CME Group’s (CME) electronic oil trading platform, among the fastest in the world, crashed on Monday afternoon after being flooded by a surge of some 12,000 price quotes in the span of four minutes. The platform remained down for more than an hour, forcing trading into the open outcry pits on the floor of the New York Mercantile Exchange, an increasingly rare scene these days thanks to the steady rise of electronic trading in recent years.
In a statement, CME cited “technical issues” as the reason behind the shutdown. Within minutes of the trading halt, the finance blogosphere was buzzing with rumors and innuendo of CME being victimized by a rogue trading algorithm “gone berserk.” Or worse: It was the work of some quote-stuffing computer program bent on sabotage.
Now that the dust has settled, to the dismay of computer-trading conspiracy theorists, it appears the crash was more than likely the result of a mere glitch in CME’s system, not some evil algorithm.
“It looks to me like CME might have rolled out some new software that hadn’t been tested sufficiently,” says Eric Hunsader, chief executive officer of market data service Nanex. Citing Nanex data (see visualization here), Hunsader says that a synchronization glitch in CME’s system caused price quotes to bunch up for three separate periods of time (35, 20, and 12 seconds), before getting released all at once. The system counted each batch of quotes as new, meaning each quote was counted over and over again, leading the system to crash.
Just as the May 2010 “Flash Crash” led to broad criticism of electronic trading, the CME crash may give fodder to those who say computer-driven trading is dangerous and unstable. “I take the opposite tack,” says Ben Van Vliet, a professor at Illinois Institute of Technology and an expert on high-frequency trading. “This is an example of something going wrong, but where the damage was minimized by the exchange acting in an expeditious, levelheaded way.” Unlike the Flash Crash, when exchanges had to break some 19,000 trades, no completed trades were busted as a result of the CME crash. “The Flash Crash was a whole lot messier,” Van Vliet says.
The incident comes as the Commodity Futures Trading Commission is starting to take a closer look at high-frequency trading. Hunsader, who has studied in detail the causes behind the 2010 Flash Crash, finds the CFTC’s recent saber-rattling amusing. “Those guys can’t even figure out how to read the data,” Hunsader says. “By the time they’ve finished forming committees to determine the definition of HFT, the market will have evolved to the point that their findings will be outdated.”