My table shows the performance of the Halloween indicator since 1986 for some asset classes. This shows that:
■ Equities are strongly seasonal, with the All-Share index doing 8.2 percentage points better in winter months than summer ones.
■ Smaller stocks are more seasonal than bigger ones. For Aim stocks, the winter premium is 12.7 percentage points. For small caps, it is 12 percentage points. But for the FTSE 100, it is 7.2 percentage points.
■ Commodities do slightly better in winter than in summer.
■ Gilts do slightly better in summer than in winter.
■ Equities underperform gilts in the summer, and Aim and small caps, on average, fall then. This violates the first rule of finance – that higher risk should mean higher average returns. "You've got to speculate to accumulate" is true only in winter.
|Assets' seasonal returns|
|FTSE small caps||-0.6||11.4||12.0|
|Summer is from May 1 to Oct 31, winter is from Oct 31 to May 1. Data are total returns since 1986, except for Aim, which is from 1997.|
What can explain these patterns? The obvious explanation is that our appetite for risk - or, if you prefer, investors' sentiment - is seasonal. As the weather improves and the nights get lighter in the spring, we become more cheerful and optimistic. This causes us to bid up share prices in March and April, often to levels from which subsequent returns are poor. And in the autumn, the darker nights make us gloomier and reluctant to take on equity risk, which causes shares to be under-priced.
Stocks which are more sensitive to sentiment, such as smaller ones, are more prone to these seasonal patterns.
Because gilts and commodities are safer haven assets, they are not vulnerable to such swings in sentiment, and so don't experience such significant seasonal swings.
This poses the question: will the "buy on Halloween" rule work this time? One reason to fear it might not is that "sell on May Day" didn't work this year; Aim and small caps has done well since May, although the All-Share generally has been closer to its average summer returns.
Precedent, though, suggests that good summers don't presage a breaking of the "buy on Halloween" rule. Aim and small caps did well in the summers of 1999, 2003 and 2009, but rose further in the following winter.
What's more, there's a good theoretical reason to suspect that seasonal patterns won't disappear as investors learn about them. It's that the same nervousness and anxiety that causes shares to be under-priced in the autumn also prevents smart investors from buying sufficiently to eradicate this under-pricing.
We're left, therefore, with a general question. Do we prefer using individual judgment about a one-off market conjuncture – there are always reasons to be cautious about buying – or do we instead favour using rules that have, on average, a good track record? My personal preference is to go with the rules. For this reason, I'll be raising my exposure to the All-Share next week.