Let's consider the good ones first. The fact that the All-Share index has underperformed cash between May Day and Halloween since 1966 - the index has delivered a real total return of minus 0.7 per cent on average - is only an average effect. There's lots of dispersion around this, and history tells us that the most common return in the summer is a positive one; 28 of the last 49 summers have seen a positive real total return. (The average negative return has occurred because big falls are more likely than big gains.)
There's also a problem here of falsifiability. 'Sell in May', like all interesting propositions about investing, is only a statistical tendency surrounded by noise. It can't be refuted by a single observation: even if the market does well between now and Halloween this would still be consistent with low average returns over summer months. It is therefore not an infallible guide to what will happen in the next few months. It's quite reasonable for a moderately risk-tolerant investor to want to bet upon the modal outcome (a small positive gain) rather than the average gain. This is especially true if you incur dealing costs or tax liabilities by selling.
What's more, 'sell in May' challenges two powerful prior beliefs. One is that greater risk must mean greater average return. The fact that equities have underperformed cash since 1966 controverts this common sense. The other is that the market follows a random walk and so should be unpredictable.
It's possible, though, that both these priors are mistaken. Eric Falkenstein has suggested that there is in fact no general trade-off between risk and return: the good performance of defensive stocks is consistent with this. And there are some indicators that can predict returns especially over long periods: the global money-price ratio; dividend yields; consumption-wealth ratios or foreign buying of US equities, for example. Why shouldn't another indicator provide a little predictability about shorter-term returns?
I fear, though, that there is also a weaker motive for scepticism about 'sell in May'. The likeliest reason why it works is that investors are prone to seasonal variations in risk appetite; they become risk tolerant in the spring and nervous in the autumn. This challenges the self-image of financial 'experts' as hard-headed men of sound judgment.
However, we know that changes in appetite for risk matter a lot. Yale University's Robert Shiller won a Nobel prize for pointing out back in 1981 that share prices were five to 13 times more volatile than their 'fundamentals', the discounted value of future dividends. This tells us that attitudes to risk are responsible for most of the variation in returns. Why shouldn't some of this variation be seasonal, given that the winter blues is a recognised medical condition and that our culture has seen such variation for centuries, for example in May Day and Halloween festivals?
The fact that you wear an expensive suit and have a pompous boring demeanour does not mean you are immune from swings in sentiment. Paul Krugman coined the phrase "very serious people" to mock so-called experts who know less than they pretend. Nassim Nicholas Taleb uses the expression "empty suits" to describe people who think they're experts but aren't: he includes stockbrokers among these. We should, then, be open to the possibility that the men in suits aren't as rational as they pretend.
Let's, though, remember the big fact. The evidence for seasonality in equity returns is stronger than it is for most financial phenomena: Ben Jacobsen and Cherry Zhang have shown that it exists in most markets over most periods. I suspect that only momentum investing has a stronger evidence base. The evidence for it is certainly stronger than the evidence for the idea that one can earn great returns simply by trusting your money to men of 'good judgment'.
