Brief price gaps in stocks cost investors $2 billion a year
Study finds nearly 24 per cent of trades may not be executed at best available prices
Nearly a quarter of U.S. equity trades may not be executed at the best price available in the market, costing investors at least $2 billion a year, according to a new study funded by the U.S. government.
The study provides new evidence of momentary pricing discrepancies that researchers say can be exploited by highspeed traders looking to make a quick profit.
The Departments of Defense and Homeland Security provided about $1.25 million for the study to better understand how U.S. markets might respond to a cyberattack, according to spokespeople for the agencies’ research arms. The Wall Street Journal first reported in 2017 that Pentagon researchers have been studying the issue.
Researchers examined all trades for stocks in the Russell 3000 index during 2016 and found that nearly 24% of them were executed during times when better prices may have been available. That equated to about 1.1 billion trades that might have been off by pennies a share from better prices available elsewhere in the fragmented U.S. stock market, the authors found. The market includes 13 different exchanges and dozens of offexchange
trading venues known as dark pools.
Critics of high-speed trading argue that the firms use insights gleaned from expensive, ultrafast data feeds sold by exchanges to make quick profits off investors trading at slightly out-of-date prices. Modern Markets Initiative, a lobbying group representing fast-trading firms, counters that market structure and regulation don’t allow anyone to profitably exploit, or arbitrage, momentary discrepancies in stock prices. “While arbitrage could happen, it’s a result of almost pure chance,” said Bill Harts, advisor to the MMI and the group’s former CEO.
The study, conducted by federally
funded research company MITRE Corp. and academics at the University of Vermont, doesn’t quantify profits from such arbitrage. But its authors say there’s plenty of opportunity for such profits to be made. “There are pure arbitrage opportunities, and they do follow a surprising structure and pattern across different tickers,” said Brian Tivnan, chief engineer at MITRE, who co-authored the research. “We see them from the smallest of the Russell 3000 all the way up to Apple.”
Shares of Apple Inc. saw one such opportunity around 9:48:55 a.m. on Jan. 7, 2016, according to data provided by Mr. Tivnan. For about 1,862 millionths of a second, there was a gap of up to 6 cents between the best prices at which investors could buy and sell shares of Apple in different trading venues. It was one of 31,935 such dislocations that happened in just the first hour of trading that day, which Mr. Tivnan selected at random for closer inspection.
Mr. Tivnan’s team examined trades completed during such dislocations to see if they were executed at the national best bid or offer.
That is the price at which brokers must seek to execute trades for their clients, thus ensuring that investors get the lowest offer when buying shares or the highest bid when selling. The quotes are disclosed on a regulatory data feed called the Securities Information Processor, which Mr. Tivnan’s team compared with quotes listed in data feeds sold directly by exchanges, as compiled by Thesys Technologies LLC. A spokeswoman for Thesys said it was the largest study of its kind.
They found that the national best bid or offer often wasn’t really the best quote available in the market at the time of the trade. One trade completed at 9:48:55.398272 sold 100 shares of Apple for $99.12—then the national best bid available in the market—but direct feeds from exchanges showed 100 shares could have been sold elsewhere for $99.16, costing the seller $4 on the $9,912 transaction.
Altogether, Mr. Tivnan’s team calculated that investors buying and selling Apple stock on Jan. 7, 2016, may have been able to save themselves at least $105,000 by trading at prices published on the exchanges’ direct data feeds. Across all of 2016, the researchers tabulated a total opportunity cost of about $2 billion, which they say is a conservative estimate.
Modern Markets Initiative said in a three-page letter critiquing the study that by the time someone sees better prices elsewhere, the opportunity to trade at those prices is already gone.
Others who reviewed it for the Journal also expressed skepticism. James Upson, associate professor of finance at the University of Texas at El Paso, said that “there is probably some opportunity cost that is occurring” because of differences in prices across data feeds. “But it’s actually pretty small. It’s smaller than what they’re saying.”
A 2017 study that tried to quantify potential profits from direct data-feed arbitrage in Dow Jones’ 30 stocks came up with a grossprofit estimate of just $11.6 million. Another estimate from 2016 using S&P 500 stocks pinned potential profits at around $3 billion.