NEW YORK (Reuters) - When Richard Peterson first started meeting with hedge funds about eight years ago to pitch using social media to predict market movement, investment managers looked at him as if he had just arrived from outer space.
Back then, what he was pitching them seemed pretty insane. Peterson, managing director of Santa Monica-based MarketPsych, said that social media can be mined for data about what people are thinking and feeling. And that, in turn, could translate into powerful investment ideas.
“People would say to me, ‘You’re crazy,'” says Peterson, who did postdoctoral studies in neuroeconomics at Stanford University. “‘You’re a psychiatrist telling me that funds should analyze social media? Come on.’ They didn’t think I was serious.”
They’re taking him seriously now. Usage of social media like Twitter has exploded in recent years, giving analysts a real-time reflection of popular sentiment. As a result, MarketPsych serves up reams of data to hedge funds (which swear Peterson to secrecy) and research firms like Titan Trading Analytics. Peterson even plans to roll out a hedge fund of his own.
“We’re champing at the bit to start trading,” says Peterson, who says his models work best in times of high volatility. “We’ve run simulations to see what would have happened by using our data in recent years, and we would’ve made 30 percent annually.”
Given the amount of irrelevant nonsense on Twitter, it’s natural to be highly skeptical of the strategy. The vast numbers of spambots, penny-stock touts and Justin Bieber fanatics aren’t helpful in generating any investment gains.
But think through the logic, and analyzing Twitter data isn’t such a bizarre idea.
A basic premise of behavioral economics is that the markets aren’t perfectly rational machines, but are expressions of human emotions like greed and fear. If you agree with that premise, and are looking for an immediate gauge of those human sentiments, then Twitter is one of the greatest tools ever invented.
“The importance of social media aggregation, and how that might influence the price of a stock, cannot be ignored,” said John Coulter, CEO of Atlanta-based Titan Trading Analytics, which uses MarketPsych’s data. “We’ve chosen to use it as one of many indicators, providing traders with alerts on events and by flagging socially expressed emotions which haven’t been picked up upon by traditional news outlets.”
The trick is how to crunch that data effectively and make some sense of the 250 million tweets generated every day. Peterson, for example, filters the data using 1,500 different factors, culling keywords to track global moods. His is essentially a contrarian take on the markets: If the public is overly bullish, it’s time to be cautious. If it is extremely gloomy, on the other hand, it might be time to snap up a bargain.
In that sense, it's much like how some investment pros look at the American Association of Individual Investors' sentiment readings as a contrarian indicator (link.reuters.com/dap66s).
But while those respondents answer a survey for a once-a-week reading, social-media sentiment analysis is immediate and ongoing.
Indeed, the Twitter-analysis trend seems to be just gearing up. Cayman-based Derwent Capital Absolute Return Fund Ltd., dubbed the first ‘Twitter Hedge Fund’ with $40 million in seed capital, was reported to have beaten the S&P by more than three percentage points in its first month of trading last July. More recent results were not available.
“It won’t make you a millionaire overnight, but it does work,” says Richard Gardner, president and CEO of Scottsdale, Arizona-based Modulus Financial Engineering, which amasses historic Twitter data for hedge funds and research firms to crunch. “The markets are moved by emotion, and I think this is going to be the future of trading. You can actually see global moods moving up and down in real time.”
Much of the excitement around Twitter trading stems from a paper by academics Johan Bollen and Huina Mao of Indiana University, and Xiao-Jun Zeng of the University of Manchester. The report found that gauging the investing public’s mood can be a startlingly predictive mechanism for the stock market. “We find an accuracy of 87.6 percent in predicting the daily up and down changes in the closing values of the Dow Jones industrial average,” the authors wrote.
Before you start examining your own Twitter feed for brilliant investment ideas, though, take a deep breath. It’s one matter for quant funds, with their highly complex mathematical models and armies of analysts, to be diving into social-media data and gauging the global mood from billions of tweets. It’s quite another for individual investors to think they can make any sense of the collective bleatings of 100 million active Twitter users.
“On Twitter, the vast majority of accounts aren’t even verified,” warns Jake Wengroff, global director of social media for consulting firm Frost & Sullivan. “If you’re going to mine Twitter data for investment ideas, at least compare it with other trading models that you’ve already built. Solely looking at Twitter signals would not be a good decision.”
As for MarketPsych’s Peterson, he’s glad that investment managers no longer look at him like he has three heads. But the downside for him is that more and more informatics wonks are crowding into the space, trying to unlock a workable trading strategy from the billions of tweets out there.
His advice? “Don’t do it. I don’t need any more competition.”
Editing By Jilian Mincer, Beth Pinsker Gladstone and Dan Grebler