked as the 13th most impactful defender in 2008).

Brent Burns is the surprising name here. He is a young guy (22) who was much improved in 2008. His contribution in the shootout caused PC to elevate him to all-star status. If not for this I would have had Campbell on the first team with Chara (by a fraction of a PC point) over Mike Green on the second team.

Copyright Alan Ryder, 2008 Hockey Analytics www.HockeyAnalytics.com 2008 NHL Review Page 35

Phaneuf missed my team because of his excessive penalty taking. The voters routinely ignore this aspect of the game, but he spent over two games in the penalty box.

Given the Hart voting it is no surprise that Malkin was voted by the writers as the first team . I had him ranked third – well behind Datsyuk and a couple of PC points behind Thornton. He does not play defense (the voters don’t care) and he takes too many penalties (the voters don’t know to notice).

We can all agree on Jarome Iginla as the first team all-star at right wing. What about the writers’ choice of Alex Kovalev for the second team? I have him ranked well off the pace (86 PC points) as the 6th most impactful right behind Iginla (141), Daniel Alfredsson (109), Marian Gaborik (106), Jason Pominville (94) and Brad Boyes (94). It was a turnaround year for him but this result is quite unfair to Alfredsson and Gaborik. He can be compared most directly to Gaborik since they have a similar propensity to eschew defense. The two players had similar points totals (Kovalev 84 – Gaborik 83) but Gaborik had more goals (42 versus 35) and especially many more goals the hard way – while even handed (31 versus 18).

On left wing I have Kovalchuk just ahead of the writers’ choice, Henrik Zetterberg, (by 3 PC points) on the second team.

Repeating from my 2007 team were Ovechkin, Datsyuk, Lidstrom and Brodeur.

East

East First Team Second Team Position Name Team PC Name Team PC LW Alexander Ovechkin WAS 162 IIya Kovalchuk ATL 124 C Evgeni Malkin PIT 113 Eric Staal CAR 102 RW Daniel Alfredsson OTT 109 Jason Pominville BUF 94 D Brian Campbell BUF/SJ 107 Mike Green WAS 97 D Zdeno Chara BOS 97 Tomas Kaberle TOR 93 G Tomas Vokoun FLA 267 Martin Brodeur NJD 259

The East dominated the NHL all star team in goal (Vokoun and Brodeur, a repeat from 2007). But the West was best on defense. I have Campbell on the East team as he spent most of the season in Buffalo. Chara edged out Mike Green for the other first team spot. Kaberle’s performance was off this year but did repeat from 2007.

Ovechkin is the only repeat performer up front. Compared to the NHL all-star team the new faces are Malkin and Eric Staal at centre and Pominville on the right wing.

Honourable mention goes to Jason Spezza (99), Vincent Lecavalier (93), Derek Roy (92) and Zach Parise (90), each of whom cleared the 90 point milestone. In goal I should note Tim Thomas (237) and Henrik Lundqvist (222).

Copyright Alan Ryder, 2008 Hockey Analytics www.HockeyAnalytics.com 2008 NHL Review Page 36

West

West First Team Second Team Position Name Team PC Name Team PC LW Henrik Zetterberg DET 121 Rick Nash CBJ 92 C Pavel Datsyuk DET 132 Joe Thornton SJ 115 RW Jarome Iginla CAL 141 Marian Gaborik MIN 106 D Nicklas Lidstrom DET 121 Brian Rafalski DET 105 D Brent Burns MIN 108 Dion Phaneuf CAL 93 G J-S Giguerre ANA 232 Roberto Luongo VAN 199

Four all stars from Detroit explain a great deal about the Wings’ success in 2008.

Datsyuk and Thornton carry over from the NHL all-star team at centre. In fact they are repeating from 2007.

Iginla is the only other repeater among the forwards. Gaborik had a great season to make the second team at right wing. Zetterberg had a greater season to be the first team left winger. And Rick Nash has returned from the dead.

The defensive all-stars look a lot like the all-NHL team. The new face is somebody named Phaneuf. Lidstrom is the only repeater from last season.

In spite of an off season, Roberto Luongo repeated in goal. But he had to settle for the second team behind Giguerre.

Honourable mention goes to Brad Boyes (94).

Rookie

Rookie First Team Second Team Position Name Team PC Name Team PC LW Nigel Dawes NYR 50 VAN 26 C Nicklas Backstrom WAS 64 Jonathan Toews CHI 49 RW Patrick Kane CHI 80 SJ 28 D Matt Niskanen DAL 62 Alexander Edler VAN 53 D Tom Gilbert EDM 57 Tobias Enstrom ATL 50 G Carey Price MON 157 Jonas Hiller ANA 95

As is usually the case, this list gets thin pretty fast. This was especially true with the forwards this season. It is hard to recognize the performances of Setoguchi and Raymond with PC scores in the 20s. Kane, Backstrom and Toews had strong rookie outings, leading the Calder voting. Peter Mueller (PHO) finished 5th in the voting and is an honourable mention at 46 PC points.

Dawes finished only 12th in the Calder voting. He had about the same PC score as teammate (who collected more Calder votes) until you get to the shootout (where he went 5 for 9 for 16 PC points).

Copyright Alan Ryder, 2008 Hockey Analytics www.HockeyAnalytics.com 2008 NHL Review Page 37

The voters placed Enstrom on the all-rookie team rather than Niskanen. I think they were impressed with his ice time (2007 minutes versus 1598) and offense. The time-on-ice differential was mainly on the power play (471 versus 171) yet PC gave Niskanen more points for offense. On the surface Enstrom’s offense (5 goals, 33 assists) does look more impressive than Niskanen’s (7, 19), but his power play offense (4, 22) needs to be dialed back a bit (PC does this) because of higher expectations on the PP. Meanwhile Niskanen (5, 13) was way better while even handed (where Enstrom collected 1 goal and 11 assists). On defense Niskanen was clearly better.

Price was fourth in Calder voting and made a clear case for the all-rookie team. Hiller got backup ice-time but performed very well (save percentage was .927).

Green (24 and under)

Green First Team Second Team Position Name Team PC Name Team PC LW Alexander Ovechkin WAS 162 IIya Kovalchuk ATL 124 C Evgeni Malkin PIT 113 Eric Staal CAR 102 RW Nikolai Zherdev CBJ 86 Dustin Brown LA 80 D Brent Burns MIN 108 Dion Phaneuf CAL 93 D Mike Green WAS 97 Dennis Wideman BOS 83 G Carey Price MON 157 Kari Lehtonen ATL 152

Up front this looks a great deal like the East all-star team. Each of Ovechkin (22), Kovalchuk (24), Malkin (21) and Staal (23) carried over from that team. Forwards tend to peak around the age of 23. The two new faces are on right wing – Zherdev (23) and Brown (23), who nosed out Patrick Kane (19).

Defenders generally take longer to hone their craft, but there are also a number of very impressive youngsters on the blueline. Burns (22), Green (22) and Wideman (24) had surprising development this year. Dion Phaneuf (22) went missing last season but has to be considered the best bet in this group to win a Norris Trophy one day.

Goalies can also take longer. But that did not stop Carey Price (20) from having a great deal of impact. Lehtonen (24) graduates from the team this season.

There are a lot of good young players in the NHL today. Only two players, Ovechkin and Lehtinen, repeated from last season. Honourable mention goes to Jason Spezza (age 24, 99 PC points), Rick Nash (23, 92), Derek Roy (24, 92) and Zach Parise (23, 90).

None of the rookie forwards made this team. But look for Evgeni Malkin (20), (18), (21) and Anze Kopitar (19) next year. Prominent names missing: Ilya Kovalchuk (23), Marion Gaborik (24), Rick Nash (22), Ales Hemsky (23) and Eric Staal (22). Each of these players earned north of $4,000,000 in 2008.

Note that, for both of the Green and Grey teams, I used the player’s age as of December 31, 2006.

Copyright Alan Ryder, 2008 Hockey Analytics www.HockeyAnalytics.com 2008 NHL Review Page 38

Grey (34 and over)

Grey First Team Second Team Position Name Team PC Name Team PC LW Brian Rolston MIN 65 Ray Whitney CAR 60 C Mats Sundin TOR 81 John Madden NJD 52 RW Daniel Alfredsson OTT 109 Alexei Kovalev MON 86 D Nicklas Lidstrom DET 121 Adrian Aucoin CAL 88 D Brian Rafalski DET 105 Matthieu Schneider ANA 77 G Martin Brodeur NJD 259 STL 173

Each of Brodeur (35), Lidstrom (37), Rafalski (34) and Alfredsson (35) were fully fledged all-stars. Rafalski is the rookie of the year for this team. Each of the others repeated from last year. When you are long in the tooth, it is hard to keep it going. Other repeaters on this team are Schneider (38), the oldest player to make the team, and Whitney (35).

In goal Manny Legace (age 34, 173 PC points) fended off resurgent (34, 116).

Adrian Aucoin (34), with a change in venue, was much improved in 2008 and made the team on defense. Honourable mention on defense goes to Adam Foote (age 36, 64 PC points) and, especially, Sergei Zubov (37) who collected 59 PC points in just 46 games.

On right wing was another player who rebuilt his game in 2008, Alexei Kovalev (at 34, another rookie). On left wing, Brian Rolston (34) is aging well and was the cheapest player on this team (2.4 million of cap hit). Mats Sundin (36) bounced back a bit this year but did not have much competition for the team. At centre on the second team was John Madden (34) with the lowest PC score (52) on the team.

Remarkable, ageless, future hall-of-famer, organizer of player unions Chris Chelios (45) nearly made the team in 2007 but was out of the running this season with ‘only’ 57 PC points. Time to hang ‘em up Chris.

All Offense

Offense First Team Second Team Position Name Team PCO* Name Team PCO* LW Alexander Ovechkin WAS 139 IIya Kovalchuk ATL 105 C Evgeni Malkin PIT 110 Vincent Lecavalier TB 84 RW Jarome Iginla CAL 110 Marian Gaborik MIN 98 D Dion Phaneuf CAL 63 Zdeno Chara BOS 51 D Mike Green WAS 54 MON 49

* PCO excluding shootouts.

We have seen most of these names already, but there are only two repeaters from last season. Ovechkin had an offensive season for the ages. In 2007 Vincent Lecavalier was

Copyright Alan Ryder, 2008 Hockey Analytics www.HockeyAnalytics.com 2008 NHL Review Page 39

the only player to amass over 100 PCO points (before the shootout). In 2008 that number rose to four. Remember that a PC point represents 0.1 points in the standings and, as the number of points awarded is very stable from year to year (it depends on the number of games going to overtime), this increase represents an increased concentration of scoring in elite talent. I think that is a good thing for the NHL.

Up front honourable mention has to go to Henrik Zetterberg (90 PCO, ex shootout) and Daniel Alfredsson (89 PCO, ex shootout).

Phaneuf led the way on defense with his rather freakish ability to draw penalties. But Green and Chara were also superior penalty instigators. Phaneuf (10 goals, 23 assists) had the second best power play offense, trailing only Sergei Gonchar (8, 38). Green had the best even-handed offense (10, 23) amongst NHL defenders.

The new face on defense is Mark Streit. He seemed most apparent on the power play (7 goals, 27 assists) but had more notable impact even handed where his 26 PCOEH points were second only to Green. An honourable mention needs to go to teammate Andre Markov who was ranked 5th amongst defenders with 46 PCO points (before the shootout). Streit had some playing time (but we don’t know how much) on the wing. Forwards get to participate in more offense and this may have boosted Streit ahead.

All Defense

Defense First Team Second Team Position Name Team PCD Name Team PCD LW Alexander Steen TOR 28 P.J. Axelsson BOS 28 C Pavel Datsyuk DET 39 Daymond Langkow CAL 33 RW Mike Grier SJ 37 Martin St. Louis TB 33 D Marc-Edouard Vlasic SJ 80 Rostislav Klesla CBJ 73 D Nicklas Lidstrom DET 78 Jan Hejda CBJ 73

The repeaters from last season are Vlasic, who moved up from the second team, Lidstrom, Datsyuk and St. Louis. Toronto now lists Alexander Steen as a left winger, a position he has been playing for years. Had he been listed there last season I would be calling him a repeater as well.

Honourable mention should go to forwards Jason Pominville (32), Boyd Gordon, Joe Thornton and Ryan Johnson, each of whom cleared 30 PCD points.

Usually the defenders on this list are two way players. Certainly Lidstrom is. But Vlasic and Klesla are stay-at-home-defensemen. Vlasic had just 2 goals and 12 assists (-1 PCO). Klesla was only a hair better (6 goals, 12 assists, 1 PCO) thanks mainly to more power play time.

Among defenders, honourable mention should go to Brian Rafalski (69 PCD) and Ron Hainsey (67).

Copyright Alan Ryder, 2008 Hockey Analytics www.HockeyAnalytics.com 2008 NHL Review Page 40

Even Handed

Even Handed First Team Second Team Position Name Team PCEH Name Team PCEH LW Alexander Ovechkin WAS 126 IIya Kovalchuk ATL 82 C Pavel Datsyuk DET 84 Evgeni Malkin PIT 81 RW Jarome Iginla CAL 102 Marian Gaborik MIN 81 D Nicklas Lidstrom DET 79 Adrian Aucoin CAL 69 D Mike Green WAS 75 Brian Rafalski DET 67

PCEH = PCOEH + PCDEH + PCOPPO + PCDSHO

The NHL’s so-called more open game really is not. The reality is that there is more scoring because there is less even handed time. But about two-thirds of the game is played even handed. And you don’t need a degree in math to figure out that this still matters more than power play time.

My definition of PCEH includes ALL of penalty drawing (PCOPPO) and penalty taking (PCDSHO). Note that this is a simplification as penalties are drawn/taken when not even- handed. But the NHL does not give us the data to split it out. So this is the best I can do for now.

This team resembles the NHL all-star team. Interestingly, last year’s second forward line of Ovechkin, Datsyuk and Ignila moved up to the first line in 2008. Two of these players cracked the 100 PC point barrier (nobody came close last season). The second line is populated by now-familiar faces. Close-but-no-prize goes to Joe Thornton (81) and Henrik Zetterberg (79).

Lidstrom and Rafalski, in spite of a change in team, repeated on defense.

Lidstrom collected 48 PCDEH points (third in the NHL). He played 1370 minutes and posted a goals against average of 1.71 (goals per 60 minutes of even handed play). Among defenders with a material workload this GAAEH was bested by only New Jersey’s Paul Martin (1.52) who enjoyed better goaltending support. Lidstrom added 24 PCOEH points and 6 net penalty drawing/avoidance points (rounding means that it does not quite add up) to make the team.

Brian Rafalski got here largely with even-handed defense (his 45 PCDEH points ranked seventh in the NHL) but his 14 PCOEH points were not insignificant. Mike Green led defenders with even-handed offense. Aucoin tied for fourth in even-handed defense.

Power Play

Remember that Player Contribution reduces to “performance” x “ice time”. Here is where playing time really matters. This is not a list of the “best” power play performers, it is a list of the biggest contributions on the power play.

Copyright Alan Ryder, 2008 Hockey Analytics www.HockeyAnalytics.com 2008 NHL Review Page 41

Power Play First Team Second Team Position Name Team PCPP Name Team PCPP LW Alexander Ovechkin WAS 37 Henrik Zetterberg DET 32 C FLO 37 Evgeni Malkin PIT 34 RW Alexei Kovalev MON 44 Jarome Iginla CAL 29 D Sergei Gonchar PIT 36 Brian Rafalski DET 29 D Dion Phaneuf CAL 30 Andrei Markov MON 29

PCPP = PCOPPP + PCDPP

The NHL’s top power play performance in 2008 came from Alexei Kovalev. He collected 17 goals and 30 assists in 341 of PP time. Ovechkin had better numbers (22 goals, 15 assists) than Jokinen (18, 15) but took a lot more ice time to get there (465 versus 312 minutes). PC found that the extra 4 goals in 153 extra minutes was a marginal performance and gave them the same score.

On defense the leader was Gonchar with 8 goals and 38 assists in 440 minutes.

Honourable mention goes to Mark Streit (who missed second team defense by a fraction of a PC point), Daniel Briere (32 PCPP), Eric Staal (31) and Pavel Datsyuk (29).

Short Handed

Short Handed First Team Second Team Position Name Team PCSH Name Team PCSH F CHI 30 OTT 21 F Daniel Alfredsson OTT 25 Mike Grier SJ 21 D Adam Foote CBJ/COL 30 Robin Regehr CAL 21 D Jan Hejda CBJ 23 Cory Sarich CAL 21

PCSH = PCOSH + PCDSHK

For forwards it is common to get to the head of this class with offense. If you ignore the offense and just look at penalty killing you get the following results (which do not look very different for defenders).

Penalty Killing First Team Second Team Position Name Team PCSH Name Team PCSH F Chris Kelly OTT 16 Ryan Johnson STL 15 F Dave Steckel WAS 15 Craig MacDonald TB 14 D Adam Foote CBJ/COL 28 Cory Sarich CAL 19 D Jan Hejda CBJ 22 Nick Schulz MIN 19

Adam Foote distanced himself from the pack in this analysis. His GAASH of 5.10 was not the best in the league (many players were under 5.00 per game) but it was very good and he got a lot of ice time (338 minutes). Only three players got more time on the penalty kill, but their results were not up to those of Foote – Duncan Keith (374 minutes, 7.22 GAASH), (368, 7.17) and Francois Beauchemin (341, 7.75). With

Copyright Alan Ryder, 2008 Hockey Analytics www.HockeyAnalytics.com 2008 NHL Review Page 42

more short handed ice time Brent Burns (212, minutes, 4.53 GAASH), Daniel Girardi (208, 4.32) and Bruno Gervais (173, 4.15) could have made these teams.

You can see that all of the forwards made the first leaderboard based on offense. Sharp and Alfredsson each had 7 goals and 2 assists while short handed. Vermette netted 3 goals and 3 assists and Grier had 3 goals and 1 assist.

Kelly had a GAASH of 4.86 in 284 minutes. Ryan Johnson put in more penalty killing time (300 minutes) than any other forward posting a respectable GAASH of 5.61. Steckel was nominally a bit worse (GAASH of 5.69) in 253 minutes, but PC assessed his goaltending (mainly Kolzig) as worse than Steckel’s (mainly Legace) and gave him the same score.

MacDonald had the lowest GAASH on the team (3.95) in front of terrible goaltending, but played only 137 minutes on the PK. Teammate Mathieu Darche had a spectacular penalty killing record (GAASH of 2.09) in only 86 minutes. And (DET) was even better (GAASH of 1.92 in just 124 minutes).

MacDonald had a lot of company at 14 PCDSHK points – each of Grier, Patrick Rissmiller (SJ), Vermette and Chris Drury (NYR) were within decimal places of the all-star penalty killing team.

Most Valuable Performances

In a sense value is a relative thing. When two players produce the same outputs, a team would prefer the player with the smaller paycheck. And, in today’s salary cap era, certain contracts are seen as liabilities because no plausible performance can justify the cost.

In this spirit I present the All-Value teams for 2008:

All Value First Team Second Team $ PC $ PC Position Name Team Cost Name Team Cost LW Milan Michalek SJ 9,122 Daniel Paille BUF 9,863 C Joe Pavelski SJ 10,335 Mike Richards PHI 10,774 RW ANA 8,328 Richard Park NYI 10,701 D Dennis Wideman BOS 7,236 Dion Phaneuf CAL 8,424 D Brent Burns MIN 7,673 Mike Green WAS 8,589 G Dan Ellis NAS 3,706 Tim Thomas BOS 3,800

PC Cost is the Cap Cost per PC point. I used the annual cap cost (in US dollars) per annum, rather than the per diem approach used in the NHL’s CBA, to screen out players with limited playing time but high per game PC scores.

Obviously a lower PC Cost is better and the table above shows the value leaders. To put these costs into perspective, a salary cap of about $50 million (for the 2006-07 season)

Copyright Alan Ryder, 2008 Hockey Analytics www.HockeyAnalytics.com 2008 NHL Review Page 43

suggests that an average player on a playoff bound team should cost at most $50,000 per annum per PC point5.

Dan Ellis was the value leader, costing $500,000 (per annum) and producing 135 PC points based largely on a .922 shot quality neutral save percentage in 44 games. Thomas was a bit more expensive ($760,000) but delivered more – 200 PC points based on a .918 SQNSV% in 57 games.

We have seen most of these defensemen before. Burns (108), Phaneuf (93) and Green (97) had three of the best PC scores of all defensemen. It turns out that they work cheap.

Among skaters the value leader was Boston’s Dennis Wideman (83 PC points in 81 games at a cap cost of $600,000). The most expensive performance from the leaderboard was that of Richards (87 PC points in 73 games at a cost of $942,400). The weakest performance was from Richard Park (43 PC points in 82 games at a cost of $462,500), but that kind of performance for a “minimum wage” is quite useful.

All Cap Roster

If all NHL players had been free agents at the beginning of the 2008 season and could have been signed for their then current cap cost, who would you want on your team? Herein I present my All Cap roster. This is a list of 23 players you might want to have on your team if you were prepared to max out your cap costs while attempting to max out performance.

This was a pretty subjective exercise. I am sure there is a mathematical solution to this optimization problem. But I know that any solution would involve a lot of variables and constraints and would tax my computer (never mind my programming skills).

So I did this by eye. This is my All Cap Roster. If you can find a more optimal one, share it with me:

5 A team with playoff aspirations needs to target a 100 point season. This translates into 1000 PC points and gives you my $50,000 average cost per PC point. A team is likely to be better than this and needs to have a lower average cost per PC point.

Copyright Alan Ryder, 2008 Hockey Analytics www.HockeyAnalytics.com 2008 NHL Review Page 44

All Cap Roster – 2008

Actual Projected Position Name Team CapCost MOI PC MOIEH MOISH MOIPP MOITOT PC LW1 Alexander Ovechkin WAS 3,834,200 1,758 162 1,100 0 350 1,450 125 C1 Pavel Datsyuk DET 6,700,000 1,655 132 1,100 100 350 1,550 115 RW1 Jarome Iginla CAL 7,000,000 1,748 141 1,100 0 350 1,450 131 LW2 Henrik Zetterberg DET 2,650,000 1,560 121 900 100 200 1,200 84 C2 Evgeni Malkin PIT 3,834,200 1,502 113 900 0 200 1,100 74 RW2 Daniel Alfredsson OTT 4,338,996 1,570 109 900 250 200 1,350 97 LW3 Patrick Sharp CHI 775,000 1,637 78 700 250 100 1,050 68 C3 Mike Richards PHI 942,400 1,428 87 700 150 100 950 56 RW3 Brad Boyes STL 1,500,000 1,480 94 700 0 50 750 57 LW4 Milan Michalek SJ 628,267 1,471 69 700 50 50 800 44 C4 Anze Kopitar LA 955,867 1,583 83 700 50 100 850 40 RW4 Jason Pominville BUF 1,033,333 1,696 94 700 150 50 900 51 F5 Dustin Brown LA 1,175,000 1,758 80 450 0 0 450 18 F5 J.P. Dumont NAS 546,667 1,655 73 450 0 0 450 25 F TOTAL 35,913,929 22,738 1,436 14,300 987 D1 Adrian Aucoin CAL 4,000,000 1,593 88 1,400 200 50 1,650 93 D1 Mike Green WAS 833,333 1,938 97 1,400 100 50 1,550 81 D2 Tomas Kaberle TOR 4,250,000 2,039 93 1,200 100 300 1,600 64 D2 Dion Phaneuf CAL 785,333 2,167 93 1,200 300 300 1,800 81 D3 Brent Burns MIN 825,000 1,894 108 900 300 150 1,350 71 D3 Dennis Wideman BOS 600,000 2,038 83 900 200 150 1,250 51 D4 Ron Hainsey CBJ 737,500 1,760 79 550 25 25 600 23 D TOTAL 12,031,167 13,429 639 9,800 639 G1 Tim Thomas BOS 760,000 3,342 200 3,700 232 G2 Dan Ellis NAS 500,000 2,229 135 1,300 81 G TOTAL 1,260,000 5,571 335 5,000 313 TEAM TOTAL 49,205,096 2,410 1,939

The rules I used to put this together and evaluate the output are:

• Cap Cost < $50,500,000

• 14 forwards, 7 defensemen, 2 goaltenders.

• Players play in position and are seeded into a depth chart for each situation.

• Players are attributed ice time (MOITOT) based on their position in the depth chart:

o The prima goalie gets 3,700 minutes (about 61 games). The second fiddle gets 1,300 minutes (about 21 games).

Copyright Alan Ryder, 2008 Hockey Analytics www.HockeyAnalytics.com 2008 NHL Review Page 45

o Even handed, first line forwards are assumed to play 1,100 minutes, second line 900 minutes and the third and fourth lines 700 minutes. The top power play unit gets 350 minutes, the second unit 200 minutes, the third unit 100 minutes and the fourth unit 50 minutes. The top penalty killing pair gets 250 minutes, the second pair 150 minutes, the third pair 100 minutes and a fourth pair 50 minutes.

o Even handed, first defensive pair is assumed to play 1,400 minutes, second pair 1,200 minutes and the third pair 900 minutes. The top power play unit gets 300 minutes, the second unit 150 minutes and the third unit 50 minutes. The top penalty killing pair gets 300 minutes, the second pair 200 minutes and the third pair 100 minutes.

o The two press box forwards are each assumed to play 450 even handed minutes and the press box defenseman is assumed to play 600 minutes (of which 25 is power play and 25 is short handed) during the injuries, slumps and trips to the coach’s doghouse of main roster players.

Goal: Tim Thomas is my starting goalie. His PC score grosses up to 232 for his slightly increased workload. This is at an all star level and not bad for $760,000. Last year Chris Mason was on this team. This year it is Dan Ellis who provided cheap ($500,000) but stellar netminding to the Predators. He would be expected to chip in another 81 PC points. These two were easy and obvious choices given their high level of performance and low costs.

Defense: There were a fair number of inexpensive performances among the best defensemen in the NHL in 208. Both Aucoin and Kaberle are about right priced for their output. However, 10 of Kaberle’s PC points came off the shootout and would be unlikely to carry forward to this team. The story was similar with Brent Burns (16 shootout points).

This was a very offensive group of players. So there is no shortage of power play talent. Kaberle was rather ineffective on the PK this year and Mike Green did not do much short handed duty. But my algorithm gave them 100 minutes of PK time.

Kaberle is the only repeat from 2007.

Forwards: My top line is Ovechkin, Datsyuk and Iginla. Ovechkin and Datsyuk repeated from last year. Datsyuk and Iginla are not cheap but they were fairly priced fpr their performance and their inclusion made this a better team. Ovechkin gets his ice time dialed way back by my model. Iginla would be my top shootout guy and that would make him the team’s most valuable skater.

Line number two, Zetterberg, Maklin and Alfredsson, is of moderate price. Alfredsson projects at 97 PC points from the second line, reflecting his PK skills.

Sharp, Richards and Boyes make a dynamite third line. Sharp would kill a lot of penalties for this team. Michalek and Pominville both repeat from last season and are

Copyright Alan Ryder, 2008 Hockey Analytics www.HockeyAnalytics.com 2008 NHL Review Page 46

joined on the fourth line by Kopitar. For the press box corps I selected Brown and Dumont.

Sharp and Alfredsson would be the team’s top penalty killers. Richards and Pominville would be pair number two.

All of these forwards can score. Nearly all could play the power play. Some of them may have to learn to play some defense for this team.

Overall: This team actually comes in over $1,000,000 under budget. This team’s very inexpensive goaltending made a bigger spend possible elsewhere. There are two pretty big contracts on this team (Iginla and Datsyuk), but these guys earned that kind of paycheque in2008. Except for goaltenders, it is only extreme performance that can justify a contract north of $5 million.

This is an offensive team (more than 50% of PC comes from forwards). This is a consequence of the availability of dirt cheap goaltending and cheap, top shelf defense. These players combined for 2,410 PC points, making this team nearly 10% better than last season’s all cap team. This might have been due to a more than 10% increase in the salary cap. But the playing time of almost all of these players needs to be scaled back and this team projects out to 1,939 PC points (or 194 points in the standings). That would not happen for two big reasons:

• There are only 164 points (1640 PC points) up for grabs!!!!

• Although winning is a linear (additive) function of individual performance over the normal range of team play, this team ain’t normal. A really good team, such as this, faces an increased headwind in its winning percentage. But this team could be a 140 point team … and a Stanley Cup shoo-in.

Sidney Crosby Watch

I would like to close by having some fun forecasting Sidney Crosby’s career. I had a look at this a year ago in a GlobeSports.com article while the Kid was on his way to the Hart Trophy. This season he had a “high ankle sprain” that kept him out of a lot of games and slowed him down thereafter.

How good is Sidney? How good could he become? The best way to answer that question is to find a group of similar players and study their performance over time.

To properly conduct that analysis it is necessary to pay attention to three large factors – age of entry into the NHL, experience and playing time. There is a huge difference between the careers of players who are prominent in the NHL in their teens and those who attain prominence at a later age. In the NHL raw skill tends to peak around the age of 23 or 24. Performance may peak at a later age as experience accumulates and ice time builds.

Copyright Alan Ryder, 2008 Hockey Analytics www.HockeyAnalytics.com 2008 NHL Review Page 47

In the case of a teenage phenom this analysis is made Sidney Crosby Peer Group easier by an earlier age of entry into the league since Player GP G A Pts raw skill development, Wayne Gretzky 1487 894 1963 2857 experience and playing time Mark Messier 1602 658 1146 1804 all tend to blossom in concert. Ron Francis 1731 549 1249 1798 Steve Yzerman 1514 692 1063 1755 But the analysis is made more Mario Lemieux 915 690 1033 1723 difficult by a limited number Joe Sakic 1363 623 1006 1629 Jaromir Jagr 1273 646 953 1599 of comparisons. The 1188 518 891 1409 comparison group I formed 1510 735 652 1387 for Crosby is comprised of Brendan Shanahan 1490 650 690 1340 players who, by their Denis Savard 1196 473 865 1338 performance, were well 1294 515 812 1327 established at an elite level of Mats Sundin 1306 555 766 1321 competition prior to the age of Dave Andrechuk 1597 634 686 1320 20. Over the history of the Pat Lafontaine 865 468 545 1013 NHL, this is an awfully small 760 372 493 865 Jimmy Carson 626 275 286 561 group of players. Sylvain Turgeon 669 269 225 494 To be included in the analysis it is also necessary to have a complete or nearly-so career. Furthermore the game changes over time and it is desirable to keep the comparison group recent. To that end most of the comparable players were born in the 1960s. Above are the 18 players I

Peer Group Performance by Age Normalized to 2008 Scoring Levels

2.00 1.75 1.50 Goals per Game 1.25 Assists per Game Points per Game 1.00 Assists per Goal 0.75 0.50 0.25 0.00 18 20 22 24 26 28 30 32 34 36 38 40 Age

Copyright Alan Ryder, 2008 Hockey Analytics www.HockeyAnalytics.com 2008 NHL Review Page 48

included in the comparison group and career summary statistics through 2008.

To use this historical data one needs to adjust for changes in rates of scoring over time. Above is a graph of the per-game scoring rates, by age, for this comparison group after adjusting scoring to the context of the 2008 season (an average of 5.44 goals and 9.41 assists per game). This graph shows the significant growth of goal scoring skills through the age of 23 and then the gradual decline thereafter. Over the course of their careers this group tended to slowly morph into playmakers (as is demonstrated by the assists-per-goal line).

It is clear from this list that Crosby is in very good company. This group averaged 1,244 games played, 568 goals, 851 assists and 1,419 points over their careers. And some of these players are still going. Is no-longer-the-kid Crosby better than ‘average’?

The table to the right shows that King Crosby has trumped Sidney Crosby versus the Peer Group the peer group handily. His injury this season pulled him GP G A Pts down to the average number Age 18 of games of his peer group Sidney Crosby 81 39 63 102 through age 20. Per game Adjusted Peer Group * 70 22 32 54 played he has scored at 134% Ratio to Peers (per GP) 150% 170% 162% of his peer group average, Age 19 assisted at 180% of the peer Sidney Crosby 79 36 84 120 group and collected points at Adjusted Peer Group * 72 25 37 62 Ratio to Peers (per GP) 134% 209% 178% 161%. Age 20 Sidney Crosby 53 24 48 72 His (per game) peer group Adjusted Peer Group * 75 29 42 71 advantage reduced in 2008 as Ratio to Peers (per GP) 119% 164% 145% a consequence of his injury (he was not the same old * Adjustments applied to bring peer averages into the scoring Sidney when he returned). context of Crosby for the year under consideration He continues to score goals at the same kind of rate he established in the first two years of his career. His playmaking, at the age of 20, is better than in his rookie year but slumped somewhat from last season. This may be a red flag. I think that injuries slow players and reduce goal scoring, but playmaking is a bit more about thinking than doing. And the peer group is getting better with age while Crosby may not be.

In 2008 Crosby’s development was slower than his peer group slowed in 2008. I think that this was the injury. As I think that Crosby is a very special player it might be useful to narrow the peer group down to one – Wayne Gretzky.

Below is a comparison of Gretzky to the peer group. To do this analysis it is necessary to compare apples to apples by translating the Great One’s achievement into today’s scoring context. When you do that you see that, although he started better, Gretzky’s

Copyright Alan Ryder, 2008 Hockey Analytics www.HockeyAnalytics.com 2008 NHL Review Page 49 development kept pace with his peers (his scoring Wayne Gretzky versus the Peer Group outpaced peer development while his playmaking lagged). GP G A Pts Age 18 Ages matters a great deal in Wayne Gretzky 79 51 86 137 this assessment. Gretzky was Adjusted Gretzky * 79 43 79 122 an older youngster. My Adjusted Peer Group * 70 22 32 54 definition of age for a season Ratio to Peers (per GP) 170% 218% 198% is a player’s age as of Age 19 st Wayne Gretzky 80 55 109 164 December 31 . Crosby’s Adjusted Gretzky * 80 41 86 127 August birthday gives him a Adjusted Peer Group * 72 25 37 62 six month performance Ratio to Peers (per GP) 152% 211% 187% disadvantage relative to Age 20 Gretzky. But this also gives Wayne Gretzky 80 92 120 212 the King some upside in his Adjusted Gretzky * 80 62 84 147 year to year development Adjusted Peer Group * 75 29 42 71 versus the Great One in the Ratio to Peers (per GP) 205% 191% 197% near term. * Adjustments applied to bring both peer group and Gretzky into the scoring context of Crosby for the year under consideration Also note that the NHL of Gretzky’s youth was the most open in modern history. Goal scoring today is about 75% of that in the early 1980s. If you look at the ‘Adjusted Gretzky’ numbers you can see that the performances of Crosby and Gretzky at the age of 19 are eerily similar. However at age 20 the Great One left the King in his developmental dust.

Just three seasons into his career it is evident that King Crosby is something special. Using the career trajectories of this peer group I have (boldly) forecast Sid’s career (see table below).

Crosby is just three years into his career and this projection may not be worth much. As a professor of mine once said, “here we are trying to estimate the weight of a dog by studying its tail”, an obviously imprecise effort. Nevertheless, this projection should elicit a ‘wow’ from the audience – a career of 1281 games and over 650 goals and 2000 points. That would have him retire as the NHL’s second highest (point) scorer. Note that my assumption that scoring levels remain at today’s low levels over the rest of his career makes this an even more impressive pro-forma career.

A big assumption in this forecast is the number of games played. Here I assume that Crosby will play as much as the average player in this peer group. So baked into this is some chance of a Sylvain Turgeon, Jimmy Carson or Eric Lindros type career, although this seems unlikely to me. On the other hand, there is a survivorship bias in this projection. Per game goal/assist rates are for those players still playing at that time. Taken together these two assumptions may understate games played (a healthy, durable Crosby could easily play more) but probably fairly project the scoring statistics.

Copyright Alan Ryder, 2008 Hockey Analytics www.HockeyAnalytics.com 2008 NHL Review Page 50

Another key assumption is that I have anchored this projection in the average of Sidney Crosby Crosby’s three seasons. This averages out Career Projection his strong 2007 and weaker 2008. If his 2008 performance was held back by recovering Age GP G A Pts from an injury this approach may have 18 81 39 63 102 seriously understated future performance. 19 79 36 84 120 20 53 24 48 72 It seems that King Crosby has a chance of a 21 73 39 78 117 50 goal season. Sid also has a god shot at 22 72 40 78 118 100 assists in a season. But my bet is that 23 73 46 86 132 24 70 39 79 118 achieving these thresholds will depend on the 25 70 40 79 119 size of goals and goaltenders going forward. 26 67 37 81 118 The NHL’s desire for more offense suggests 27 66 37 81 118 that future scoring inflation could pump up 28 62 34 73 107 Crosby’s career. Around him the Penguins 29 62 29 69 98 are only getting better (as the Oilers did 30 59 35 70 105 around Gretzky). 31 59 31 68 99 32 55 24 60 84 Since I first did this, during the 2007 season, 33 57 26 56 82 Sid’s career forecast has fallen more than 400 34 52 23 56 79 points (this is the tail of the dog problem). 35 51 23 51 74 36 46 20 45 65 His development slowed in 2008. I blame 37 42 17 44 61 the injury but could be wrong. But another 38 33 12 25 37 culprit was significant erosion in scoring in 39 0 0 0 0 the NHL again in 2008. I expect that the 40 0 0 0 0 NHL will act on this soon and Crosby’s Totals 1282 651 1374 2025 numbers will balloon.

At this very early stage in his career, I give Crosby a better than 50% chance of finishing his career behind only Gretzky in both career assists and points. Yes, Sidney is very good. The Great One is in the past, long live the King.

Copyright Alan Ryder, 2008 Hockey Analytics www.HockeyAnalytics.com