Separating the Market-Moving Tweets From the Chaff

NOVEMBER 11, 2013, 7:29 AM

By WILLIAM ALDEN

  Peter Wynn Thompson for The New York TimesFILTERING TWITTER’S FIRE HOSE Joe Gits, the president and chief executive of Social Market Analytics, at his company’s office in Chicago.

 Carl C. Icahn wields it as a megaphone. Traders and analysts use it to chat publicly about stocks. Eyewitnesses employ it to post messages about news events before they make headlines.

Wall Street has undoubtedly recognized the value of Twitter as an investing tool. But a question remains about how best to harness the stream of half a billion messages that flow through Twitter daily — and use that information to gain a market edge.

A quest to mine Twitter for investing gold, which began several years ago in universities and start-up companies, is entering the financial mainstream. This week, the New York Stock Exchange plans to introduce a faster version of a service that gives investors a window into how particular stocks are being discussed on Twitter, complete with “sentiment” scores that reflect whether the chatter is positive or negative.

That effort, a partnership with a start-up called Social Market Analytics, is among several social data projects making inroads in an industry that has kept Twitter and Facebook at arm’s length. Faced with strict compliance requirements, employees of firms like Goldman Sachs and Morgan Stanley are blocked from accessing these sites on work computers, though many discreetly check Twitter on their smartphones.

“Every time I went to a buy-side firm, they would ask about social media,” said Joe Gits, the president and chief executive of Social Market Analytics. Working with the Big Board “is a big credibility boost for the whole concept of social media and trading,” he added.

Some computing experts are skeptical. Ilya Zheludev, a doctoral student in the financial computing department at University College London, who has studied whether Twitter can lead the market, said he had concluded that, for the most part, not enough data exists to reliably trade on signals gleaned from Twitter messages.

“It’s somewhere in between being able to print money and having nothing at all,” Mr. Zheludev said.

A boutique hedge fund in London, Derwent Capital Markets, started a fund in 2011 to trade on trends emanating from Twitter, licensing software from Indiana University, but closed it after just one month.

And yet social media is rapidly becoming a feature of life on Wall Street. In April, after companies like Netflix were already making use of social media, the Securities and Exchange Commission set rules for how executives could use those outlets to send messages that would be of interest to investors. That same week, Bloomberg L.P. began incorporating Twitter messages into its data service.

Twitter’s initial public offering has focused attention on the company’s business of selling data. A select group of companies, including Gnip, DataSift and Dataminr, buy data directly from Twitter, selling it to other analytics companies or using it to produce analysis for clients. Twitter generated $47.5 million in revenue from licensing its data last year, an increase of 66 percent from the year earlier, the company said in its I.P.O. prospectus.

The start-up favored by the New York Stock Exchange, Social Market Analytics, which was founded last year, tracks about 400,000 Twitter accounts that are deemed relevant to the market, using data from Gnip. After scanning for words and phrases that reveal sentiment, the service produces scores that show how the chatter about a company’s stock is changing in a positive or negative direction. Traders can receive email alerts and track the data through a N.Y.S.E. communication system.

Mr. Gits said Social Market Analytics alerted subscribers to unusually positive sentiment in Apple’s stock before Mr. Icahn said via Twitter on Aug. 13 that he had accumulated a stake in the company. “A lot of times, you’ll see rumors breaking here first,” Mr. Gits said.

But even if it does have a measure of predictive power, Social Market Analytics will lose its usefulness over time as a wide pool of investors incorporates the company’s information into stock prices — a point acknowledged by Thomas Watson, vice president for global market data at the exchange.

Before signing a pact with Social Market Analytics, the exchange talked with Dataminr, a four-year-old company it was introduced to by Twitter, Mr. Watson said. Dataminr, which has raised $46.5 million from investors, takes a different approach, alerting clients when potentially market-moving news events are starting to unfold.

One hedge fund trader who uses Dataminr’s service said it helped his firm avoid losses when stocks dropped after the Boston Marathon bombings in April. “It wasn’t on TV yet,” said the investor, who spoke on the condition of anonymity because he was not authorized to speak to the news media. “We got that alert early, we saw it and it worked beautifully.”

When it comes to individual stocks, Dataminr is useful for tracking companies that sell retail products, like Microsoft or Apple, the investor said. “People are tweeting about the products, or tweeting about different aspects of the business,” he said. “When you get to the more arcane situations, while the technology works, you don’t get the same volume of data coming through.”

Other companies are trying to offer something more basic: a version of Twitter that does not make compliance officers nervous.

Emmett Kilduff, a former investment banker in London, started Eagle Alpha after leaving Morgan Stanley last year. The company recently began offering Social Sonar, a compliance-friendly version of Twitter, to traders and brokers. The service, which does not allow traders to post messages, filters for posts from executives and government officials with the aim of alerting traders to information that appears on Twitter before it hits news wires. The most basic version costs $100 a month.

In a sign of the demand for these services, an academic who published a widely cited paper on Twitter data is looking to cash in on its investment potential. Johan Bollen, an associate professor at Indiana University’s school of informatics and computing, contended in a 2010 paper written with a doctoral student that an analysis of Twitter data could predict the Dow Jones industrial average with 87.6 percent accuracy.

Mr. Bollen is setting up a company called Guidewave Consulting to offer his Twitter data insights to clients in finance and other industries. His method is to measure collective mood, as distinguished from sentiment about a particular company, he said.

One of the fans of Mr. Bollen’s 2010 paper was Derwent Capital, the London firm that started a short-lived hedge fund based on its ideas.

Now the manager behind that fund, Paul Hawtin, is trying again. He said the Derwent fund closed because of market turbulence in the summer of 2011, when Standard & Poor’s lowered the credit rating of the United States, prompting prospective investors to flee.

This time, instead of licensing technology, Mr. Hawtin has developed his own software to detect sentiment as well as events in real time, he said. His new firm, Cayman Atlantic, opened on Nov. 1, with 50 private managed accounts, totaling $10.9 million.

“It’s like the Twitter hedge fund, Take 2,” Mr. Hawtin said.

Julia Werdigier contributed reporting.