In the current era of financial services re-regulation, high frequency trading has gotten caught up in part of the wave and landed under the microscope. This article is intended to address the misperceptions and a degree of hysteria that has surrounded high-frequency trading of late–although the hype has subsided as more industry professionals speak up in favor of high-frequency trading.

First, to correct a common misunderstanding about high-frequency trading: many traditional brokers with little electronic trading experience tend to confuse “high-frequency trading” with “low-latency trading.” Low-latency trading involves routing trades to the exchange with a minimal delay with the goal of arbitraging potential mis-pricings. Few buy-side institutions deploy low-latency strategies, as they find it difficult to compete with broker-dealers in this space. In contrast, “high-frequency trading” refers to accurate short-term forecasting, typically executed in a systematic, computerized fashion.

According to the survey of hedge fund managers we conducted at FINalternatives this past summer, 84% of fund managers define high-frequency trading as investing with position holding times of up to one day. Overall, there are four broad classes of high-frequency trading, two of which involve market making in which positions can indeed be held in seconds and sub-seconds. The other two types are statistical arbitrage and event arbitrage that retain positions for several minutes and up to one day.

A trading frequency is defined as the position holding time–the period of time between position open and close times. In this article we measure the profitability of three trading frequencies against one trading strategy, trading the S&P 500 on the announcements of the U.S. Leading Indicators Index.

We define low, medium, and high trading frequencies as position holding periods of one month, day, and hour, respectively. We find that trading the S&P 500 on leading indicators is profitable at low trading frequencies and high trading frequencies, but is not cost-effective at the most common medium trading period. To capture maximum profitability for their clients, advisors should therefore consider trades with frequencies other than the most common daily holding periods.

Now, let’s talk about ways to trade the S&P 500 on the basis of the U.S. Leading Indicators Index. The index is a composite of ten indicators that historically preceded peaks and troughs in the U.S. economy. The composition of the Index changes through time, and may include the following indicators: interest rate spread between a U.S. 10-year note and the Fed funds rate, average weekly initial claims for unemployment insurance, average weekly manufacturing hours, index of supplier deliveries (vendor performance), stock prices, and manufacturers’ new orders for non-defense capital goods.

To assess the impact of the Index announcements on the S&P 500, we conduct event studies. We first identify all dates and times of leading indicators in the past two years. Next to the date and time of each announcement, we record the following three figures: change in the leading indicators index in the previous month, consensus forecast developed by several economists prior to the announcement, and the actual realized change in the index that was revealed in the announcement.

In addition to the event announcement dates and content, we also record the changes in the S&P 500 around the event announcement “window.” These are the changes in the S&P 500 price:

  • From the last trading day of the month prior to the announcement month versus the last date of the month when the announcement occurred
  • From the evening of the trading day immediately preceding the announcement day versus the evening of the announcement day
  • From the opening price on the day of the announcement versus the closing price 1 hour after the announcement occurs

Next, to estimate the impact of the announcement on the price changes in the S&P 500, we conduct a regression analysis. This analysis can be performed using the regression functionality in the Excel. We regress the changes in the S&P 500 on different aspects of the event announcements.

From the regression results, we find that at the monthly portfolio rebalancing frequency, the S&P 500 moves in tandem with the previous month’s change in the leading indicator index with probability of 87%. Thus, if the leading indicator index increased in the previous month, S&P 500 is expected to rise this month.

At daily trading frequencies, such a relationship does not hold: neither prior nor concurrent changes in the leading indicators have any bearing on the daily changes in S&P 500.

At hourly frequencies, however, the price of the S&P 500 moves with the unexpected component of the announcement: if the actual figure announced is higher than the consensus, S&P 500 rises with 90% probability; else, if the actual leading indicator index shows lower values than the expected value, then S&P 500 is likely to fall. Our results are consistent with several academic studies written on the topic, including one published in 2003 in the American Economic Review, volume 93, “Micro Effects of Macro Announcements: Real-Time Price Discovery in Foreign Exchange,” by T.G. Andersen, T. Bollerslev, F.X. Diebold and C. Vega.

The moral of the story is simple: the medium-frequency daily trading space is just too crowded with traders to generate positive profitability. To make money, one needs to break from the herd and wander off to greener pastures, of the lower- or higher-frequency variety.

Steven Krawciw is a vice president of private banking at Credit Suisse AG. Irene Aldridge is a managing partner, Able Alpha Trading, LTD. She is also the author of High-Frequency Trading: A Practical Guide to Algorithmic Strategies and Trading Systems, (Wiley 2010 in the U.S.)