Reliance on bots for trades makes fake news a Wall Street problem
The story was explosive if true: Google planned to buy Apple for $9 billion, according to a Dow Jones Newswire headline earlier this week.
The prospect, of course, is absurd not only because Apple is worth more than $800 billion but because the story said “Google Chief Executive Larry Page had secret talks with the nowdeceased Steve Jobs in 2010 to firm up the deal.”
Dow Jones deleted the story after just a few minutes, explaining that it was accidentally published as part of a technology test. And sophisticated stock traders likely didn’t even have enough time to digest the news before it was corrected, much less take any action.
Yet that brief bit of fake news was blamed for a minor jump in Apple’s stock price to about $158 a share before falling back to around $155. But if the headline was too absurd to be believed, who was likely momentarily tricked? Bots.
An increasing portion of stock trades every day are controlled by computer algorithms, many of which scan Twitter feeds, news headlines and other social media looking for tidbits that can move markets. And much of that trading is taking place in a blink of an eye with highspeed traders who measure time in microseconds.
In a recent research report, JPMorgan Chase estimated that just 10 percent of daily trading is done by human stock pickers.
The growing reliance on technology rather than humans to make stock trades has made identifying disreputable news a bigger challenge, market industry veterans have said. In 2013, a fake tweet about an explosion at the White House sent markets on a roller-coaster ride. In August, a false tweet that White House economic adviser Gary Cohn was resigning sent stocks tumbling.
“This is why the machines, no matter how smart, are never going to be as sophisticated as a human,” said Tom Lin, a law professor at Temple University, who has studied the impact of technology on financial markets. “The bots cannot discern humor or nuance. They have no real context. They are just going to execute it on whatever they see.”
The news headlines appeared for only a few minutes, from about 9:34 a.m. to 9:36 a.m. Tuesday. But that is now enough time for bots to move markets significantly, said Lin. “Two or three minutes is a long, long time,” he said.
For long-term investors, such glitches are not a major problem, said Lin. The markets will correct themselves over time. But on a day-to-day basis it is potentially problematic. “It is alarming because it calls into question the integrity of the financial system,” Lin said.
Traders who work on the floor of the New York Stock Exchange are expected to spot humor in social media posts, while computer programs are unable to spot false news.