Economics Sep 3, 2019
Some High-Frequency Trading Strategies Can Damage the Stock Market’s Health
But a small tweak to how trading orders are processed could help.
Think nothing can happen in 64 millionths of a second? You’d be wrong: a trade can be processed on a major global stock exchange in that time.
Over the past 10 years, many exchanges have cut trade-processing times dramatically. The stock exchange BYX, for example, increased order-processing speed by more than seven times in that period. And this new, lightning-fast speed can earn high-frequency traders big money.
High-frequency trading represents an advantage for those who can act quickly on new market information. But how does it affect the market itself?
Joshua Mollner, Kellogg assistant professor of managerial economics and decision sciences, wanted to find out.
“One of the big changes related to stock markets over the past 10 to 15 years has been the rise of high-frequency trading,” Mollner says. “So we want to understand its impacts and, perhaps more interestingly, whether we need to rethink the mechanics of stock markets and how they are set up and regulated, based on those impacts.”
He was especially interested in how different strategies used by high-frequency traders might affect the overall health of the market.
In a study with Markus Baldauf of the University of British Columbia, Mollner shows that high-frequency trading, when used to power what’s known as an arbitrage strategy, can reduce the overall health of the stock market on two key measures—how liquid it is and how informative stock prices are. As a solution, they propose small but meaningful changes to how stock exchanges process orders.
The Need for Speed
High-frequency trading represents a major shift in how stocks are bought and sold.
“Once upon a time,” Mollner says, “the stock exchange was a physical place where humans would go and trade with one another. But today everything is automated and done by computers. That means trading decisions are much faster. ‘High-frequency trading’ refers to the extreme end of that spectrum. Even a few microseconds slower or faster can make a big difference for a trader.”
High-frequency traders use market knowledge and predictions to program an algorithm aligned with their trading strategy. Some, for example, may set the algorithm to buy shares of a given tech stock at a specific price and sell that same stock at a higher price the same day.
“It’s not about making drastic, sweeping changes. Small surgical changes, like the delays we’re proposing, can have a big impact.”
Not surprisingly, major hedge funds or investment banks are best poised to harness high-frequency trading because they can afford the necessary technology. “Being the fastest trader in the market requires a lot of resources and technological investments,” Mollner says.
In this context, the researchers sought to understand two things: how market “health” changes as trades happen more quickly, and whether those impacts warrant policy changes in how markets are set up.
Measuring the Impact of High-Frequency Trading on Market Health
So what makes a stock market healthy?
According to Mollner, there are two main components.
Healthy markets are liquid, meaning they involve small transaction costs. More liquid markets mean more participants—from large institutions to individual investors—and a higher volume of mutually beneficial trades, which promotes greater overall economic efficiency, Mollner says.
The second component is informativeness, which means that stock prices relate meaningfully to the fundamentals of the companies that offer them. “Informative stock prices create signals that help channel capital to its most productive uses,” Mollner says.
The researchers built a mathematical model (not using actual market data, in this case) to examine the impact of high-frequency trading on those stock-market health measures.
To measure liquidity, they focused on the “bid–ask spread,” or the difference between the prices for which high-frequency market makers would buy and sell a given share. “If that number is small,” Mollner says, “that means it’s a market where transactional costs are small, or a more liquid market.”
The model measured informativeness by estimating how much fundamentals-focused research was being done by investors to make investment decisions—such as predicting the success of a new product by using artificial intelligence to read product reviews and social-media posts. “The idea is if more research is being done, then that’s going to be reflected in stock prices and those prices will be more informative in the end,” Mollner says.
Market Maker and Arbitrage Traders
The impact of high-frequency trading, the researchers found, depends on the specific type of investment strategy being used.
On the one hand are the high-frequency market makers, or traders who offer to buy and sell a given stock and make money from the price difference, or the spread.
“They’ve gotten really good at managing the risk of trading with someone who might know more than they do—someone who might buy only when the price is likely to go up and sell only when the price will go down,” Mollner says. “That ability helps them operate profitably even while charging a fairly small spread. Their presence leads to more liquid markets.”
On the other hand are high-frequency arbitrage traders, who assess where prices are going literally in the next milliseconds, then try to trade in the most profitable direction. “One of these traders might be monitoring the market and get the sense that some big trader out there is buying, so prices are going to go up, and they’ll start buying as well,” Mollner says. “Effectively they’re amplifying whatever the informed traders are doing.”
While this form of arbitrage can be profitable for the trader, the researchers’ model showed that it can cause problems for the market.
Why? Because that amplification of better-informed traders’ moves, in turn, makes things riskier for market makers, forcing them to charge a larger spread to be profitable and ultimately reducing market liquidity. And in addition, high-frequency arbitrage also leads to less informative prices.
“If I’m a trader who’s done some fundamental research and acquire some information but I’m in a world of super-fast high-frequency traders, I’m not going to be able to trade much volume before they figure out what I’m up to,” Mollner says. “That weakens my incentives to do the research in the first place.”
Small Change, Large Impact
Because the research suggests that high-frequency arbitrage reduces market health, it makes sense to do something about it.
It turns out a small tweak to how exchanges process trading orders can help.
The researchers found that introducing a short processing delay—a slight pause before the order is executed—for certain order types could ultimately reduce the negative impact of high-frequency arbitrage. The researchers propose delaying everything except cancellation orders, which would be processed immediately, as they are now.
Mollner explains why this works: “Say there’s some strong signal that prices for a stock are going to go up. That sets off a race. The market makers race to cancel their current orders to buy and sell. Meanwhile, the arbitrage traders race to send orders to trade at the market makers’ originally quoted prices, to take advantage of what they see as mispricing. So if we process cancellations immediately but delay everything else, that tilts the scale in favor of the market makers and against the arbitrage traders.”
Of course, the high-frequency arbitrage traders won’t embrace this idea, Mollner says: “It’s safe to say not everyone is going to be in favor of this.”
However, it’s already happening: several exchanges, including the Cboe-owned EDGA, have proposed exactly what Mollner suggests; the Cboe approached the U.S. Securities Exchange Commission in June 2019 about delaying EDGA non-cancellation orders by as little as four milliseconds to reduce the negative impact of high-frequency arbitrage.
“It’s not about making drastic, sweeping changes,” Mollner says. “Small surgical changes, like the delays we’re proposing, can have a big impact. The tiniest speed bumps can make a big difference.”