Recently, there's been a challenge to hedgehogs. Philip Tetlock, a political scientist at the University of Pennsylvania, has shown that in forecasting foxes beat hedgehogs by a long way.
He established this through an experiment called the Good Judgment Project, in which volunteers are asked questions such as: will Assad still be president of Syria in March 2017? Who'll win the French presidential election? Or what will be the price of oil in November 2016?
A few people, he's found, are "superforecasters", who answer these questions better than others. They do so, he says, by being foxes - gathering fragments of information from many sources and aggregating different perspectives. Hedgehogs, by contrast, do badly because they see the world through the lens of a grand theory and ignore evidence that doesn't fit that theory. The talking heads you hear in the media are usually hedgehogs.
This suggests that investors should be foxes. We should abandon big theories and try to gather dispersed information.
Except it doesn't. There are big differences between the Good Judgment Project and stock-picking.
One difference is that investors already have a fantastic device for gathering dispersed information and aggregating diverse perspectives. It's called the market. As Friedrich Hayek said, prices represent the aggregation of "dispersed bits of incomplete and frequently contradictory knowledge". The fox's job is therefore done for us if we buy a tracker fund.
From this perspective, there's a massive difference between the Good Judgment Project and stock-picking. The GJP is an attempt to find the right answer. But in stock-picking we need to avoid errors. Our task is not so much to find the very best stock as to avoid unnecessary mistakes. And we can do this by using hedgehogs' grand theories. The efficient market hypothesis helps us avoid wasting our money on high-charging funds. And the maxim 'people are overconfident' warns us not to take big positions and to diversify.
There's another difference. In the GJP information is either useless because it's irrelevant or of value. In investing, however, it can be worse than useless. If a fact is already in the price but we trade as if it were not, we'll lose. Information can give us an illusion of knowledge and overconfidence. And it can mislead us in another way: frequent price data can give us an impression that an asset is riskier than it in fact is.
To put this another way, in the GJP information is a parameter; it has given value and the trick is to discern that value. In investing, however, information is strategic; its value depends upon whether others have it or not. If we assume they don't have it when they do, the information might lose us money. This happens in flotations; investors fail to see that the businesses' owners have inside information which they use to sell at the best possible time and so they pay too much for those businesses.
Investors therefore face a much tougher challenge than Professor Tetlock's subjects. The latter must merely gather data. We must ask: is this already in the price or not? This is an altogether trickier task.
Here, the hedgehog has an advantage. Our grand theory, the efficient market hypothesis, warns us that investors know everything. The hedgehog will therefore avoid underperforming the market except to the extent of tracker funds' fees. That's not too bad.
And we can do better. The hedgehog who knows another big thing - that momentum stocks usually beat the market - knows that investors typically underweight information about past performance. This gives him an edge. And he gets another edge for knowing another big fact - that defensives also tend to outperform.
In this sense, the hedgehog who knows one big thing - that markets are usually efficient except for momentum and defensives - should do well.
We should not, therefore, infer from Professor Tetlock's work that foxes always beat hedgehogs. It all depends on context.
In saying this, I am, paradoxically, taking a foxy kind of stand. I'm denying the classic hedgehog view that there are law-like generalisations in the social sciences. Instead, what's true in some domains is false in others. Political forecasting is a different domain from investing.
