Thursday, February 21, 2013

IU prof patents his 'Twitter predictor' for stock market swings | Indianapolis Star

IU prof patents his 'Twitter predictor' for stock market swings | Indianapolis Star | indystar.com


Written by
Stephanie Wang



Johan Bollen's 'Twitter Predictor' is a formula for gauging society's mood based on Twitter or Facebook posts.
Johan Bollen's 'Twitter Predictor' is a formula for gauging society's mood based on Twitter or Facebook posts. / Indiana University photo
Millions of your 140-character tweets — that mindless drivel and those snippets of wisdom — have already predicted the bobbing of the stock market.
Soon, they could divine the onset of the flu or the movement of politics.
Since his 2010 research article set people atwitter, Indiana University researcher Johan Bollen has broadened his formula for gauging society’s mood based on Twitter or Facebook posts. This week, he won a patent for what many have dubbed “the Twitter predictor.”
What you ate for breakfast? Your favorite sports team’s big win? That funny thing your cubicle neighbor just said? Crank it all through a formula and the tweets can hint at society’s collective mood of the moment, based on certain words about how you feel.
“Public sentiment is a pretty ghostly concept,” said Bollen, an associate professor of informatics. “You’re trying to quantify something that maybe people feel isn’t inherently qualitative.”
Other researchers have examined tweets to determine how happy people are in different cities or states. Marketers have looked at tweets to understand whether people like their products.
What Bollen found is that these moody tweets can be used to peer into the near-future, as indicators of what could happen socioeconomically or financially.
The first test tied tweets to the Dow Jones Industrial Average. Bollen’s research with doctoral student Huino Mao claimed close to 90-percent accuracy in predicting the stock’s short-term changes. When people seemed calm, the Dow went up. When they became anxious, it sank.
The recent patent brings his startup, Guidewave Consulting, closer to selling the invention for wider use. IU’s Research andTechnology Corp., a Guidewave shareholder, can collect royalties from sales.
But not everyone buys into it.
“I’m still somewhat skeptical,” said Mark Foster, chief investment officer for Kirr Marbach & Co. in Columbus, Ind. “The short term is sort of a random event.”
“It could be a coincidence,” said Bill Wendling, chief investment officer for Indianapolis-based Bedel Financial Consulting. “In real live testing with actual money, it’ll take a couple of years to find out whether it’s a valid option or not.”
Both investment experts say the invention is better geared toward day trading or hedge funds, as opposed to longer-term investments.
Bollen said two hedge funds have tested the Twitter predictor. He acknowledges a risk but uses this analogy: “If you’re a farmer, we’re not telling you how to grow corn. We’re telling you what the weather will be like.”
The next step is to steer the data analysis to examine what drives certain feelings. It’s easy to see when people aren’t happy, Bollen said, “but the big challenge is to explain why.”
Star researcher Cathy Knapp contributed to this report. Call Star reporter Stephanie Wang at (317) 444-6184

.

No comments:

Post a Comment