CC BY
47
SUSTAINABLE INFLUENCE OF SOCIAL MEDIA ON VOLATILITY OF STOCK PRICES. STATE OF ART
A. V. Dubko
Q. Boissier
The paper introduces the growing interest toward social media from investment funds. Article discusses the most vivid examples of social media influence on volatility of stock prices that stimulated the practical interest toward the problem. Also the paper reviews the existing academic researches. In the end it presents the authors' model of the factors, defining the influential power of the message in the social media.
Key words: social media, social networks, Twitter, volatility pf stock prices.
Social media for the last 10 years have become a rightful part of life. And for the generation of 2000, it will be as natural as a car or hot water. Firstly treating social media as an entertainment sites, soon many professional spheres revealed its potential: firstly, marketing and advertising, then recruiting - nowadays recruiting through social media, allowing the HR to reach passive candidates is becoming more and more popular. Then there was a number of reports on the role of social media in growing and “heating up” the protest movements. (Gonzalez-Bailon et al., 2011). Now it is the turn of finance. The critical condition for social media to be influential is number of people engaged. As soon as critical mass of professional investors and economists started sharing their thoughts through social media, it began to affect the market. According to Colt Technology Services, in 2013 about 45 percent of financial-services workers regarded social media as a trailing indicator and used them to validate their trading decisions (Sukumar N., 2013).
Structure of social media allowing message to go viral in an hour is in line with long-discussed danger of high-frequency traders who gain tiny profits for milliseconds and thus employing fast and almost autonomous machines that rely on algorithms and lack critical thinking. First significant events revealing potential influence and potential threat of social media in relation to financial sphere happened in 2013. Hacked Twitter account of Associated Press shared the tweet about injury of Barack Obama as a result of the explosion in the White House. Message was instantly re-tweeted, reaching 4000. Although the hack was soon revealed and blocked, it caused shock waves through the market making S&P500 decline 0,9%. (Matthews C., 2013). Another vivid example happened a few months later, when famous investor Carl Icahn wrote in his twitter: “We currently have a large position in APPLE. We believe the company to be extremely undervalued. Spoke to Tim Cook today. More to come.” And after that stock price of Apple moved from $475.76 to $494.66 for just one hour (Pressman A., 2013).
However, academicians started researching the topic earlier. In 2011 one of the first researches on the topic appeared devoted to predicting the stock price changes by the tweets of common people. Assuming that market movements depend greatly on the people’s expectations, researchers analyzed vast layer of tweets, then they allocated the dominating expectation of the majority of people and compared it to actual market movements on the next day. They managed to
490
find significant correlations. (Zhang X. et al., 2011). In the same year Bollen et al. also found that the public’s mood on Twitter can predict the Dow Jones Industrial Average. (Bollen et al., 2010). In 2014 Sul, Dennies, and Yuan collected all Twitter posts that mentioned S&P 500 firms and claimed that positive or negative expectations or statements about a firm significantly correlated with that firm’s stock returns. (Sul et al., 2014). While most of researches applied overall approach: taking general data and looking for mentions of firm stock tickers, names and keywords, in 2015 appeared one of the first papers that used firm-specific twitter metrics to predict homogeneous groups of stocks that have similar movement of returns. (Liu L. et al., 2015).
There also can be a different approach - not through the analysis of mass opinion, but through the analysis of the posts of key figures that actually form the mass expectations. As we could see in real-life examples, in both cases major changes in market were caused by single initial source whose message later was shared by others, but what determined the changes was the one single trigger.
Now it is important to briefly observe key factor groups that define the influential power of social media message. We can single out three groups of factors: (1) factors of the recipient of the message; (2) properties of the technology; (3) properties and qualities of the source - the sender. First group includes personal traits of the recipient that define how easily he is influenced by the information from social media. Here we include (a) risk-taking - general readiness to take risk and participate in activities with high level of uncertainty; (b) propensity to trust - basic level of trust that is formed in early childhood and serves as base for local trust and (c) locus of control - an extent to which individuals believe they can control events happening to them. As we can observe, very different people are involved in the cascading processes of information dissemination, so we can conclude that this group of factors is not the most significant when we are studying mass processes.
Second group refers to the technological differences of social media. Here the most important are (a) safety police: social networks differ greatly on safety and privacy settings. In business networks (e.g. LinkedIn) an attempt to add a friend is a complex procedure that demands efforts and limits fraudulent actions. At the attempt to add a complete stranger to contact list - the initiator will face number of trustworthiness checks long before the person he is trying to get in contact with will even know about these attempts. On the contrary, in VKontakte adding a friend procedure is a simple choice between: “Add/ Don’t add”. Such choice is often made impulsively, by influence of situational factors. Such difference in privacy and safety measures leads to differences in trust formation to the messages arising inside the networks; (b) available instruments of information sharing. For example, some networks allow uploading audio, while others not; (c) algorithm of information sharing - networks differ in what types of information they by default include in the user’s news feed. For example, Facebook’s news feed reflects not only uploads and actions of the friends, but also everything that they liked or commented, while Vkontakte - reflects only direct actions. The key point in influencing stock prices is speed. From this point of view, the best instrument of existing social networks is Twitter. It provides minimum information, it is unified that makes easier the mechanical processing. According to some professionals, it is not even a social
491
network, but informational network (Ben Parr, 2010), as it focuses not on social ties but on information dissemination. While difference in number of users is not critical, Twitter for now is the perfect instrument for emergence of cascading behaviors.
Finally, the third and probably the most interesting for us group of factors -properties and qualities of the source - the sender. (a) Verifiability and trustworthiness of the source. In the earlier example, Associated Press (AP) was a trustworthy source which news, even such incredible, at first were treated with some credit of faith. (b) Properties of the message itself - structural and semantic features that makes one message more likely to be shared than another (Лепехин
Н.Н., Дубко А.В., 2011). (c) Influential power of the source, that is defined by (1) size of personal network; (2) activity and loyalty of the followers - only number of followers is not enough if the followers are not active participators and are not loyal or interested enough to re-share the message of the initiator. Otherwise, the message will die in the first circle of friends while power of cascade is in re-sharing. (3) Density of the network. The more close and strong are ties with followers - the more likely they share the same interest and consequently are more likely to share the information further. (4) Position in the network. The source is located within tightly-interconnected network or has also large number of weak ties granting him access to other dense communities and increasing the chance that his message will reach vast number of people. Returning to our example, if AP wasn’t a popular source, it won’t cause such effect as time was crucial - hack attack soon was revealed and blocked, so only its position within diverse network let its message to go viral before counter-actions.
Now let us explain what is the practical implication. In portfolio management we can distinguish two kinds of managers: active and passive. The latter refers to managers whose goal is to replicate the performance of an index whereas active managers select asset classes and securities to include in their portfolios. The performance of active managers has two components. The first one is security selection, which refers to their ability to value and select securities to add in their portfolios, and then their strategy can be for example to short sell overvalued securities and being long on undervalued ones (arbitrage or market neutral strategy), or to overweight the stocks with the highest risk-adjusted return in an equity long-only strategy. The second component is called market timing management, which is the ability to anticipate market trends and to thus modify the portfolio risk exposure by increasing or decreasing the beta of the portfolio. Beta referring to the risk arising from the exposure to general market movements.
Alpha that can be seen as the added value allows to evaluate and compare fund managers’ performance. It refers to the outperformance (or underperformance) compared to a benchmark such as an index. Hence, passive managers will have an Alpha equal to 0 because the performance of their portfolio will be derived from an index, whereas active managers will perform differently from the benchmark.
Social media analysis is one mean to improve the alpha of managers by giving them information in incredibly fast way to share information thus enabling fund managers to anticipate even more faster hence reducing time decision and enhancing market timing management. Thus, since its inception, social media analysis has been widely and quickly adopted by the investment industry mainly by
492
hedge fund managers in order to obtain or improve the Alpha in their investment strategies. Growing interest to the topic both from academicians and practitioners proves potential impact of social media to the stock market. We believe, that when tested, our model will help to reveal existing correlations between opinions of key figures expressed publicly and stock prices movements. That is a good example of sustainability and globality of the world processes.
Sources
Bollen, J., Mao, H., & Zeng, X.-j.. Twitter mood predicts the stock market. Journal of Computational Science, 2(1), 1-8. (2010).
Gonzalez-Bailon, S., Borge-Holthoefer, J., Rivero, A. &Moreno, Y. The Dynamics of Protest Recruitment through an Online Network. Sci. Rep. 1, 197; DOI:10.1038/srep00197 (2011).
Liu L., Wu J., Li P., Li Q. A social-media-based approach to predicting stock comovement / Expert Systems with Applications 42, 3893-3901 (2015).
Sul, H. K., Dennies, A. R., & Yuan, L. I. Trading on twitter: The financial information content of emotion in social media. In: Hawaii international conference on system sciences (HICSS) (p. 10) (2014).
Zhang X. Fuehres H., Gloor P. Predicting Stock Market Indicators Through Twitter “I hope it is not as bad as I fear”/ Procedia - Social and Behavioral Sciences 26, 55 - 62 (2011)
Лепехин Н.Н., Дубко А.В. Доверие к виртуальной идентичности в Интернет-среде // Вестн. С.-Петерб. ун-та. Сер. 12. 2011. Вып. 4. С. 145-151.
Matthews C. How Does One Fake Tweet Cause a Stock Market Crash? (2013) // http://business.time.com/2013/04/24/how-does-one-fake-tweet-cause-a-stock-market-crash/ Pressman A. Carl Icahn's Multibillion-Dollar Tweet Boosts Apple Stock (2013) // http://finance.yahoo.com/blogs/the-exchange/carl-icahn-multibillion-dollar-tweet-boosts-apple-stock-205938760.html
Sukumar N. Twitter, Facebook Influence Share-Price Moves, Study Says (2013)
http://www.bloomberg.com/news/articles/2013-02-12/twitter-facebook-influence-share-price-
moves-study-says
Parr B. Facebook, Twitter and The Two Branches of Social Media (2010) // http://mashable.com/2010/10/11/facebook-twitter-social/
493
