Need Stock Tips? Read Your Tweets Social scientist Johan Bollen and his fellow Indiana University researchers analyzed emotional cue words in Twitter messages during the 2008 market collapse. And, he says, his model predicted with 86.7 percent accuracy whether the Dow would go up or down.

Need Stock Tips? Read Your Tweets

Need Stock Tips? Read Your Tweets

  • Download
  • <iframe src="" width="100%" height="290" frameborder="0" scrolling="no" title="NPR embedded audio player">
  • Transcript

If you're looking for a way to leverage your investments in these tough economic times, Indiana University professor Johan Bollen has at least one investment tip you probably haven't heard before: Get a Twitter account.

Bollen studies the links between Twitter and various social indicators -- including the Dow Jones Industrial Average. He discovered that one could predict the other after assigning some graduate students to analyze millions of tweets for emotional cue words.

For example, he tells NPR's Guy Raz, "I'm very sad. I'm angry. I'm at ease."

Based on that method of mood measurement, Bollen and his students initially thought they'd have a way to observe the reaction of the Twitterverse after the stock market had moved one way or the other.

"We were expecting that if the market goes up, afterward people will be happy about it," he says. "If the market goes down, afterward people will be sad about it."

What his measurements actually revealed was slightly more interesting.

Mood analysis and market movement did overlap. But the mood dimension was not "happy vs. sad." It was "calm vs. anxious."

"And that kind of surprised me," Bollen says, recounting how a colleague first shared the results.

"And then she said, 'There's one more thing you should know. I had to shift the mood curve forward in time by three or four days.' "

That's when Bollen knew: It wasn't that the Dow Jones could be used to predict the mood on Twitter -- it was that Twitter could be used to predict the Dow Jones.

"This is astonishing," he says.

Bollen's model, he says, predicted with 86.7 percent accuracy whether the Dow Jones would go up or down. Those results came from a period of analysis lasting several months in the fall of 2008, a dark and volatile time for the market, he says. "But we decided to test it specifically for that period, because if it worked for that period, then we figured it would work for later periods as well."

So why would Twitter be a better economic indicator than more traditional economic fundamentals, like unemployment or consumer confidence?

"My best conjecture is that these user communities have become so big that a lot of investors are actually involved in Twitter," Bollen says. "And somehow they may pick up on the national or international zeitgeist.  And then -- unconsciously, perhaps -- that works itself into their investment strategies."

So will Bollen and his team make their analysis available publicly on a daily basis?

"We're scientists, so we're not interested in making a quick buck," he says.  "We're interested in what this phenomena tells us about the social and economic relations between these online communication environments and real life."