International Journal of Pure and Applied Mathematics Volume 118 No. 22 2018, 941-946 ISSN: 1314-3395 (on-line version) url: http://acadpubl.eu/hub Special Issue ijpam.eu
GOOGLE TRENDS AND STOCK RETURNS – A STUDY OF INVESTOR SENTIMENTS USING BIG DATA
1Hari Krishnan.A.V, 2Gopakumar.V, 3Varsha Sureshkumar School of Business, Amrita Vishwa Vidyapeetham, Coimbatore. [email protected], [email protected], [email protected]
Abstract: This paper tries to check for the presence or capitalization in the exchange. The adjusted daily absence of a relationship between the individual returns closing price for all the 50 stocks of the Nifty50 index of the stocks in NIFTY50 and Google trends data for was extracted for a period of 5 years starting December the word ‘buy ’. We have employed 2012 and ending December 2017 from the NIFTY Granger causality test to investigate the association website. This was then averaged into the weekly between Google Trends data and individual returns. For closing price for analysis. the study, the adjusted closing prices of Nifty 50 shares for a period of 5 years is used to compute the weekly 2. Literature Review stock yield and the Google trend data from India for the keyword ‘buy’ is taken. It was Reference [1] examined the influence of information observed that the returns of some individual stocks contained in news on the stock market, by sourcing showed a bi-directional causality while some showed a news articles from Lexis Nexis, a searchable database unidirectional causality. Among NIFTY 50 stocks, of digitized print media. Results showed that the many did not show any causal relationship between the content in the recording revealed a substantial weekly returns and the relevant Google Trend keyword: connection with the stock yields. ‘buy’. Reference [2] applied the technique of collecting data from social media platforms like Facebook, Keywords: Google trend data, Unit root test, ranger Twitter and Google. The study performed a sequence causality test, Buy, Nifty 50 returns. analysis and transmission mechanism of how social media is a good reflector towards the capital markets. It 1. Introduction also put forth the relevance of reach of data from social media as a tool in monitoring the society’s behaviour. Investor sentiment acts as a good indicator for other Reference [3] tested for the presence of any investors for trading in stock markets traditionally. relationship between online search intensity and stock With the advent of the Internet and easy availability of trading behaviour. The test was carried out in Japanese information, market sentiment has recently been stock markets for 189 companies. All of them were observed online too, where investor sentiments are Japanese stocks, which were searched from 2008 to studied based on the search keywords about companies. 2011 in google.com. The conclusion drawn was that the The keyword ‘buy’ is usually used along with a relationship between search intensity and stock price is company name in websites like google.com when a positive but weak. But the relationship between search person searches for information, indicating an interest intensity and trading volume was positive and strong. to buy stocks of that company. Prospective investments Reference [4] examined if stock returns could be decisions are usually taken with the help of technology. forecasted by using data from Google Trends. Previous Google search is one such major tool for collecting studies stated that the returns were high for the first 1 to such data. The data from Google search is aggregated 2 weeks with Google search volumes, followed by a and available as Google Trends, which is an open- subsequent price reversal. The dataset used for this source and searchable database. By using the search study pertains to 2008 to 2013. The key conclusion of keyword ‘buy’ along with the company name, data is the paper was that higher Google search volumes for a extracted from Google Trends for a period. We then particular stock yielded negative returns. The paper test for any possible relationships between the word also attempted a trading strategy by selling the stocks ‘buy’ and Nifty50 stocks. with higher search volumes and buying those with less NIFTY is a broad-based index representing 50 frequent Google search volumes. large and liquid stocks on the National Stock Exchange, Reference [5] tested the relationship between India. Nifty50 stocks were chosen for the study because Twitter feeds and financial markets by collecting data it represents more than 60% of free float market on the volume of data churned out and performing a
941 International Journal of Pure and Applied Mathematics Special Issue
sentiment analysis for 30 companies from the Dow Causality test, a statistical technique used to predict the Jones Industrial Average (DJIA) index for 15 months. relationship between variables using time series data. The study found that Twitter sentiment had a positive One prerequisite for using Granger Causality test is that relationship with the stock returns when the volume is the data has to be stationary. So using unit root test, at peak. It also identified that Granger Causality and weekly returns of Nifty50 and Google Trend data are Pearson Correlation were lower in the time-series for tested for statistical stationarity. The unit root test is the given sample time frame. directed to find whether the dataset is stationary or not. Reference [6] presented an overview of the Statistical stationarity implies that the properties of mechanisms operating between social media tools and statistics (mean, variance and autocorrelation) are the capital markets explaining the rationale using perpetual for the given particular period of time. theories of behavioural finance. It also contemplated on Equation (1) and (2) explains the causal relationship less rational factors (investors’ sentiment/public mood) between ‘Buy’ and ‘NReturns’ as a powerful indicator of asset pricing and capital = market volatility. ∑ ∝ + ∑ + Reference [7] identified the emergence of cyclical (1) patterns, as a result of linearity. The NIFTY data was = ∑ ∝ + used from 1994 to 2013.The study found a cyclical ∑ + (2) pattern where Indian stock markets were observed to
have shifted between periods of efficiency and Where, inefficiency, and settling to becoming efficient from BUY= ‘buy’ search term used in 2003. Google trends Reference [8] analyzed changes in the volume of NRETURNS = Nifty 50 index weekly returns search terms in Google. The findings were patterns that can be interpreted as ‘early warning signs’ of stock , , , = coefficients of the model (i.e., the market moves. contributions of each laggedobservation) Data from January 2004 and February 2011 was , = residuals (prediction errors) for each time used. It was found that adding huge behavioural data series. sets (financial trading and search query volumes) can bring up new understandings into the different phases 4. Hypothesis of large-scale collective decision making Reference [9] used Granger Causality to The subsequent hypothesis was formulated which is comprehend the association between stock returns and tested using Granger Causality test. selected macroeconomic variables like oil prices and H01a: GT does not Granger cause RET exchange rates from India and US for the years 2000 to H01b: RET does not Granger cause GT 2015. It was found that the GDP growth rate of India and USA are significant predictors of NIFTY returns. Where, Similarly, reference [10] used Granger Causality to GT = ‘buy’ search term used in study the relationship between exchange rate returns Google trends and stock indices returns. The paper established that the RET = Nifty50 individual stocks weekly returns indices (commodity, IT, MNC and energy) significantly caused exchange rate returns. Figure 1.
3. Methodology 50 40 The study has taken weekly closing data of Nifty 50 30 stocks for the period starting December 2012 to 20 December 2017.Weekly index returns of Nifty50 individual stocks were then calculated by the following 10 log normal formula 0 = ln ( / ) 12131313141414141515151516161617171717 Where, = Nifty 50 index closing price of present Bosch HCL Technologies week