In English football (soccer) an "own goal" is when a player scores against his own team (usually accidentally). Sports videos and books are full of own goal blunders and the players almost never live them down. The term has become synonymous with anything that backfires upon the perpetrator. So when BATS experienced an embarrassing technology problem on the day that it was launching its first IPO - on its own firm no less - it joined the ranks of some of soccer's most infamous players. BATS cancelled its IPO and is working hard to repair its damaged reputation.
It seems one of BATS' servers malfunctioned on that fateful day, taking its share price from $15.25 to just 2 cents within minutes. The same server controlled trading for Apple shares, which dropped over 9% and triggered exchange circuit breakers and halted trade for five minutes nationally. As I told Forbes last Friday, the triggering of circuit breaker in Apple is a reassuring sign that there are better precautions in place to prevent a repeat of the flash crash that slammed the broader market on May 6, 2010.
However, the fact remains that an issue like this was - and is - bound to happen. Multiple market venues, innovative technology and complex algorithms all add to the primordial soup that is the new market structure. One glitch on a big trading venue like BATS can do a lot of damage to the market overall. With BATS and other ECNs looking at exchange status and planning to nab listings away from bigger exchanges, the issue becomes more salient.
BATS told Bloomberg that the coding for opening IPOs was new and, although it had been tested in the lab, it had not been tested in "real-world production." It is difficult to test a new code in the real world of course, because it would impact the real market. The question is, did BATS do adequate stress testing in the lab? Did BATS have back-up procedures in place if it went wrong in the real world?
BATS' response to the glitch was admirable, having spotted the issue in early trading they alerted the SEC and then stopped trading (and cancelled the IPO) a couple of hours later. It could have been a lot worse. But in the meantime confusion reigned - and the market does not like to be confused. As more trading venues garner more volume and become more exchange-like with IPOs, the danger of technical glitches increases. Add to this the market's continuing reliance on algorithms, which can malfunction and go rogue, and you get the recipe for a perfect flash crash storm.
As I noted in an earlier blog, a new study found more than 18,000 instances of ultrafast mini-crashes over the past five years - almost one per trading day on average. These mini-crashes took place in under 1.5 seconds, with many happening in less than 1/10th of a second, and moved the stock price by more than 0.8%. And there is some evidence that the faster the trades, the higher the likelihood of an incident. With the SEC sniffing around high frequency trading firms (trying to see if they give clients an unfair advantage over others) and making progress with a consolidated audit trail for monitoring HFT activity, the timing could not be worse for issues like BATS' to happen.
To be truly responsive, exchanges and trading destinations have to take greater responsibility for extensive back-testing and market simulation, so that before algorithms (and new codes) go live it is possible to see how they would perform in production. Real-time market monitoring and surveillance is crucial and allows more rapid response to potential crises and market abuse – potentially allowing ultrafast action to prevent or minimize any market impact. Hopefully the BATS incident serves as a wakeup call.