Robert Metcalfe and colleagues at the University of Chicago asked foreign exchange traders to test a new trading platform. Alongside the usual assets on this platform, they put an artificial asset which they called a leveraged dollar fund. This was constructed to be an especially attractive trade, as it rose a lot when the US dollar rose but fell only a little when it fell.
They then split the traders into two groups without their knowledge. One got second-by-second prices of the artificial dollar fund; the other got prices only every four hours. After two weeks of trading - long enough to learn that the fund was a good trade - the traders who got four-hourly price data held much more of the dollar fund than those who got second-by-second data, and made 50 per cent higher profits as a result.
This is striking evidence that professionals are prone to a bias known as myopic loss aversion - the tendency to avoid good assets because they look bad over short-term horizons.
To see how this works, imagine an asset with 10 per cent annual returns with a standard deviation of 20 percentage points, with these returns being serially independent. Such an asset is twice as good as equities have been since the early 20th century, so it would be a great investment. However, if you look at its prices every day, you'd see a 49 per cent chance of a loss, whereas it has only a 17 per cent chance of a loss over a three-year period.
If you look at its price every day and are especially concerned not to lose money you'd infer that this was a bad investment. You'll therefore own less than you would if you looked less often.
This is myopic loss aversion. It's myopic because you look only at the short term. And it's loss aversion because you're focused on avoiding losses.
This matters. Back in 1995 Shlomo Benartzi and Richard Thaler suggested that this was the reason why shares did so well over the long run; potential investors who looked at daily prices saw a big risk of a fall and so avoided them, with the result that shares were generally underpriced, thus offering nice returns to the investors who were brave enough to buy them.
The fact that professional traders are indeed prone to myopic loss aversion increases the credibility of this explanation.
But there are two other implications. One is that the so-called experts can be as error-prone as the rest of us. Professor Metcalfe's work corroborates a claim made by Richard Nisbett and Lee Ross back in 1981 in one of the first studies of cognitive biases, Human Inference - that experts make inferential errors just as the untrained do. You might want to bear this in mind next time you take financial advice (or read one of my articles).
Secondly, this warns us that even reliable information can mislead us when it interacts with fallible judgment. Those traders who got second-by-second data were better informed than those who got four-hourly data. But they made worse decisions as a result. Professor Metcalfe says: "Even though traders prefer more frequent feedback, they are certainly worse off when they receive such information."
