Friday, 24 March 2017
Last updated 9 hours ago
Dec 23 2009 | 10:13am ET
By Irene Aldridge -- High-frequency trading has generated plenty of controversy over the past few months. Even after many heated on- and off-air discussions, the distribution of opinions on the subject remains highly bimodal. Some argue that high-frequency trading is a parasitic form to be exterminated, while others (the author of this article included) contend that high-frequency trading will play a core role in the next generation of investing strategies.
Numerous financial studies and theories have noted that short-term forecasting invariably works better than long-term predictions. Shorter horizons are accompanied by a lower degree of uncertainty, enhancing the predictive power of forecasts. Viewed from this angle, high-frequency trading is nothing more than a technology-fueled exploration of how small time horizons can actually improve the predictive power of securities pricing.
Contrary to popular misconceptions, high-frequency trading strategies are defined by the frequency of position holding, not by the speed of order execution. (The latter strategies based on the fastest order execution speed are known as low-latency strategies.) While fast execution is essential in today’s markets, and many brokers guarantee order execution speed in 20 milliseconds or less, it is the frequency of buy and sell signals that separates high and low-frequency trading decisions. As such, many conventional quantitative and systematic strategies can easily be transformed into profitable high-frequency strategies. Take a systematic trading strategy, reduce the holding period, et voila! (To be fair, the low-to-high frequency adoption is often a one-way street: some types of high-frequency strategies do not translate to lower frequencies.)
Naturally, high-frequency trading relies heavily on technology, which is often absent at many traditional hedge fund shops. Many funds have depended on analysts to generate trading signals using low-tech methods: a persuasive analysis to justify investments into a particular security or a manually triggered MatLab program. This approach has its limits. As conventional statistical arbitrage strategies are sped up to holding periods counted in minutes, it becomes difficult or impossible for a human trader to manually generate forecasts at such speeds.
The costs of computer hardware required in an automated high-frequency platform are often very reasonable, and certainly lower than those put forth in the media. To put it simply, a computer equipped to run today’s generation of online video games is perfectly suitable for running a high-frequency trading system, and will set you back only a few hundred dollars. Yet, the sky is always the limit as far as the technology costs are concerned, and some may prefer to buy advanced servers for several hundred thousand dollars apiece, although the marginal benefit of such expense cannot be justified in many cases.
Co-location, a popular topic, can also be integrated into a trading platform without much expense. Many brokers now offer this service free of charge, hoping to profit from sell-side transaction revenues.
Perhaps the most costly component in building a high-frequency system is finding and retaining specialized staff capable of programming trading ideas. Total compensation schedules for such personnel can run upward of $500,000 per year, and, worst of all, once someone lures them away with an even higher paycheck, their intellectual capital leaves with them. In response, some firms require strict non-compete agreements to discourage their employees from transferring their knowledge to other firms. Others use brokers’ “plug-and-play” solutions, where instead of building the whole infrastructure from scratch, traders extend the code offered to them by their trading counterparties. Some providers offer “drag-and-click” visual interfaces for building high-frequency trading strategies, although sophisticated strategies may still require expertise in a programming language.
Given these technological issues, traditional hedge funds have been particularly cautious of the high-frequency tide. Those funds with existing quantitative strategies are likely to be ideally positioned to move into the high-frequency space: speeding up the portfolio holding periods is all it takes in many cases. Finding a profitable high-frequency strategy, therefore, may only be a few clicks of the mouse away.
Irene Aldridge is a managing partner at ABLE Alpha Trading and the author of the new book High-Frequency Trading: A Practical Guide to Algorithmic Strategies and Trading Systems (John Wiley & Sons).