Using Twitter to Predict Investor Decisions
The Ohio State University
Since the stock market’s inception in the 17th century, people’s thoughts and feelings have played a part in a stock’s success in trading. Obviously, company performance and an investor’s rigorous analysis of a stock drive most valuation, but it has been proven that, especially in the short term, investors’ cognitive biases drive some decisions as well. What if an investor knew how others felt about a company? What if they could see a facet of those biases? With this kind of information, investors and companies could make more informed and profitable decisions every day. With technology today, we may have a tool that shows how people feel in regards to a company: Twitter. I ask the question: can Twitter be used to predict how an individual stock will move on a given day? Using DiscoverText, an application that collects Tweets based on keywords, I collected data on Tweets about three major corporations: Home Depot, Starbucks, and Southwest Airlines. WordStat, an application that counts words in text data, was used to code positive and negative sentiment for Tweets. SPSS was then used to develop a Time Series regression model. Results indicate predictive relationships between the stock price of a company and positive Tweets, negative Tweets, and the number of words in each Tweet. The study finds a statistically significant relationship between the sentiments, volume of Tweets, and stock price, but the relationship differs between companies. Future research needs to determine if this is because of difference in product or some other factor. Going forward, my research has the ability to play a role in larger models and allow investors to make more educated and more profitable investing decisions.
Twitter, Stock Market, Predict, Sentiment