Something very important happened in federal appeals court this week. Though it was completely unrelated to the current Obamacare rumble in the Supreme Court, its repercussions could nevertheless have a similarly outsized effect on the market and its stability.
If you remember 2010's flash crash, you know that high-frequency trading, or HFT, can (and will) occasionally break down, with dangerous consequences. The trading firms behind this electron-fast method of penny-plucking have advantages we mere mortals could only dream of. But no matter how fast and no matter how smart the algorithms behind HFT happen to be, they're still designed and perfected by human beings. Now a high-profile ruling in a major criminal case has opened the door to the possibility that these imperfect systems could fall into the hands of the unprepared.
Let's start with what's wrong about this precedent within the small group of HFT firms that control much of the world's automated trading, and then take a brief look at what a broader interpretation could mean. I hope that, in the end, you'll disagree (as I do) with the ruling from the Second Circuit Court of Appeals. Heck, even the Second Circuit found its decision a bit unpalatable.
A bit of background
I wrote last year about efforts to rein in HFT's more egregious errors, focusing briefly on the case at issue. Sergey Aleynikov, a onetime Goldman Sachs
The National Stolen Property Act, and to a more important degree the Economic Espionage Act, or EEA, were used to charge Aleynikov, but the language of those statutes also wound up exonerating him. Under the EEA, Aleynikov was charged with downloading a "trade secret" product designed for interstate commercial purposes. Under the NSPA, he was charged with taking "stolen goods" across state lines.
Here's where things get wacky. Aleynikov's original conviction found that the code was indeed a valid product and a stolen good under those statutes, and he was thus sentenced to 97 months in prison. But in overturning the conviction, the Second Circuit sided with Aleynikov's argument that the specific HFT software was neither a "product" nor produced for or placed into any form of interstate commerce. Words matter to lawyers and judges, and the wording of each statute allowed enough wiggle room to overturn the original conviction. The majority opinion made certain to point out that the wording of the EEA could stand to be improved and urged Congress to revisit it in the near future.
Seeds of a bigger problem?
The most immediate concern over this ruling should be regarding HFT software. As the court's decision states, "It usually takes years for a team of programmers to develop an HFT system from scratch." Not only that, but large companies may pay millions per year to have their technology as close to a stock exchange's servers as possible, a concept called colocation that can shave invaluable milliseconds off each trade. NYSE Euronext
The recent implosion of the supposedly tech-savvy BATS Global Markets IPO on its own exchange underscored what many might suspect but few understand: Software-related market glitches happen with startling regularity, with stock exchanges having suffered communications breakdowns nearly once every other trading day last year. Trading halts, which kick in when a stock spikes or drops by large amounts in seconds, happened more than once every trading day. And that's just with the current crop of extremely well-financed and well-connected market movers. Creating a free-for-all in HFT software among lesser financial firms could be a disaster.
Agree to disagree
We're living in a 21st-century world, but many of our laws are built on a 20th-century understanding of technology. Goldman Sachs should be understandably incensed to find that a highly complex software product, developed at significant expense and used to conduct millions of trades a day with stockholders around the world, is not considered a product serving interstate commerce under federal law.
But other companies should be aware of this ruling as well. The efficient operation of in-house software is absolutely critical to the success of nearly all the world's best companies. At least under current law as interpreted by this court ruling, this loophole could put that proprietary software at risk -- and with it, the success of entire companies.
Laws move slowly, but technology advances with frightening speed. This latest ruling is only another in a series of fumbles by legislators, lawyers, and regulators as they attempt to keep the machines under control. Let's hope that something can be done to remedy the holes this ruling creates, before some enterprising programmer with substandard ethics decides to drive a truck through one of them.