Understanding markets is not easy, while pulling meaning from financial markets is especially hard. Stock prices move alarmingly fast and far. In our view, the elementary economist’s tool kit of ‘supply’, ‘demand’ and ‘equilibrium’ is not much use. Instead, we believe the tools provided by cognitive science are much more useful; in particular how and why people can be wrong.

Essentially there are two types of error. First, there is random error, or ‘luck’. This is unpredictable. A lot of forecasting is prone to luck. Second, there is bias. Bias is predictable error. We believe bias is just as prevalent as luck in forecasting.

It isn’t easy to differentiate bias from bad luck, but we think three questions can help.

First, what type of problem is it? Hard problems involving prediction and lots of information are more susceptible to bias than easy problems with limited prediction or information.

Second, what type of person is making the prediction? In our view, different types of people are prone to different types of biases. Third, what are the circumstances? Quick decisions are often more prone to bias than slow ones.

We replace the usual economist’s dichotomy of ‘buyers’ and ‘sellers’ with a trichotomy of ‘investors’, ‘analysts’ and ‘managers’.

An investor’s task is to decide which stocks to buy or sell, and when. They tend to frame this task in terms of reward and risk, but we think they are particularly susceptible to bias in the latter.

Most investors think of risk in terms of how much they can lose, rather than the chances they are wrong, and they tend to think harder about risk after they have thought about reward. Both, in our opinion make them susceptible to leaning on the wrong sort of information to form predictions on risk, making them more susceptible to bias.

We think investor bias is more likely to be found at the extremes, where there are signs of unusually high (or low) views about risk. A simple way, thus, to look for investor bias, is to trawl the relative extremes of anxiety, or risk indicators. When looking for investor bias we are especially interested in stocks with low valuations and/or poor relative stock price changes. Both of these are indicators of relatively unusual anxiety.

An analyst’s task is to forecast the fortunes of stocks and advise others to act accordingly. We view their task as subtly, but importantly, different to investors. They are advisors, not risk takers. We think their task attracts a different type of person to investing and throws up different types of biases. We think analysts are much more prone to over-confidence type biases. So while investors think about risk, but often in the wrong way, we think analysts don’t really think about risk at all, and certainly not in the way they should – “what is the risk I am wrong?”. This makes them bad at dealing with their own errors.

A manager’s task is to run their business. Running businesses involves reacting to the environment and making investment decisions along the lines of “should we grow, or contract, or fight to survive”. Such judgements require prediction and the weighing up of rewards and risks. We believe managers are peculiarly susceptible to bias when considering risk, but in a completely different way to investors. They behave more like analysts: they often fail to see risk at all. In our view, they are unusually self-confident and tend to frame most risks as “opportunity”.

The type of people who want to run large companies and the types of behavioural traits required to fight through a complex hierarchy to a position of power, produce, in our view, an element of Darwinian natural selection for bias. Their bias equips them well for their personal ambition, but poorly for the tasks once they are at the top. Hence we view them as the main source of risk.

This is where circumstances become really important. We think managers are, in some sense, hardwired to take too much risk. Fortunately there are circumstances where either they are restricted from following their impulses, or where the consequences of their impulses are relatively benign. This is where we feel safe, and being natural cynics we think such situations are relatively rare.

In our view, the best places to look are in stocks where management’s actions look unusually low risk. A good starting place for this is capital allocation and its outcome – how careful does it look and how successful has it been? This is our framework for understanding stock markets.

Jeremy Lang is a fund manager and co founder of Ardevora Asset Management