They might be only 140 characters long, but the mood swings captured in those tiny “tweets” can predict the movement of the stock market, says a new study. Measuring Twitter’s “calmness score” can predict how the Dow Jones Industrial Average will change three days later – with 86.7 percent accuracy.
“We were pretty astonished that this actually worked,” Johan Bollen, a social computational scientist from Indiana University, Bloomington, told Wired Magazine.
Bollen and a graduate student stumbled on what Wired calls a “computational crystal ball” while analyzing Twitter feeds for mood data. Bollen’s original goal was to build a nuanced barometer of America’s mood (he found that the West Coast is happier than the East coast, to the surprise of no one).
To that end, the researchers beefed up a standard psychological test called the Profile of Mood States, and applied the test as if Twitter were a patient in a psych ward, analyzing 9.8 million tweets from 2.7 million tweeters between February and December, 2008. The test picks out emotionally-charged words and adds each word’s score to one of six dimensions: calmness, alertness, sureness, activity, kindness and happiness.
The method was working smoothly in test runs. Then, just for kicks, they compared the national mood to the Dow Jones Industrial Average. They found that calmness lined up astonishingly well with the rises and falls of the stock market – but three or four days in advance? No one knows why. Many tweets come from teenagers who are unlikely to play the stock market. Still, factoring in tweeted emotions made a prediction algorithm 13.4 percent more accurate.
But it’s probably too early to drop out of school and invest everything based on Bollen’s test: these results only cover 11 months in 2008. Better finish that degree in economics, just to be safe.
– Emily Goyins