Saturday, May 30, 2009

Sell in May? A Follow-up

At the beginning of May, I wrote a post questioning the validity of the old trader's axiom of Sell in May. Although it's a widely-held truism, the previous 29 years of market action in May reveal the exact opposite to be true.

With May 2009 in the books, the contradiction has extended itself.

The NASDAQ has now closed higher in May a remarkable 20 out of the past 30 years -- a 66% occurrence rate since 1980. The SPX has done even better, with a positive close in 21 of the past 30 May cycles (70%).

The streak has strengthened in recent years. Since the 2002 bottom, the NASDAQ has climbed in May six of the past seven years. The gains are statistically significant as well. Since 1980, the average May for the NASDAQ is +1.87%; for the SPX it's +1.51%.

However, despite this track record, predictions of May weakness are trotted out each year as part of the established canon of trader wisdom.

Why the discrepancy?

Perhaps the biggest reason is that Sell in May is the opening gambit in a longer seasonal trading strategy known as the Halloween Indicator. Part of trader lore for generations, in July 2001, Sven Bouman and Ben Jacobsen examined the financial markets of 37 countries, some as far back as 1694. They established that, across three centuries and multiple economies, winter has produced better market results than summer.  So, for 300 years, selling in May and buying back in after Halloween has produced better returns.
The point here is that for 30 years dumping stocks in May has reduced these seasonal returns, and it's a fact inexplicably unnoticed by trading mavens, the mainstream press and the blogosphere alike. For example, in Oct 2007 Mark Hulbert -- normally excellent with these types of observations -- considered ways of improving the Halloween Indicator. However, Hulbert focused on MACD confirmations and never mentioned May performance. This year, Hulbert insisted that the Halloween Indicator remains strong, while again neglecting the May anomaly. It will be interesting to see when the investing community finally recognizes that Sell in June is the better rule of thumb.

It's important to understand that, by their own admission, Bouman and Jacobsen could never establish why financial markets outperform in winter. Looking ahead, it's possible that globalization and the 24/7 interconnection of world economies could impact this centuries-old market pattern to an even greater degree.

As I write, the profitable winter season is just three weeks away  -- in Brazil, Australia, Indonesia, South Africa and every other market south of the equator.

Time to sell?


Anonymous said...

Sell in may means sell by the end of May. Thus, May performance should be included in positive period. Your post suggests sell before May or on May 1st, which is confused.

dk said...

Hi JY...thanks for the comment. My point is that, since statistically May is a productive month, why steer investors out of it at all? To your point, few practitioners say, "Sell before the end of May". Generally, at the first sign of selling, advocates intone the wisdom of adjusting portfolios for purely seasonal reasons. I've seen no one point out that historically May is an above-average month.