Twitter Can Predict the Stock Market
A new study finds that Twitter can be used to predict the stock market with 87% accuracy. Essentially, the “mood” of the Twitterverse is seemingly correlated to movements in global markets, with “calmness” an apparently accurate predictor of movement. The study’s authors theorize that societies are experiencing “collective mood swings” throughout the day, which influence their interest in buying or selling.
Tweet My Tastes
This study comes on the heels of earlier evidence showing that Twitter could predict box office sales and demonstrate the global “happiness level” in songs, blog posts, and presidential elections. (Among other interesting data, the latter study found the “happiness” of song lyrics trending steadily downward for the past 30 years, and that bloggers have become much happier since 2005.)
So how does it work?
The so-called “global mood” was the result of analyzing the emotional “content” of over 9.8 million tweets. From February to December of 2008, the tweets of 2.7 million users were collected to build a database of “emotional phrases”– over 70 in total– which fall into seven categories: happiness, kindness, alertness, sureness, vitality and calmness. The resulting data is pretty intriguing. For example, one emotion–”calmness”– seemed to correlate to a rise in stock market value. Except it did so three or four days in advance.
The algorithm developed by Johan Bollen and Huina Mao (both of Indiana University-Bloomington) was originally used simply to predict stock market movements. Given only the past three days of data, this “predictor” was right about 73% of the time. But when data on global emotion was added, the predictor shot up to 86.7% accuracy.
It’s obviously pretty early to start trading based on this, but Bollen is confident he’s on to something.
“ was probably one of the most difficult periods to predict,” he said. “We had a presidential election, we had what looked to be financial Armageddon, we had the start of what has been the deepest and greatest recession since the 1930s… If our algorithm was able to predict Dow Jones Industrial Average in that period, we figured that may establish some kind of lower baseline. It could do a lot better in other periods of time.”
The researchers say their next move is to test the theory by applying the algorithm in realtime.