A rogue algorithm used by the Knight Capital Group set off huge swings in prices and trading volumes for many stocks yesterday, and The Wall Street Journal is reporting that the fiasco might cost Knight $440 million in losses. For the operations staffs of the counterparties to these rogue-inspired transactions, though, I suspect the nightmare is just unfolding.
I am being a little facetious but, from an operations standpoint, it will be very difficult to discern what exactly happened in a transaction environment that is functioning at the speed of milliseconds, let alone at unprecedented volumes. Just exactly who is responsible for trading activity that is algorithmic and thus, by definition, automatic? Can you hold responsible a software glitch, a poorly written algorithmic rule or a blink in a low-latency connection? How exactly do you trace the full impact of even one rogue algorithm? If you could piece it together, where do you place the blame? Is a software developer responsible because he or she inadvertently created a software glitch? Or do you blame the quality assurance/testing staffer that missed the glitch?
Part of the problem with the black art of algorithms is that they are created in great secrecy and most firms that engage in high-frequency trading (and their customers) would prefer to avoid transparency. Yet with this latest crisis regulators and the many investors hurt by the rogue algorithm may want to pry open the locked door on the algorithm lab and justifiably ask for some answers.
But it would be a bad precedent if there were attempts to legislate or regulate away advanced trading technologies. It would be smarter for all firms affected by high-frequency trading to get their middle- and back-operations out of the Dark Ages of Faxes and into the Age of the Millisecond. As for the SEC, the best we can hope for is that it also picks up the pace, which it has been trying to do via circuit breakers and a centralized database that will help it monitor suspicious market activities. The SEC will also have to lure more smart people to its side to keep track of advanced trading technologies. Of course, there are many firms very willing to also hire smart people but pay them very well.
While it would be nice to fantasize about a return to the days when major trading errors were caused by an individual or a group, retro only happens in movies, music and bars, not on Wall Street. No matter what the regulators or industry groups say, high-frequency trading complete with algorithms-upon-algorithms and ever-lower latency links are a part of the landscape, rogue or not. The Knight Capital fiasco is a warning call that all market participants have to come up to speed or they will continue to lose investor confidence.