New research by Ben Jacobsen and Cherry Zhang of Massey University in New Zealand shows just how powerful is the "buy on Halloween, sell on May Day" rule. They studied the entire history of world stock markets and found that in 81 of 108 national markets, shares did better on average in the six months from Halloween to May Day than they did in the six months from May Day to Halloween. In only two markets was the opposite case in a statistically significant sense: Nepal and Bangladesh.
Averaged across all markets and all history, equity returns are 4.5 percentage points higher from October 31 to April 30 than they are from April 30 to October 31. For example, in the UK since 1693, returns have averaged 2.4 per cent in the November-April months and -1 one per cent in the May-October months - a gap of 3.4 percentage points. In the US since 1791, the gap has been 1.7 percentage points. In Germany since 1870, it has been 5.3 percentage points. And, in Japan since 1914, it has been 8.3 percentage points. And so on.
This is no mere historical quirk. The Halloween effect "has strengthened rather than weakened in recent years," says Professor Jacobsen. Averaged across all markets, returns in November-April have been 5.6 percentage points higher than in May-October since 2001. Though this is a smaller gap than between 1981 and 2000, it's bigger than for any decade since 1711-20.
Of course, this doesn't mean Halloween is an infallible. In the UK, it failed in the winters of 2007-08 and 2008-09, for example. But, in volatile markets, no indicator works perfectly. It's just that the odds favour Halloween as a time to buy.
But why? One possibility is that as the nights draw in we suffer from the winter blues and become anxious and depressed. It's no accident that Halloween has its origins in the festival of Samhain, when our pagan ancestors prepared for winter, unsure of whether they'd have enough food to see them through. Because the human mind is shaped more by adaptation to the tens of thousands of years we spent as hunter-gatherers than it is to the few decades of industrial society, it's likely that we share the instincts of our ancestors. And these instincts tell us to be cautious in the autumn and to prefer safe assets to risky ones. This means that share prices are low and so subsequent returns should, on average, be high.
This explanation runs into a question. Why, then, has the Halloween effect been more powerful since the 1960s than it was before. It can't be that we are more susceptible to the winter blues than our grandparents were.
We're probably not. What might have changed is not so much our emotions, but the stock market. If markets contain more hard-to-value stocks - such as younger, speculative ones without much earnings history - they will be more prone to shifts in sentiment. The winter blues effect will therefore be stronger.
Two facts corroborate this theory. One is that the Halloween effect was especially strong in the 1990s when speculative tech stocks had a high weighting in western markets. If you'd bought on Halloween and sold on May Day between October 1995 and October 2003, you'd have made a return of 149 per cent, compared with 67 per cent from being in equities all the time.
The other is that Aim stocks have an even stronger Halloween effect than mainline ones. Since its inception in 1996, the Alternative Investment Market (Aim) market has risen an average of 4.3 per cent between Halloween and May Day, while the All-Share has risen 2.9 per cent. But Aim has lost an average of 2.5 per cent between May Day and Halloween, while the All-Share has lost 0.9 per cent. This suggests that sentiment-driven stocks - which Aim stocks tend to be to a greater extent than main market ones - are more vulnerable to seasonal shifts in investors' moods. If such stocks are more significant now for mainstream markets than they were before 1960, then you'd expect to see a bigger Halloween effect in stock markets.
The message here is simple. Halloween is a surprisingly robust - though not infallible - buy signal. And it's more powerful for stocks which are more sensitive to investors' sentiment. In this sense, the Halloween signal favours growth and cyclical stocks over defensives.
