Follow the money: Can Twitter predict the stock market?
Predicting the stock market is a multi-billion ‒ or perhaps
trillion-dollar ‒ industry, and companies like iSentium LLC want
a piece of the pie. The Florida-based business uses a relatively
simple method that is reminiscent of blackjack strategy: They
assign a +1 or -1 weight representing the positive or negative
tweets that investors and financial professionals make concerning
a particular stock. After analyzing over a million tweets a day,
the resulting numbers are then aggregated to get a feel for the
market sentiment about the company in question.
“What we’re telling you is what does the mob or the crowd say today,” Gautham Sastri, president and CEO of iSentium, told the Wall Street Journal. “Twitter is a big pipeline of emotion and we’re providing a snapshot.”
Firms analyze tweets to gauge stock sentiment, charge $15,000 per stock symbol http://t.co/IMz6L6JF4U
— Wall Street Journal (@WSJ) July 6, 2015
The financial sector has always been on the leading edge of technology, from Thomas Edison’s ticker tape to algorithms using complex mathematics, to squeeze information out of the stock market’s numbers. Today, technology can use natural-language processing to actually read social media and news to convert it into meaning for a computer.
Subscription-based services, such as Dataminr, that scan Twitter
and other social media sites, are used by news agencies to get
quick, automatic tips for breaking stories and by investors to
detect events that could warrant actions on the stock market to
gain a profit.
Other companies that are trying to get in on social media analytics include TheySay Ltd., which was founded by Stephen Pulman, an Oxford University professor of computational linguistics. The company has its own take on the art of using computers to draw meaning from text using the principle of “compositionality,” examining not only the meaning of words but their arrangement in relation to one another.
“Words in isolation may have a positive or negative sentiment but once you put them together they can often mean something else,” Pulman told the Wall Street Journal.
This use of information cuts both ways, though. On April 23, 2013, the Associated Press had its Twitter account hacked to make a tweet that falsely said there was an explosion at the White House. The S&P Index lost $136 billion in a matter of four minutes, though the value was recovered just as quickly.
“Most trading of securities and derivatives is accomplished using supercomputers wired directly into exchanges and other venues. They operate at trading speeds well below milliseconds so no human is involved,” former investment banker Walter Turbeville noted on the Demos blog. “I suspect that a tweet that comes from a ‘verified’ Twitter account and includes ‘Obama’, ‘White House’, and ‘bombs’ might qualify as a sell-triggering word combination.”
Naturally, Twitter, Inc. itself isn’t going to let everyone else
use their massive social media empire without getting in on the
game themselves. It sells data directly to banks and hedge funds
to conduct analysis. The company has also recently signed an
agreement with IBM, granting the computer giant privileged access
to the hundreds of millions of tweets that are posted every day.
IBM said that the uses for this technology go beyond merely predicting the stock market, taking it a step further to use consumer preference information to help banks and retailers create services and products more in line with the needs of customers.