Knight Capital has admitted that it was a "technology issue" in its market-making unit that affected the routing of shares of about 148 stocks to the New York Stock Exchange, causing havoc to the markets on Wednesday and riling investors as stocks swung more than 10 percent in early trading.
The incident was reminiscent of the 2010 Flash Crash which temporarily wiped out almost $1 trillion in market value in minutes -- although stocks in the new market fiasco swung up and down, rather than just crashing.
In a statement, NYSE said it believed that its systems and circuit breakers had operated normally during early trading, a fact that likely prevented the Knight Capital fiasco from degenerating into a full-blown market crisis. Still, how could a rogue algo have been allowed to disrupt the markets just two years after the infamous Flash Crash and after a number of other more recent high-profile technical glitches such as the botched Facebook IPO and the failed public offering of BATS?
Experts point the finger at the failure to implement efficient risk controls.
A technology-enabled market will witness glitches and challenges from time to time. But that is why it is critical to test any changes in technology before they are implemented to make sure they work properly, notes Larry Tabb, CEO and founder of TABB Group.
Further, people should be monitoring executions and should put a stop to the trades as soon as they realize something has gone awry. There should be checks and balances within the trading firm, as well as at exchanges.
“Whoever’s order it was should be monitoring the execution, whether it’s a buy-side algo or a market-maker algo. You can’t monitor transaction by transaction, that’s humanly impossible. But there certainly should be control room monitoring of the health of all [the firm’s] algos,” Tabb says.
The August 1 fiasco raises a number of questions, he adds: “Where was limit up and limit down? Where were the risk gateways and the circuit breakers? Why wasn’t it caught before it went into production?”
Asked about the implications of the Knight incident, Tabb says: "I think it is very serious."
The wild trading went on for half an hour before action was finally taken. Knight Capital eventually told brokerages to send their orders elsewhere due to technical problems. “This shows the complete collapse of every control that there is supposed to be in the market,” Tabb warns.
He notes that risk controls, such as those outlined in the newly-updated Fix Protocol guidelines which call for more scrutiny of large orders by brokers to ensure that they aren’t fat-finger mistakes, the validation of an order’s symbol to make sure the correct stock is being traded and checks to ensure a fund manager hasn’t accidentally duplicated an order, can prevent these market incidents – but only if firms actually implement these controls.
[Read: FIX Protocol Risk Checks Take Aim at Fat Finger Trades, Errant Orders to learn more.]
John Bates, CTO at Progress Software, agrees that the August 1 mini-flash crash was entirely preventable, had the right risk controls been implemented.
“While it is reassuring that the issue was caught, preventing a larger-scale crash like the May 2010 Flash Crash, the fact remains that this may have been prevented entirely had the firm responsible been utilizing adequate controls such as real-time surveillance and risk monitoring,” Bates argues.
He points out that algos can move so quickly that by the time a human has noticed an error, the damage is likely to already have been done.
“We can only hope that this incident serves as a reminder of how seemingly minor technical glitches can cause drastic market moves, and that more firms deploy technology to focus on control and safety – not just speed."