“This strategy will not be helpful in predicting the direction of particular stocks,” said Julie A. Chandler, a securities lawyer with K&L Gates in Dallas. “Any information that hits social media is already public, and therefore the market will have already accounted for that information and the price of the stock in question will already reflect that information. In other words, by the time you learn of the information, it will be too late to get a ‘leg up’."
That is not, seemingly, deterring fund managers at Derwent Capital Markets in London, which launched e £25 million hedge fund based on research by Johan Bollen, a professor of informatics and computing at Indiana University. By sampling 10 percent of all Tweets and comparing negative versus positive comments, Bollen’s program was able to predict the daily close of the Dow Jones Industrial Average with 86 percent accuracy.
And Twitterers seemed to buy the notion.
“When people are greedy,markets go up. When fearful they go down,” Rich Rocks tweeted.
“I Tweet, Therefore I Profit,” Kerry Yi of New York City said in a Twitter post.
At StockTwits.com, there’s an entire community of investors who use Twitter to trade ideas and track news.
If only it were that simple. Bollen’s program relies on some rather complex analysis, and also uses a program developed by Google to rate whether individual tweets can be classified as calm, alert, sure, vital, kind, or happy.
The new hedge fund is also more than likely relying on its large size. Having more capital makes it easier to hedge investments.
But Twitter traders may fall into the same trap the Chicago Tribune’s editors did in 1948, when they printed a headline that declared “Dewey Defeats Truman!” Harry Truman won the presidential election that year. That prediction was made on the basis of telephone polls which proved inaccurate, because polling methodologies didn’t take into account biases in the sample of households with telephones. What’s accurate now may become inaccurate over the time—especially if Twitter’s demographics shift, or Twitter changes the data it makes available in its feeds.
Alan Pak, who describes himself as an “algorithmic quantitative trader” said that while the information on Twitter and other social networks is public, it comes out fast enough to make a difference for investors. Pak said the key is to figure out the correlation between events on Twitter and events in markets.
He gave the simple example of “volcano in Iceland = #icantfly hash tag posts on twitter = short airline stocks.”
“It seems obvious written this way, but by data-mining Twitter, you can draw those conclusions faster and on a larger scale,” he said.
Whether or not the strategy will work does not diminish that more people are tying to use social media to predict the future.
In the 2010 election cycle, Globalpoint Research analyzed a wide stream of data from blogs, Twitter and Facebook, and was able to predict the Kentucky primary race for a U.S. Senate seat to within a single point of the actual result.
Wall Street fortunes are made on being just a few degrees more right than the rest of the market. And any edge will do. For now, Twitter seems pretty edgy.
Photo by Helico