“The Twitter Predictor”: Tweets to Predict the Stock Market
Meet Johan Bollen: Associate professor of informatics and computing at Indiana University, and the man who claims that Twitter can help predict the stock market.
Dubbed by the media as “The Twitter Predictor,” Bollen’s method for predicting economic trends is based on a formula that evaluates society’s overall mood on the social media site.
Millions of tweets from millions of users are selected and analyzed based on a series of key words related to how one might be feeling. These words are then used to measure the overall mood of the public along six different “mood states”: calmness, alertness, sureness, vitality, kindness and happiness.
Bollen and his collegues found that the correlation between this national mood and the Dow Jones Industrial Average was actually quite strong. In fact, adding the mood factor to a trained machine-learning algorithm to predict the stock market increased the accuracy from 73% to 86%.
In response to this unique discovery, Indiana University’s Research and Technology Corporation recently received U.S. Patent No. 8,380,607 – “Predicting Economic Trends via Network Communication Mood Tracking.” Bollen’s startup company, Guidewave Consulting, will also collect royalties.
It may be an interesting, and now protected, idea but not everyone is buying into the new system. Until a wider range of tweets over a larger span of time are found to have a consistent tie to the stock market’s rise and fall, some feel that Bollen’s discovery could just be coincidence.
Twitter users hail from all over the world, and it’s unrealistic to think that even the large majority have their hand in stocks, and Bollen agrees. Until more research is done and the system is tested with real money in real time, it’s hard to know just how much potential there really is here.