January was traditionally a good time to pick up bargains, on the stock market as well as anywhere else. This “January effect” was that US stocks rose much more in January than in any other month. Sidney Wachtel discovered the phenomenon in the 1940s, but it wasn’t until the 1970s that anybody took much notice. Subsequent researchers have made refinements and produced ingenious explanations, but the main interest in the January effect is as a challenge to the efficient markets hypothesis.

This hypothesis suggests you can’t beat the market without inside information. That’s because all publicly available information is taken into account in the market price. A rule which says “buy on 31 December and sell on 31 January” shouldn’t yield spectacular returns. But it has.

A simple way to understand the efficient markets hypothesis is to think about the problem of finding the shortest queue at the checkout. I never bother. The way I see it, if there were an obvious shortest queue, other people would have gone to stand in it already and it wouldn’t be the shortest anymore. Similarly, cheap shares have been snapped up and are no longer cheap.

But taken to extremes, the efficient markets hypothesis must be false. If it were impossible to beat the market then nobody would try, and if nobody tried it would be easy to beat the market. In other words: if nobody bothers to look for ₤50 notes dropped on the pavement, there will be more than enough lying around to justify the effort of doing so.

Yet the efficient markets hypothesis is much more convincing than it looks at first glance. In the supermarket, the efficient market hypothesis only holds if most supermarket shoppers are rational and the remainder make mistakes only at random. The random mistakes would tend to cancel each other out, and the rational shoppers would keep the queues equal by seeking out any short ones.

In fact real people make systematic mistakes, not just random ones. That might ruin things in the supermarket, but not in the stock market. In the supermarket a group of elite queue “arbitrageurs”, trying to exploit different queue lengths, could not equalise queues when faced with hordes of ignorant shoppers who irrationally favoured aisles three and four.

In the stock market, when smart investors can take advantage of other people’s stupidity (for instance by buying cheap shares in December and offloading them in January) then these smart investors get richer and more influential. The irrational investors may be numerous but they will also be impoverished and inconsequential.

This arbitrage behaviour is only limited by its inherent risks. Risk-free arbitrage would require some shadow portfolio that was free of whatever error was being made - they would simply sell the “no-January-effect Wilshire 5000” in December and buy the “January-effect Wilshire 5000” to replace it. The shadow portfolio doesn’t exist.

True arbitrage does not involve taking risks, but without a shadow portfolio, exploiting errors in the stock market does: rather than guaranteeing a profit, suddenly the smart investor is simply hoping for one.

That is why markets are almost efficient, but not quite. And in case you’re wondering why I didn’t tell you about the January effect in December, when the knowledge might have made you a little bit of money, don’t worry: the January effect appears to be a quaint piece of history. The arbitrageurs have made sure of that.