The Fast and the Incurious: How High-Speed Stock Trading Pushes Out Smart Money

Automated trades are tactically brilliant, but these “speed demons” can discourage serious market research.

August 20, 2021

| by Edmund L. Andrews
A trader reacts as he works on the floor of the New York Stock Exchange in New York City. Photo by: Reuters/Andrew Kelly

Algorithms now account for half of all stock trading in the U.S. | Reuters/Andrew Kelly

Does high-speed algorithmic stock trading make the stock market dumber? When it comes to smaller companies, the answer may be yes.

Automated trading, in which computers spot split-second opportunities to outmaneuver the competition, now accounts for more than half of all stock trading in the U.S. “Flash boys,” as Michael Lewis dubbed them, use algorithms to anticipate and profit from big moves caused by slow-moving funds. These algorithms can detect when mutual funds are about to buy large volumes of a particular stock, for example, and direct the system to immediately start buying up shares. The algorithmic traders can then cash in on a brief but predictable jump in the stock’s price as the mutual funds make their move.

In theory, automated trading makes the market more efficient in two ways. First, it speeds up price discovery, because the algorithms are essentially inferring from past trade and quote patterns what the smart-money people have already figured out. Second, it leads to higher trading volumes, which can potentially reduce the cost of individual transactions.

But a new study finds strong evidence of a downside: Automated traders also appear to drive out traders who get their edge from fundamental research. For small-company stocks, this crowding-out effect leads to less fundamental research on particular stocks as well as less information in the marketplace. “Slowing down those speed demons could actually be good for the markets and for the economy,” says Charles Lee, professor of accounting at Stanford Graduate School of Business, who coauthored the study with Edward Watts, a former Stanford GSB doctoral student who is now a faculty member at Yale University.

Tick Talk

In their study, Lee and Watts explore what happened when automated traders hit a temporary speed bump. They took a fresh look at vast amounts of data from a two-year pilot project that the Securities and Exchange Commission conducted on the effects of “tick size,” the minimum increment by which a stock’s price can go up or down.

Slowing down those speed demons could actually be good for the markets and for the economy.
Charles Lee

Before 2005, when stock prices were quoted as fractions of a dollar, the minimum tick was often one-sixteenth of a dollar, or 6.5 cents, or higher. Since 2005, stock prices have been decimalized and the minimum tick size has dropped to 1 cent. That change gave a big boost to automated trading, because it’s much less costly to move in and out of stocks at high speed when the incremental price changes are smaller. To see if the lower tick size was good or bad for small-company stocks, the SEC launched an experiment in 2016. In a randomized test, 1,200 small companies (with market capitalizations between $1 billion and $3 billion) had their tick size raised to 5 cents, while a control group of similar companies kept the 1-cent tick.

In line with what many experts had predicted, Lee and Watts showed that the higher tick size led to a substantial drop in algorithmic trading. Automated trading dropped by almost 11% more for the higher-tick stocks. On top of that, however, they found evidence of an increase in fundamental research activities: Searches for those companies’ filings on the SEC’s website surged, especially in the weeks before their earnings announcements.

The real news, however, was that the stock prices of companies with less automated trading began to reflect upcoming earnings news more accurately. To gauge that, Lee and Watts measured how much of the upcoming earnings news each quarter is anticipated by the 60-day stock returns leading up to each announcement date. If increased fundamental activities in the firms with higher tick sizes resulted in more informed stock prices, they reasoned, then these firms’ pre-announcement stock returns should better anticipate the upcoming earnings news. Sure enough, they found that in firms where tick sizes were increased to 5 cents, the pre-announcement stock returns contained more information about the upcoming earnings news.

Gnats on an Elephant

Why would algorithmic trading cause the market to be less knowledgeable? Lee’s theory is that automated trading is driving out traders who get their edge from doing fundamental research. Think of it this way: If you’re a portfolio manager trying to eke out an edge through fundamental analysis, your insights are a lot less valuable if an automated trading system can infer your planned trades and act on them before you get a chance to carry them out.

Trading algorithms enable tactics such as “back-running,” in which a system looks for signs that a big institution is getting ready to buy or sell a particular stock, and then jumps ahead on its own. Large fund managers know they cannot purchase huge blocks of stock at once, so they often divide up their purchases into smaller orders. For orders that are more difficult to execute, they often need to start buying first thing in the morning. Algorithmic trading systems quickly recognize those early purchases and immediately start buying the same stock in anticipation that the larger fund’s bulk purchases will temporarily drive up its price. “They’re like gnats on an elephant,” says Lee, referring to the algorithmic traders.

In effect, says Lee, automated trading enables a form of poaching. An algorithm doesn’t care about fundamental information or analysis, but it can extrapolate someone else’s insights from their trading behavior. If there are enough algorithmic traders around, they can create a big disincentive for other traders to invest in original research, especially when it comes to smaller companies with lower trading volume.

Lee cautions that the study is focused on smaller companies by design, and that the effect of tick sizes could be different for stocks of companies with high market valuations. For giant companies with millions of investors, a change in the minimum tick size may not have the same effects.

But at least for small companies, the drop in automated trading during the pilot project seems to have made the market more astute. “It seems to have had a cleansing effect on pricing efficiency,” Lee says. “Once the gnats went away, the elephants came back.” It remains to be seen, however, whether the SEC will make any permanent changes to the tick size as a result of the pilot project.

For media inquiries, visit the Newsroom.

Explore More