Those who “sell in May and go away” are the very same people that assign significance to this otherwise absurd rule. Throw the rule around and let it grow to a principle, wait for a few months for prices to fall and then come back and buy at lower prices. This will stop because, as I will show in this blog and contrary to common misconception, this strategy has worked only recently and it is a condition that strong hands will make disappear.
To make a long story short, the strategy “sell in May and go away” can serve as a means of manipulating prices to the benefit of those who follow the strategy and to the detriment of those who do not. In order to analyze this strategy it may be best to consider its compliment, i.e. the strategy “buy in June and sell in October”. A statistical analysis of this strategy may provide clues about its significance.
I used daily S&P 500 Index data and I backtested the strategy. The interesting result was that the win rate of the strategy came out 64.52%, meaning that in the 62 years involved in the test, 40 were winning and only 22 were losing. So why do people think that this is a losing strategy and as a result its compliment, “sell in May and go away”, is a winning one?
The answer lies in just a few outliers in terms of magnitude of losses that all have occured in the last 12 years. Below is the Amibroker (Charts created with AmiBroker – advanced charting and technical analysis software. http://www.amibroker.com/”) trade report for the last 22 years sorted for higher loss:
It may be seen that the biggest losses occurred in 2008, 2001, 2011, 2002, in that order. Out of the 22 years, 15 were winning and just 7 were negative but the losses were high enough to provide a false sense of a general rule. Those few large losses have compelled people to declare this a rule. Are these losses just random outliers or due to structural modes of the market?
I believe these are just random outliers because after using data for S&P 500 since 1950 and repeating the study I ran a bootstrap analysis on the percent price changes (from June to October) for each year. The p-value came out equal to 0.1940 meaning that there is no evidence against the hypothesis that these results are random. Actually, I think they are but some are also trying to use the random outliers to create a profitable condition for them by raising this rule to an investment principle. In this way they can try to buy earlier, possibly during September before the rest do and make profits. I believe this can be taken care of easily by buying a lot earlier, maybe in August. Then, some may try to buy even earlier in July, and you see now, healthy market forces will slowly and naturally close this artificial arbitrage some have tried to create by taking advantage of a few statistical outliers. I give it a high probability that those who will try to take advantage of the rule “sell in May and go away” in the next 5 years will regret it.
Disclosure: no relevant positions.