Trading in a High
The Flash Crash and beyond
It happened in a flash. Proctor & Gamble stock was trading at its normal price of about $60 a share at 2:40 p.m. on May 6, 2010. Three and a half minutes later, its price had dropped to $39.37. A minute after that, it was back at $60. But the Dow Jones Industrial Average kept falling, losing about 500 points between 2:41 and 2:46 p.m. Then it snapped upward, and by 3:00 p.m. the Dow was back where it had been at 2:40. Traders watched in disbelief. Even though stock and futures markets quickly regained their balance, everything looked different after it was over. The markets had failed, and no one knew why.
The market’s May 6 seizure quickly became known as the “flash crash.” The actual financial losses it caused were fairly minor, because almost all of the stock and futures trades that happened during the plunge were canceled. The crash was important because it exposed structural weaknesses that must be corrected, quickly, to prevent a real catastrophe. “It was a great wake-up call,” says Maureen O’Hara, Johnson’s Robert Purcell Professor of Finance. She talked about the causes and effects of the flash crash at the November meeting of the Finger Lakes chapter of the Johnson Club.
O’Hara is a specialist in microstructure, the study of stock markets and trading structures, and she served on a task force that was charged by the Commodities Futures Trading Commission (CFTC) and the Securities and Exchange Commission (SEC) with investigating the flash crash and recommending fixes. Their September 30 report says that three recent, largely unregulated changes in equities markets combined to produce a “perfect storm” on May 6. The first is high-speed trading, which uses fast computer connections to get orders from buyers to sellers in milliseconds. The second is the increased role of high frequency traders, which has changed how market making occurs. The third is changes in routing, or how buyers and sellers get together in particular markets.
There are only about 400 high-speed trading firms in the United States. They make very small amounts of money on each stock trade and exploit small differences between bid and ask prices by making thousands of trades per second. The trades are so fast that no individual trade is actually done by a human. They are done by computers that are programmed with commands that trigger buying and selling when markets move up or down. Despite their small numbers, high-speed traders account for almost three-quarters of the trading volume in equities and half the volume in futures contracts, O’Hara says. As a result, they play a dominant role in how equities markets are made. So if all the high-speed trading computers suddenly switch to sell orders at the same time, trouble might develop so quickly that no one could stop it.
The day of the flash crash started badly, with the Dow Jones Industrial Average losing value because of investor fears for the solvency of European debt. Sell orders had been outpacing buy orders all day, eroding the integrity of futures markets. Then at 2:32 p.m., a high-speed firm in Kansas City placed an order to sell 75,000 contracts in E-Minis, the futures exchange for the S&P 500, at a value of $4.1 billion. That is when the flash crash started. “Computers saw too many sell orders,” says O’Hara. “And when the computers see things they don’t understand, they revert to human control. The problem is that in many high-speed firms, there were three guys sitting there, and they had no idea what was going on. They were like deer in the headlights, with no idea of what to do. So they stopped trading. And once you stop making markets, there is no one to buy what others want to sell.”
The New York Stock Exchange (NYSE) and other exchanges have “circuit breaker” rules that can slow down or reroute trades when things are going haywire, and these could have stopped the flash crash. But fewer than 30 percent of trades now go through the NYSE. There are now 13 exchanges for U.S. equities. There are also more than 30 “dark pools,” where mutual fund managers get together privately to match buy and sell orders; four major “internalizers,” which sell buy orders and buy sell orders from online firms like Charles Schwab and E*Trade Financial; and over 100 brokers who clear trades at their own desks. “Equities trading is highly fragmented,” says O’Hara.
At 2:41 p.m., selling in futures markets spread to stocks, and the market plunge began in earnest. The Dow’s loss in value went from 2.5 percent at 2:30 p.m. to 9.2 percent at 2:46 p.m., a decline of 998 points for the day. But trading in futures contracts is not as fragmented as equities. At 2:46 p.m., the Chicago Mercantile Exchange paused trading in E-Mini futures for five seconds, allowing buy orders to accumulate. With equilibrium restored, the price of the Dow snapped back to its 2:41 p.m. price before 3 p.m. By the close of trading on Monday, the Dow was less than 1 percent lower than it had been at the opening bell on Thursday.
O’Hara doesn’t blame the trader in Kansas City. “High speed traders do a lot of good, because they make markets,” she says. “That trade might have been the straw that broke the camel’s back, but the camel was in very bad shape when it arrived.” Although the flash crash was not caused by bad actors, she says that that the internalizers didn’t do themselves any favors that day. “As soon as the crash started, they re-routed all their orders to the stock exchanges,” she says. “When the going got tough, they were out of there. So what are people paying them for?”
Beginning in May, the SEC started testing a system that applies circuit breaker rules to individual stocks. “These have been playing to mixed reviews,” says O’Hara. When the price of a company’s stock goes haywire, it is almost always due to a mistake in the buy or sell order –what financiers call “fat finger” trades. “These would have been cancelled anyway,” she says. “But when there’s a circuit breaker, it can shut the whole market down.”
Another possible solution is called “limit up and limit down,” which means putting a collar on a stock. If pending trade orders would cause that stock to rise or fall beyond the limits, those trades would not go through. “The terrible thing about the flash crash was that 20,000 trades were canceled,” says O’Hara. “So if you can stop a trade that should not go through anyway, why not do it?” Limit rules might be coming along by the spring of 2011, she says.
O’Hara says that markets in general are in very good shape, with robust liquidity and safety measures. She adds that individual buy-and-hold investors were completely unaffected by the flash crash. But she also makes several pointed conclusions. “If you’re still dreaming of making a fortune by sitting in your garage and making stock trades all day, forget it,” she says. “There is no way for an individual to compete with high-speed traders.” She also says that the crash exposed a weakness in Exchange-Traded Funds (ETFs), which buy and sell like stocks but are actually index funds. “ETFs are supposed to be index funds you can buy and sell all day,” she says. “But 70 percent of all the cancelled orders on May 6 were from ETFs. They had problems maintaining liquidity.”
O’Hara also says that investors should look closely at brokers before hiring them: “Do you want to get where you’re going in a rickety car with bad brakes because it’s really cheap, or are you willing to pay a little more to ride in a car that’s safe? So far, we have not put a spotlight on the behavior of brokers.”