Economists at the University of Siena studied over a million bets placed upon Serie A games and found something curious – that bets were more likely to lose the closer they were made to the time the game kicked off.
What's going on here is what Alexander Todorov, a psychologist at Princeton University, calls the "illusion of knowledge". More knowledge, he says "can decrease accuracy and simultaneously increase prediction confidence". For example, gamblers who place bets many days before games use simple but powerful heuristics, such as 'Milan are better than Cagliari'. But gamblers on the day of the game have more knowledge about team selection, the referee, the mood of the squad and so on. Such knowledge, though, does not give them predictive power. It's just noise. But mistaking it for useful information, people trade on it.
Sometimes, then, knowledge can increase our confidence more than it increases our ability, with the result that we lose money.
What's true for Italian gamblers can also be true for British stock-pickers. It's possible that as we research stocks or economies more, what we gather is not knowledge that helps predict results, but rather the sort of noise that makes us overconfident. It gives us what the Nobel laureate Daniel Kahneman calls the illusion of skill.
Research by Christophe Merkle of the University of Mannheim has found that investors are prone to this. He studied the behaviour of clients of Barclays Stockbrokers (who are unlikely to differ much from other investors), and uncovered three different types of overconfidence, although the three are only slightly correlated with each other:
■ Over-estimation. Investors expect higher returns than they actually get.
■ Over-placement. Most people believe they are better informed and more skilful than the average – which, while theoretically possible, is unlikely.
■ Over-precision. Investors attach too narrow confidence intervals to their forecasts for returns; they underestimate risk and uncertainty.
These three types of overconfidence have material effects. Investors prone to over-precision – and 88 per cent of those surveyed by Mr Merkle claimed to have better knowledge of financial conditions than the average person – were more likely to trade frequently, something which is known to lose money. Those prone to overestimate returns or to be overly precise about such estimates tended not to diversify much; why spread risk if the market's going to rise a lot or if there's little danger of it falling?
Now, the sins of over-trading and under-diversifying are not necessarily mortal ones. Many shortfalls from fully optimal behaviour have quite small costs. But they can be avoided, if we realise that some cognitive biases are ubiquitous. The problem, though, is that overconfidence blinds us to its existence.
