Twitter tracker

Informatics professor Johan Bollen and colleague Huina Mao began with the premise that people feel happier when the market is up and that stock prices are predictive of the national mood. Ten million tweets and some calculus later, they found the opposite: The collective Twitter mood actually predicts up-and-down movements in the Dow Jones by four days.

Their published results made national headlines last October — Bollen even had a go on CNBC’s “Squawk Box.” In March, a London hedge fund announced its $40 million investment in the model, and the fund opens for trade in April with Bollen and Mao as private consultants.

Inside: Can I give you a scenario to see how this works? Say people are very anxious on Monday. What will the Dow Jones do in four days, most of the time? GO.

Bollen: It depends on the dimension of mood. When people get nervous, we’ve observed that the market responds three or four days later — which is quite a shock because you’d think this would happen immediately, but it’s like the Titanic. It’s difficult to turn the boat. It may take a couple days for the public and traders to process their moods and then start making decisions.

Inside: So you have all these Twitter users who are predominantly under 30, and then you have all these investors who are above 30 predominantly. How do you justify or defend that young people’s moods are affecting how older people are buying stock?

Well, we’re actually not defending that at all. We have no idea of why this works, we really don’t. But it works.

Inside: Science teachers everywhere are screaming correlation doesn’t equal causation. Is that what you’re saying about this study?

Yeah. There may very well be some mechanism through which these are related. The correlation is very high: mood from three or four days past is correlating with present Dow Jones evaluations.

Inside: So what do you think is happening?

What I think may be happening is that somehow these 30-year-olds — Twitter is heavily news-driven — are like a canary in a coal mine. It’s an early indicator of what happens and how we all feel. When the market finally adjusts, it’s very difficult for people to get an edge on the market because by then we all share the same information. So it is possible that we’ve tapped into an alternate early source of information. The other hypothesis is that traders’ moods are affected by the public’s mood. One of our recent papers investigates how mood can be contagious online.

Inside: Have you met with any of the Twitter co-founders?

Not with the co-founders, but we talk with representatives for Twitter. Right now they’re changing their data access policies, and at least for commercial applications you will no longer be allowed to access their data feeds. You have to purchase a license to the data. They have an exclusive partner, Gnip, and they license out the data – it’s not cheap.

Inside: I didn’t know Twitter could do that. Does Facebook have an arrangement like that?

No, not for the raw data, which is too bad because that would be highly valuable data. Although I think Facebook fulfi lls a different function: People post updates, but it’s not so news-oriented. It’s more a sort of communal sharing. Twitter is much more fastpaced – people sharing news and very ephemeral status reports.

Inside: What are you working on now?

Right now we’re trying to model how emotions spread through these networks. With Libya, Egypt, and Tunisia, where Twitter and Facebook played an important role, it’s not inconceivable that emotions played a tremendous role in how people got organized and how they felt about the regime. If there’s contagion of emotion, it means people can get infected with anger and all kinds of other emotions and that may play a role in organizing these kinds of protests.

Inside: Will you be able to get access to tweets from those specific geographic areas?

Only the Egypt tweets or only the Libya tweets? Yeah, it’s possible, because some of the tweets are geolocated. You can also use indicators like, “I’m at so-and-so Square right now.”

Inside: Several years ago, when the Internet was just beginning to develop, you compared communication networks to the role of a brain in an organism. Do you still think this?

I still stand by that. I wouldn’t have called it “the brain,” but at least “the central nervous system.” If you look at the globalization of our economy that has taken place, you look at information technology like Facebook and Twitter, it allows us to be part of phenomena — social, cultural, economic — that occur in different parts of the world. And to be so aware of these, that sometimes it’s difficult to determine whether it’s happening to someone else or to us.

If you look at the disaster in Japan, you can see the pain and the suffering. We’re so close to it that it affects how we feel about ourselves. I think it’s tying us all very closely together, and I think it’ll have a beneficial effect on the “world organism,” if you can look at it that way.