By Sy Harding
October 5, 2012
Twice a year, in April and October, I remind you of the market’s remarkable seasonality, the popular version of which is known as ‘Sell in May and Go Away’. It calls for getting out of the market on May 1st each year and back in on November 1st.
As with most investment strategies, most investors have only short-term thoughts regarding it. If it worked out the previous year or two, “Well just maybe I’ll consider it for next year.” And if it didn’t work out the previous year then clearly it’s either just a silly theory, or a strategy that may have worked in the past but the pattern has obviously come to an end.
And like all strategies, especially buy and hold, it doesn’t work in every individual year. But it doesn’t have to in order to produce remarkable outperformance over the long term. That’s because in years when the market makes more gains in the unfavorable season when a seasonal investor is out, the seasonal investor doesn’t have a loss, but merely misses out on additional gains. But when the market does have a correction in the unfavorable season, its losses can be well into double-digits, which the seasonal investor avoids.
It’s a shame more investors don’t take the time to obtain the facts.
The seasonal effect is so pronounced that investing based solely on those calendar dates succeeds in the difficult task (even for professionals) of outperforming the market. And it does so while taking only 60% of market risk, a very important consideration.
You don’t have to take my word for it. Independent academic studies provide indisputable proof.
For instance, a 27 page academic study published in the American Economic Review in 2002 concluded, “Surprisingly we found this inherited wisdom of Sell in May to be true in 36 of 37 developed and emerging markets. Evidence shows that in the U.K. the seasonal effect has been noticeable since 1694. . . . The risk-adjusted outperformance ranges between 1.5% and 8.9% annually depending on the country being considered. The effect is robust over time, economically significant, and unlikely to be caused by data-mining.”
And a new 54-page study by Ben Jacobsen and Cherry Y. Zhang at Massey University in New Zealand, was just released a few days ago. It’s titled The Halloween Effect: Everywhere And All The Time. It refers to the ‘Sell In May’ pattern as the ‘Halloween Effect’, selling May 1 and re-entering the day after Halloween, October 31.
It confirms and adds to the findings of previous studies. A few quotes from it: “Observations over 319 years show November through April returns are 4.5% higher than summer returns. The effect is increasing in strength. Over the last 50 years the difference between the two periods is 6.2%. It does not disappear after discovery, but continues to exist even though investors may have become aware of it. . . . It is significant in 35 countries . . . stronger in Europe, North America, and Asia than in other areas. . . . The odds of the strategy beating the market are 80% for horizons over 5-years, and 90% for horizons over 10-years, with returns on average of around three times higher than the market.”
I’ve always given credit for the discovery and coining of the phrase ‘Sell in May and Go Away’ to researchers in the 1970’s. But this new study reports “a mention of the market wisdom “Sell in May” in the May 10, 1935 issue of the Financial Times, and the suggestion that at that time it was already an old market saying.”
But the market obviously does not roll over into a correction exactly on May 1 each year, or begin a new favorable season rally on November 1 each year.
So the Street Smart Report seasonal strategy, developed in 1998, incorporates the MACD technical indicator (Moving Average Convergence/Divergence) to more closely identify the seasonal exits and re-entries.
It is a significant improvement over the basic Halloween Indicator. Under its rules an exit signal can come as early as April 16, but will be delayed if MACD remains on a buy signal at the time. In the fall, the re-entry can take place as early as October 20, but will be delayed if MACD is on a sell signal at the time. Of interest as we enter October, its re-entry signals have been as early as October 20, but also as late as November or even early December.
Mark Hulbert, of Hulbert Financial fame, has been tracking various versions of seasonal timing strategies since mid-2002. In an update in a current article on MarketWatch, he reports that the Street Smart Report version of seasonal timing has gained an average of 8.5% annually since mid-2002, compared to the Halloween Indicator’s average annual gain of 6.9%, and the market’s average gain of 5.7%, and while taking only 60% of market risk.
This year, seasonality seemed to be working out right on the button when the Dow topped out on May 2 and by June 4th the S&P 500 was down 9.1%. But the big rally off the June low has the Dow now recovered and 2% above its May 1 peak. So this might be one of those years when seasonality does not work out.
And yet, with at least the Street Smart Report’s seasonal signals out of the unfavorable season sometimes coming as late as late November, there’s still plenty of time for a correction first, before a favorable season rally to new market highs by next spring.
Either way, investors would do themselves a big favor by checking out the facts about seasonality. Click here to read the latest independent academic study http:/ssrn.com/abstract=2154873
Sy Harding is president of Asset Management Research Corp., and editor of the free market blog Street Smart Post. Follow him on twitter @streetsmartpost.