How To Predict Stock Market Trends Through Social Media
Social media is a vast space, and predictive analytics has proven to be a useful resource when trying to decipher seemingly unpredictable stock market trends from social media trends. Whether you’re trying to figure out the US stock market trends or general stock market trends in 2015, social media might be an important criteria which needs to be factored into your investment checklist.
A study conducted in 2011 by a PhD student from Pace University showed how popularity on social media can forecast stock market prices on a daily basis. The study was done along with Famecount (now called Starcount), and took a close look at three big brands – Nike (NYSE:NKE), Starbucks (NASDAQ:SBUX), and Coca Cola (NYSE:KO) for 10 months in 2010 and 2011. The number of fans, followers, and video views were compared against stock price changes for each brand, in relation to consumer stocks—which acted as an indicator of the general status of the stock market. Keyword search trends were also factored into the study, and as a result a connection emerged between the stock prices for these brands and popularity on social media on a per day basis. This pattern was apparent even when a delay of 10 to 30 days was brought into the study.
Even though the study mentioned above may have provided only a microscopic view of the relationship between social media trends and stock market fluctuations, stock traders are using these strategies like text analysis and sentiment analysis to understand the stock market better.
Sentiment and Text Analysis
Sentiment analysis looks at whether the tone of text is negative or positive and text analysis looks at the meaning of a chunk of text. The information gathered from these techniques have been used as actionable data for algorithmic trading, and this type of automated trading accounts for almost half of the trades that take place in USA. Text and sentiment analysis comes in handy when you’re trying to understand large data sets related to a particular stock. It helps convert text into numbers so that the data can be used in a trading system.
If a company announces a change in one of their products and there has been a statistical connection between stock price rise and related news at an earlier period, then the trading system will take action and buy shares.
The data provided by these two types of analysis also brought in fresh information to the trading system, since it was independent of other factors affecting stock market seasonal trends. Contrary to popular belief, high positive sentiment levels may not always be the most useful for traders. It is a stock with average positive sentiments that usually go unnoticed and which can reap benefits when you buy them.
Twitter and the Stock Market
Predictive analytics has been useful in predicting the frequency of trading and the stock price for the next day, based on data from Twitter. This model was developed by a professor at the University of California, Riverside and other researchers. A trading strategy supported by this model was created by three researchers at Yahoo!, an associate professor at Bourns College of Engineering, Vagelis Hristidis, and his graduate student. Their trading strategy performed 1.4%-11% better than other baseline strategies. During a 4-month simulation, this strategy also outperformed the Dow Jones Industrial Average.
Hristidis’ study looked further than the effect of negative and positive sentiment on stock prices, it looked at the quantity of tweets and the interrelationship between tweets, users, and topics. The professor did point out flaws in the study, but the model they developed lost an average of 2.4%, while the random model lost an average of 5.5%.
Facebook and Stock Market Trends
Arthur J. O’Connor used to work in risk management on Wall Street and he went on to write a paper that looked at the relationship between Facebook popularity and consumer brand stock prices. He identified 30 brands that had the maximum number of followers, and tracked the likes these companies got for a period of one year as well as their share price on a daily basis.
The premise of his study aimed at finding out whether popularity (measured in Facebook likes) affected performance (share prices). O’Connor found that 99.95% of the changes in share prices could be explained by the change in the number of fans. He did not see a direct relationship between likes and an increase or decrease in share prices, but stock market trends were affected by the appreciation a company got on social media. The majority of a change in a particular company’s stock price was linked with the likes that brand received for that period or day.
Social Media and Stocks
These studies have prompted people on Wall Street to use sentiment feeds in their existing algorithms that trade stocks. In 2011, WiseWindow—a social market analytics company—started a service that forecasts stock market trends supported by consumer sentiment conveyed on message boards, blogs, and social media sites. They provided this as a Bloomberg data feed. The increased predictive value was apparent in the airline industry and Emerald Logic, an advanced analytics consulting company, said that using the data provided by WiseWindow improved returns by more than 30% on an annualized basis.
Social media sentiment was just one of the variables that was factored into this analysis. Note that the data is most applicable for stocks that largely value customer sentiment and that it only acts as a filter on the results that other predictive models produce.
WiseWindow and Reuters.com share a strategic relationship that runs a real-time sentiment index. The index is called Social Pulse and it brings together the sentiment and volume of a particular company’s mass online opinions to compare sentiment shifts and stock price percentage changes. Social Pulse looks at the major companies that are traded publicly from finance, technology, industrial, energy, cyclicals, and healthcare sectors. WiseWindow’s Mass Opinion Business Intelligence™ (MOBI) is another useful tool that examines online conversations to create a consumer sentiment score to each company in the index. The scores are revised every hour and acts as an indicator of stock performance.
Deep research has to go into stock market analysis. Social media sentiment is just one factor. An investor must always look at a company’s fundamentals, be up to date on the news that can affect stock prices, and avoid some of the common investing mistakes that people usually tend to make.