IU prof creates, patents software that uses Twitter to predict stock market performance

Johan Bollan now trying to sell software

An Indiana University professor says he knows how to use Twitter to help predict the stock market.

Professor Johan Bollen has patented his software and is planning to get it into the hands of industry professionals.

It started two year ago with a basic algorithm that tracks every tweet in Twitter nation, compiling and categorizing them based on their emotional content.

"If I post something meaningful like, 'My car broke down. I hate this. This really sucks,' then a computer looks at that and finds individual words, phrases," Bollen said.

Bollen said his software takes those words and phrases and uses them as social and economic predictors of the stock market.

"My student and I were sitting in my office and I told her this is huge. This is big. I mean, you're telling me that this data actually predicts the market," Bollen said.

Bollen said the accuracy of his software varies, but certain tests have shown it to be 86 percent accurate in predicting the Dow average.

"A very large hedge fund looked at it and in their projections they could have significantly outperformed the market," he said.

Now that the technology has been patented, Bollen will increase efforts to sell the data to industry professionals.

"We're always looking for more revenue and more clients, but I think the demand is definitely there," he said.

The money made from this software will be split three ways among the university, Bollen and his graduate assistant.

Print this article Back to Top