One of my new side-interests is the phenomenon of High Frequency Trading (HFT) that has taken hold of world financial markets in the last half-decade. HFT is a subset of Algorithmic Trading, the practice of allowing a computer program to place market orders based on current market conditions as well as a pre-determined strategy. Market making and statistical arbitrage are two of the general strategies employed for this purpose.
However, HFT takes algorithmic trading to a whole new level. Proprietary trading firms now use custom software to make millions of split-second trades per day. In fact, analysts believe that HFT comprised 73% of all trades in 2009. Wow. That’s quite the feat in an industry where just decades ago all trades were placed by actual human beings waving their arms around [see http://tradingpithistory.com/].
Of course, I look at this topic and see an excellent computer science problem (didn’t see that one coming did you?!). How do you reliably execute trading strategies in an environment where milliseconds of latency can cause millions of dollars in profits to swing to your competitors? By co-locating your servers as close as possible to those that process market orders, that’s how. By paying computer engineers millions of dollars to squeeze every ounce of performance out of server operating systems. And the list goes on. In fact, I’ve read that Goldman Sachs has a internal department dedicated to finding and leasing valuable datacenter space close to electronic financial exchange centers such as those in northern New Jersey. Do you think the founders of Goldman ever envisioned that their firm would be interested in thousands of square feet of windowless, freezing-cold computer rooms?
The net result of all this effort is that HFT makes investment banks and proprietary trading firms billions of dollars per year in profits. However, the debate over whether HFT is good or bad for individual investors rages on. Some claim that it provides necessary liquidity to the markets while others believe that HFT simply causes prices to be more volatile than ever before.
Regardless of the answer, consider my mind blown.