Monday, October 11, 2010

A solid primer on high-frequency trading

For those who are unfamiliar with high-frequency trading (HFT) robots and how they've taken over the market, this 60 Minutes piece does a terrific job of laying out the key issues. It's the best summary I've seen yet, placing things in layman's terms for non-professionals. If you're invested in this market in any way, it's essential that you understand how HFTs have changed market dynamics and the relevant risks.

I could have excerpted pretty much the whole piece here (it's that much of a must-read), but I focused on some of the basics.
It may surprise you to learn that most of the stock trades in the U.S. are no longer being made by human beings, but by robot computers capable of buying and selling thousands of different securities in the time it takes you to blink an eye. 
These supercomputers...are operating on highly secret instructions programmed into them by math wizards who may or may not know anything about the value of the companies that are being traded...
The players range from firms like Goldman Sachs, Barclays, Credit-Suisse and Morgan Stanley to hedge funds and smaller operations like Tradeworx, which is the only high frequency trading firm that would talk to us or let us in. 
It's run by Manoj Narang and a small group of mathematicians and scientists called "quants," which is short for quantitative analysts. Their high speed computers trade 40 million shares every day. What Narang and other high frequency traders tell their computers to do is to make a profit of a penny or less, 40 million times day. 
They scan the different exchanges, trying to anticipate which direction individual stocks are likely to move in the next fraction of a second based on current market conditions and statistical analysis of past performance. But the computers have no real understanding of who these companies are and what they do. 
Asked if it's all math, Narang said, "It's all probability and statistics - a procedure that you can define precisely."
The problem, of course, comes during particularly volatile periods, when traditional "probability and statistics" break down. During these times of panic, most of these algorithms freak out, and behave unpredictably. In other words, the robots that we rely on to provide 70% of market liquidity are at their worst when we need them the most. That's why any attempt to explain May's "flash crash" without focusing on HFT activities is inherently flawed. As long as HFTs dominate the market, we can expect many more flash crashes.

Note: I tried to embed the video here, but CBS's embed function was screwy, so click here to watch.


[CBS News]

2 comments:

  1. Hello, I just wanted to take a minute to tell you that you have a great site! Keep up the good work.

    ReplyDelete