I always thought that when a really, really, big market order mistake was made that it would be made on an electronic exchange by an algorithm. It appears I was wrong -- so far. A mistaken order on Japan's over-the-counter stock market, worth over $600 billion, looks to be the fault of a very human fat finger.
According to Bloomberg, more than 40 requests to deal shares worth 67.78 trillion yen (US$617 billion) were cancelled before they could be matched. The Japan Securities Dealers Association explained that one of its members had mistakenly entered a combination of volume and price, rather than simply volume. The biggest order was for 1.96 billion shares of Toyota, 57% of its outstanding shares, and for Honda, Canon, Sony, and Nomura.
Had the orders even begun to be matched, the result could have been catastrophic. The trades would have been displayed upon completion, sending algorithms monitoring the OTC markets into a trading frenzy. I agree with Larry Tabb, founder of research firm Tabb Group, who said it could "absolutely" happen here in the US, due to a lack of checks and balances on OTC markets.
I still believe, however, that the biggest possibility for error comes from rogue algos on electronic exchanges. The immediate knock-on effect, such as happened on the May 6, 2010, Flash Crash, which temporarily wiped nearly a trillion dollars off the market, would be enormous.
Almost three years ago, I said that, although financial institutions have tightened risk processes to prevent human fraud, they are still not adequately monitoring their algorithmic trading. This was part of my 2012 capital markets predictions, which also included a forecast that a financial institution would take a billion dollar (or more) hit when a rogue algorithm went wild.
[Read more market observations from WS&T Thought Leader Dr. John Bates: HFT's Death by a Thousand Cuts.]
Algos triggering algos can cause much more damage than a fat finger, because by the time they are detected and stopped a lot more instruments and exchanges would be impacted. Witness one of the biggest single mistakes made on August 1, 2012, which resulted in a rogue command from Knight Capital's new software. The command unleashed a flood of orders to the New York Stock Exchange, unrestricted by volume caps, and cost the company around $10 million a minute -- $440 million. (Knight had to be rescued by a consortium that included HFT firm Getco.)
Detecting unwanted behavior such as system errors, insider trading, fat fingers, front-running, wash trading, or quote stuffing will help financial institutions prevent trade-related meltdowns such as Knight Capital, or what nearly happened in Japan. Monitoring trader behavior can also help avoid another LIBOR or FX fixing scandal.
As I see it, not enough progress has been made by financial institutions in the monitoring and surveillance of trading -- human or machine. I stand by my $1 billion hit prediction, though I may have to throw fat fingers into the equation.Dr. John Bates is a Member of the Group Executive Board and Chief Technology Officer at Software AG, responsible for Intelligent Business Operations and Big Data strategies. Until July 2013, John was Executive Vice President and Corporate Chief Technology Officer at Progress ... View Full Bio