Join our community of smart investors
OPINION

The merits of randomisation

The merits of randomisation
March 10, 2016
The merits of randomisation

In fact, he's echoing an old idea. Back in 1973 Princeton University's Burton Malkiel claimed that "a blindfolded monkey throwing darts at a newspaper's financial pages could select a portfolio that would do just as well as one carefully selected by experts". Subsequent research, however, has shown that this was too harsh a judgment - on the monkeys. Economists at Cass Business School have found that portfolios of US stocks picked at random would, on average, have beaten the market most of the time between 1968 and 2011. This is because during this time small stocks did well which implies that most shares outperformed capitalisation-weighted indices which were dragged down by big stocks' underperformance.

You might object that Professor Malkiel's claim is just a variant on the idea that stock markets are informationally efficient - that share prices immediately embody all relevant information so there is no point trying to research stocks.

 

 

This misses the point. The best evidence we have that markets are inefficient is the fact that some types of stocks do well: defensives, momentum and quality stocks (defined by objective factors such as profitability). We could pick stocks at random from those that satisfy these criteria; this is a form of constrained maximisation. For example, as I write the IC's stock screen shows that there are 86 stocks with a beta below 0.8; positive earnings per share growth in the past five years; and which have risen 20 per cent or more in the past 12 months. One could pick at random from these.

Doing so exposes you to those factors that tend to pay off over the long-run while also protecting us from some of the cognitive biases that have traditionally cost investors money. I'm thinking of three in particular:

Overconfidence. If you use judgment, brief periods of good performance - which are likely through dumb luck - will lead you to believe you have skill: this is especially likely because we all want to believe we're clever. This can encourage you to become dizzy with success, and so trade too much and take on too much risk, thus worsening your performance.

Herding. Our judgment can be influenced by others: it takes a strong mind (or a pigheaded one) to bet against the crowd. This can lead us to buy stocks when they are fashionable and overpriced, and so sell when they are out-of-favour and underpriced.

The disposition effect. Investors tend to hold on to losing stocks in the hope of getting even. This is a bad idea because it exposes them to negative momentum - the tendency for stocks that have fallen to continue falling. One reason investors do this is that selling at a loss means admitting that they were wrong. Nobody likes doing this. Taking judgment out of stockpicking, however, would allow us to take a more objective view of our shares and so be readier to sell losers.

The case for constrained randomisation in stockpicking, therefore, seems strong. Which poses the question: why then do very few people do it?

One reason, of course, is that some people simply enjoy the process of researching shares. This is entirely reasonable - as long as you remember that only very lucky people get paid to enjoy themselves.

There is, though, another reason. It was discovered by Harvard's Ellen Langer in the 1980s in a photocopying room at the university's library. She asked people waiting in line: "may I use the Xerox machine?" 60 per cent let her go to the top of the queue. But when she asked "may I use the Xerox machine because I have to make some copies?" 93 per cent let her in. This is despite the fact that the second question conveyed no useful additional information.

What's going on here is simple. As Columbia University's Jon Elster has said, "people want to have reasons for what they do". That word "because" gave them a reason, even though it was an otiose one.

So it is with stockpicking. We want to have reasons for the stocks we buy. In a world of uncertainty, limited knowledge and bounded rationality in which a lot of information about stocks is already in the price, many of these reasons will be as irrelevant as Professor Langer's "because" - or perhaps even worse than useless. However, because we find it so difficult to get by without reasons for what we do, we are reluctant to embrace randomisation despite its merits.