A "kill switch" would enable market participants such as exchanges or broker dealers to switch off immediately trading algorithms that are causing instability or erroneous trading in markets. The concept of a kill switch might work in the following way: an exchange could set a limit on the amount and/or value of trades that a particular firm could execute over a certain time limit. If it exceeded those limits, that might trigger a kill switch that would allow the exchange or broker to shut down the trading session.
The subject was given fresh impetus in August 2012 when Knight Capital Group rolled out new trading software that caused erroneous trading in more than 140 stocks on the New York Stock Exchange. It took Knight three quarters of an hour to locate and resolve the problem, and the $460m loss it subsequently recorded for trading out of its positions meant the electronic market maker effectively lost around $10m per minute.
A subsequent industry roundtable in the US debated whether kill switches should be introduced. Some worried that applying kill switches could hurt market liquidity and sentiment. Others pointed out that adding another layer of software did not necessarily solve the problem. In Knight's case, the problem was "old dormant code" with an undetected error and was triggered by the updated software. It then began generating orders that were unrestricted by volume limits.