On Wednesday the stock market went through a period of extreme volatility, one that was eventually traced to one trading company, Knight Captial. Further analysis traced the problem to a test of a new trading software program that ended up costing Knight $440M and may bankrupt the firm.
While the actual mechanics of the failure require an in-depth knowledge of trading mechanics, the best way to summarize the problem is to say that the software started to do the opposite of the typical stock-trading mantra. Instead of “buy low, sell high” the software bought high and sold low - and did so at an alarming pace. The article quotes one observer saying “Do that 40 times a second, 2,400 times a minute, and you now have a system that’s very efficient at burning money.” The net result of this is that Knight may have completely wiped out all of its capital in one fell swoop. Reports are now emerging that Knight did not have a board-level focus on technology risk , something that is becoming increasingly common in most major corporations.
One thing I am struggling to understand is that Knight’s woes appear to be due to performing software testing with live data. As someone who now works in a limited fashion with software testing, I am amazed that an untested trading algorhythm was allowed into production, considering its ability to do major damage to the firm. I understand that no software can be fully tested, but surely the program could have been fed market data via some kind of firewall to see how it reacted before turning it loose on the company’s funds. Putting this software into production could have been the fault of an IT team failing to follow proper testing protocols, or could have been due to pressure from business partners in the firm demanding the new software be implemented without full testing.
I’ve written about the impacts of automation on the stock market before , and what happened Wednesday seems to be another indication that adoption of technology in the world of financial trading is happening at a greater pace that the ability of firm executives (as well as investors) to grasp its overall implications. The financial blog Zerohedge has an excellent interview with a High-Frequency Trader that illustrates some of the side effects of this lack of understanding. I think it is safe to predict that the failure of Knight Captial will not be the last high-profile financial event traced to automated trading or other technologies.