high-frequency trading

One of the many things that continue to intrigue me is high-frequency trading or algo trading, as it is more popularly known.

Fundamentally, algo trading is automated trading wherein computers are used to enter trading orders, sometimes totally without human intervention. A bigger rationale behind this form of trading – apart from the speed advantage due to computer usage – is that it spreads risk for the institution by executing a large number (to the order of millions!) of low value trades. So even if the markets were to swing heavily on a particular day, algo trading is designed to minimize that impact. How beautiful!

High frequency traders use computers that execute trades in milliseconds; this was some years back. “The execution period has come down to microseconds these days and the latency (professional jargon!) will continue to go down as algorithms get more sophisticated and more powerful computer systems emerge”, says a friend of mine who works with a US-based hedge fund in Gurgaon.

Ultrafast stock trading relies upon computer programs that take years to develop and are closely guarded secrets. Typically spending millions and millions of dollars, companies pay hefty bonuses to their top algorithm makers. A piece of this top-secret software can be worth millions, may be billions in case the whole software is out on sale. This is best illustrated by the recent Sergey Aleynikov – Goldman Sachs (GS) case.

Sergey Aleynikov, who worked in the software unit of GS, allegedly copied 32 MB of their 1600 MB top-secret trading software and uploaded it on a server based in Germany. This was some days before he quit his $400,000 a year job to join a rival paying 3x that amount. How fishy does this sound? Very! Alarm bells ringed all across GS as the theft was discovered and Mr. Sergey was arrested by the FBI in no time. The ensuing battle continues till date.

This is a glimpse of the high-stakes game of high-frequency trading!

-Sumeet Seth (sumeets29@gmail.com)