<!-- Swiftype Variables -->


Academics tackle high-frequency trading problem

Eric Budish believes that high-frequency trading profit-taking is a violation of the efficient market hypothesis.

Trio proposes trades at discrete intervals such as every millisecond

Three academics have identified a market structure they contend would enhance liquidity and stability, overcoming detrimental impacts to investors of high-frequency trading.

The men call for scrapping the current market structure of so-called continuous buying and selling of stocks at any time that has led to the technology “arms race” to trade ever faster that harms liquidity and raises costs for institutional investors.

They propose instead trading in discrete time intervals — of, say, every second or every millisecond — they claim would strengthen the position of investors.

“Stocks trade in pennies,” or discrete prices, said Eric Budish, associate professor of economics at the University of Chicago Booth School of Business, in an interview. “If you want to bid more than me, you have to bid at least a penny more than my bet. You can't bid a millionth or a billionth (penny) more than my bid. And that's the sense that prices are discrete. There is a minimum price increment.

“There are good reasons for that. And we're in a sense arguing for a minimum time increment” for trading, he said.

Mr. Budish — along with Peter Cramton, professor of economics, University of Maryland, College Park, and John J. Shim, Chicago Booth School doctoral candidate in finance — co-authored an unpublished paper outlining their market restructuring idea. The paper, “The High-Frequency Trading Arms Race: Frequent Batch Auctions as a Market Design Response,” was the winner of the $100,000 AQR Capital Management LLC Insight Award, announced May 21.

The Budish team earlier this year was beginning to generate interest in their potential solution, even before Michael Lewis' controversial book, “The Flash Boys: A Wall Street Revolt,” was released and drew significant public attention to the issues raised by high-frequency trading.

“There is a lot of interest in the work” proposing discrete trading intervals, Mr. Budish said.

Mary Jo White, chairwoman of the Securities and Exchange Commission, said in a June 5 speech that “today's technology-driven market ... may work against, or at least not optimally for, the interests of investors and companies. ... I am receptive to more flexible, competitive solutions that could be adopted by trading venues. These could include frequent batch auctions or other mechanisms designed to minimize speed advantages.” Ms. White's comments come after comments in May by Mr. Budish before officials at an SEC conference on financial market regulation and at a meeting of the Financial Industry Regulatory Authority.

Mr. Budish recently also discussed his ideas with representatives of pension funds, investment managers, broker-dealers, exchanges, high-frequency trading firms and regulators, he said. He declined to identify the private entities.

CFTC testimony

In February, Mr. Budish testified at a Commodity Futures Trading Commission technology advisory committee meeting. The “HFT arms race is a symptom of a basic flaw in modern financial market design: continuous-time trading,” he said. The “arms race is a constant of the market design” and “harms liquidity” in terms of spreads and order depth and is “socially wasteful,” Mr. Budish testified. The “arms race profits come at the expense of liquidity providers, which ultimately harms liquidity. All of the money spent in the arms race comes out of the pockets of investors.”

Markets should “replace continuous-time limit order books with discrete-time frequent batch auction,” he recommended.

“Eric and his team really help shape the discussion in a more analytical ... framework to get better conclusions,” said David Kabiller, founding principal at Greenwich, Conn.-based AQR.

Donald B. Keim, professor of finance at the Wharton School of the University of Pennsylvania, Philadelphia, and director of its Rodney L. White Center for Financial Research, thinks the Budish team idea has some promise.

“Generally, I think this is an idea worth considering,” Mr. Keim wrote in an e-mail. “I'm not so sure it would enhance the current levels of liquidity in the market, but it might enhance stability. I don't necessarily see this as a replacement to continuous trading (but instead) more a supplemental trading mechanism. I doubt the additional cost of waiting to trade would be material enough to matter to pension funds.”

Discrete time has “beneficial effects” over continuous time, Mr. Budish said. For one, “it stops the race for speed. It stops investments that are in the order of hundreds of millions or billions of dollars annually for what seems like economically trivial speed improvements. It stops the speed race,” while enhancing liquidity and market stability. The cost to investors is having to wait a tiny bit of time to trade, Mr. Budish said. For the other, discrete time “transforms the nature of competition from competition based on speed to competition based on (stock) price,” he said.

As the market is structured now, continuous trading enables traders “to make money from symmetrically observed (public) information, information that many market participants observe at the same time,” Mr. Budish said. Active investment managers seek to make money from asymmetric information, “information they have that the rest of the market doesn't have.”

“If I'm smarter about a company's next-quarter earnings than the rest of the market ... I can make money from that information,” Mr. Budish said. But even in an efficient market, faster traders gain an advantage over other investors, even though all have new information simultaneously and the market instantly incorporates news about valuation into stock prices.

“You make money from symmetrically observed information if your ... trade gets processed by an exchange a tiny fraction sooner” than the next trade, Mr. Budish said. “Ultimately in a competitive market that money (gained by the advantage of faster trades) comes out of the pockets of investors, especially (institutional) investors looking to trade large quantities” of stock.

Violation of efficient market

The HFT profit-taking is a violation of the efficient market hypothesis built into the market design, Mr. Budish said.

Eugene Fama, Nobel economics laureate and Chicago Booth School finance professor, has declared “these technical arbitrage opportunities are built into the market design,” Mr. Budish said. These “kinds of arbitrage opportunities aren't supposed to exist in a well-functioning, efficient market. But they're built into the continuous-time market design.”

Under the existing market structure, “high-frequency traders are acting rationally with respect to the rules of the market as they are currently given,' Mr. Budish said. “Our research suggests, though, there is a structural flaw in those rules.”

That is continuous trading, “meaning that one can buy or sell stock or other securities literally in an instant during the trading day, where an instant is measured as finely as computers allow, less than a nanosecond currently,” Mr. Budish said.

“These (technical) arbitrage opportunities induce a race for speed.” The paper “makes a case against continuous time (trading) and in favor of discrete time,” he said.

“Guys digging high-speed fiber optic cables or ... building state-of-the-art microwaves (for trading) would lose from this proposal,” Mr. Budish said in his remarks to the CFTC. “Speed-based strategies are no longer going to be profitable under this (discrete) market structure.”

At what discrete time interval should trading occur? In his CFTC testimony, Mr. Budish said the interval would be at a “discrete time at very high frequencies ... faster than humans can perceive.”

The paper did not recommend a specific interval. “We purposefully don't reach a conclusion in this paper what the optimal discrete time interval is. We think the important conceptual point is moving away from continuous time to (move to) discrete times. ... This is the first paper in what hopefully will be a series of papers” that might offer a definite time increment, he said.

This article originally appeared in the June 9, 2014 print issue as, "Academics tackle high-frequency trading problem".