Mark Hulbert writing over at Barrons.com reviews the twenty year performance of Norman Fosback’s Seasonality Timing System. Fosback created the model in the early 1970s and popularized the system in his newsletters and his book, “Stock Market Logic”.

While there are a number of seasonality systems, Fosback’s is relatively simple. It takes positions in the stock market prior to market holidays and over month end. Otherwise the system remains in cash. Surprisingly this system has been able to outperform the market with significantly less risk (some 1.4% a year). According to Hulbert it has the best risk-adjusted performance of any system his newsletter tracks.

Of course, no system is perfect. The seasonality system has some significant negatives. Hulbert notes four of them:

First, the strategy involves a lot of transactions — around 17 round trips a year, in fact. So you need to pursue it using investment vehicles that have little or no transaction costs.

This probably means using one of the handful of no-load index funds that levy no additional fees for frequent switching, such as those offered by the Rydex, Potomac, and ProFunds families of funds.

The strategy’s second vulnerability, related to the first, is that it generates lots of short-term capital gains. So, ideally you should follow this strategy in a tax-deferred account. The strategy’s risk-adjusted returns are still attractive on an after-tax basis, but a lot less so than on a tax-deferred basis.

Third, the strategy may stop working once enough investors follow it. Success eventually will undermine all profitable strategies, of course, and there is no reason to expect this one to be an exception.

Having said this, however, I must also concede that this strategy has so far withstood a lot of favorable attention. Fosback prominently mentioned the strategy in his book “Stock Market Logic,” which he wrote in the mid-1970s, and hundreds of thousands of copies of the book reportedly have been sold over the intervening three decades.

Also, I was in the audience at several well-attended seminars that Fosback gave in the late 1980s and early 1990s in which he described his system and its results. And yet, according to HFD data, some of its best years have occurred over the last decade. I also have written about it before, and the system continues to outperform.

Finally, the pattern may be a statistical fluke. I point this out not because I believe it’s true, but because you should always ask whether what has happened in the past is a matter of luck or coincidence.

As far as systems go this one has had a long incubation period. As Hulbert notes it could stop working at any time. There is a great deal of research on seasonality, including our post here.