Tuesday, December 29, 2009

Aldridge, High-Frequency Trading

First, what Irene Aldridge’s High-Frequency Trading: A Practical Guide to Algorithmic Strategies and Trading Systems (Wiley, 2010) is not. It’s not an idiot’s guide to high-frequency trading, and it’s not a do-it-yourself manual for the small fry self-directed trader who wants to transform himself into the next Renaissance Technologies. (At least not unless he gets a pretty hefty infusion of cash into his trading business and has highly developed quant skills.) Instead, this book is a description of some of the key elements of high-frequency trading—order execution, ways to find trading opportunities, backtesting, portfolio optimization, and risk management. It draws on academic research (some 20 pages of bibliography) and explicates many ideas with mathematical and statistical formulas.

Aldridge’s book is, I think, particularly valuable for the intraday trader who will often have a high-frequency trading system on the other side of his trade. It’s imperative to know how these systems operate and how they can sometimes in what seems a blink of the eye exploit a market inefficiency and, presto, return the market to efficiency. The book also offers useful pointers for those seeking to develop, automate, and monitor their lower-frequency trading strategies.

There’s a lot of meat on the bones of this book. To profit from it directly you have to be quantitatively savvy. To profit indirectly, you need only be intellectually curious and not mathematically challenged.

For the latter group (yes, I include myself) let me share two of Aldridge’s broad-based hypotheses. I suspect that if a trader truly understands and acts on the issues involved in these two hypotheses, whatever his reasoned conclusion as to their validity, he will leapfrog ahead of his competition, high frequency or snail-paced.

First, Aldridge claims that “in the long term, none of the markets is a zero-sum game. The diverse nature of market participants ensures that all players are able to extract value according to their own metrics.” (p. 47)

Second, all traders seek to differentiate predictable price moves from random moves. Aldridge describes some of the tests that can be performed to determine market efficiency—for instance, non-parametric runs tests, autoregression-based tests, tests based on the martingale hypothesis, and cointegration-based tests. She concludes that “the same security may be predictable at one frequency and fully random at another frequency. Various combinations of securities may have different levels of efficiency. While price changes of two or more securities may be random when securities are considered individually, the price changes of a combination of these securities may be predictable, and vice versa.” (p. 89)

I will never be a high-frequency trader, but even for the “slower and duller” Aldridge’s book has a lot to offer. What works at warp speed is sometimes, slightly modified, a winner for those who are entering their trades (even manually) via a cable modem far from the exchange.

1 comment:

  1. While price changes of two or more securities may be random when securities are considered individually, the price changes of a combination of these securities may be predictable, and vice versa.

    True.

    ReplyDelete