The case for doing so is simple; it has worked well historically. Since 1986, the All-Share index has risen by an average of 7.2 per cent from Halloween to May Day and it has fallen in only six of those 31 periods. By contrast, the market has dropped by an average of 0.4 per cent from May Day to Halloween even including this year's 11 per cent rise. And in those summers, we've seen 14 falls.
This is in fact part of a wider pattern. Ben Jacobsen of Tilburg University and Cherry Zhang of Nottingham University Business School show that much the same is true for most national stock markets since records began.
But there's a problem here, of course. The 'sell in May, buy on Halloween rule' has failed recently. Real returns were slightly negative last winter and the market rose strongly this summer. Does this mean the rule has broken?
I don't think so. The rule has never been infallible, but merely a statistical tendency, albeit a strong one. One or two failures should slightly weaken our confidence in the rule, in Bayesian fashion. But it should not undermine it completely, especially as this summer was unusual in other respects.
However, even if we discard the 'buy on Halloween' rule, there's another rule that tells us to buy now. It's the 'buy when prices are above their 10-month average and sell when they are below them' rule proposed by Meb Faber of Cambria Investment Management.
My table shows how this rule would have worked had it been applied to the All-Share index on May Days and Halloweens since 1986. Looking at Halloweens, the average rise in the index after prices have been above their average on Halloween has been 8.3 per cent. That's even better than the performance after all Halloweens. And the index has fallen on only two of the 20 occasions when the Faber rule told us to buy on Halloween: in the winters of 2000-01 and 2007-08. This implies that falls are much less likely than they are in other six-month periods.
| The 10-month average rule on Halloween and May Day | ||||
|---|---|---|---|---|
| Winter | Summer | |||
| Average | 7.2 | -0.4 | ||
| % of falls | 19.4 | 45.2 | ||
| Price>average | Price<average | Price>average | Price<average | |
| Average | 8.3 | 5.2 | 0.2 | -2.7 |
| % of falls | 10.0 | 36.4 | 41.7 | 57.1 |
| Based on changes in the All-Share index since 1986 | ||||
We don't therefore need the 'buy on Halloween' rule to have a reason to buy now.
I confess that this conflicts with my intuition. I fear that markets are pricing in the good news that a weak pound will raise overseas earnings, but not the bad news that higher import costs will squeeze profits.
The historic record, though, rejects this theory. It tells us that momentum matters: when the market has risen, it often continues rising.
I therefore face a dilemma. My instinct tells me one thing, but rules and history tell me another. However, mindful of the fact that there are tight limits upon all our powers of judgement (and especially mine!) I prefer to trust rules rather than my own judgement.
Perhaps, though, the conflict isn't so acute. The same rule that tells me to buy now would also tell me to sell when prices dip below their 10-month average. If we're lucky, the bad news of a profit squeeze won't hit equities until later.
