Undercover Economist

Why predictions are a lot like Pringles

‘Nobody thinks that there’s any great virtue in forecasts but we find them hard to resist’

In mid-December, Phil McNulty, the BBC’s chief football writer, offered us his predictions for the rest of the English Premier League season. My interest in football is limited but I found McNulty’s efforts fascinating. Even the most sceptical about football can learn a great deal from the episode.

A brief piece of context for those sceptics. Chelsea, the champions, had just played Leicester City, a team that had been relegation favourites just a few months before. Leicester won the game. This result would have been surprising had it not been set against the even more surprising pattern of the season. Champions Chelsea had slumped towards the bottom of the league after producing an unprecedentedly appalling run of form; Leicester, meanwhile, were top of the table. Nobody was shocked to see them vanquish Chelsea but it felt like a significant moment nonetheless.

What of McNulty? At the beginning of the season, he had predicted that Chelsea would be champions again, while Leicester would finish in the bottom three and be relegated from the Premier League. Both of those outcomes are now inconceivable. After admitting that his initial prediction had been about as wrong as it is possible to be, McNulty proposed a new set of predictions.

Those predictions were . . . but wait. Why on earth should you care? McNulty knows a great deal about football — far more than I do — but he had conclusively proved that he can’t see into the future. And yet he felt bold enough to offer another forecast, which many sports fans read with great interest.

This is a common pattern in football and beyond: pundits make forecasts, their audience consume those forecasts with relish, the forecasts are proved wrong, nobody is very surprised, and the cycle begins again. Why?

Part of the explanation is wishful thinking: we like to believe that the world runs on rails, and to trust in experts who claim to have decoded the timetable and can therefore explain what is going to happen, when, and why. Forecasters with a record of some success — such as data-driven political and sports analyst Nate Silver — soon find themselves saddled with unrealistic expectations.

Silver correctly predicted the fine details of the 2012 presidential election but he is happy to admit three things: that US elections are data-rich environments and much more predictable than most; that he had some luck; and that the bar for forecasting success had been set very low by partisan pundits much more interested in cheerleading than accuracy.

Sure enough, when Silver and his colleague Ben Lauderdale tried to predict last year’s UK election result, their performance was woeful. This was partly because the seat-by-seat polling data are far less detailed than in the US and partly because Silver’s good luck didn’t last.

We would be wise to have more realistic expectations, even of careful data-driven forecasters such as Silver. But perhaps our expectations are irrelevant. Even when we know that the forecasts are useless, when the pundits have no track record, when the events in question have always been unpredictable (the stock market; geopolitical shocks; recessions), we remain hungry for opinions about the future.

The truth is that forecasts are like Pringles — nobody thinks that there’s any great virtue in them but, offered with the fleeting pleasure of consuming them, we find it hard to resist. I am not sure quite why this should be so, but I have a couple of theories.

Possibility one is that the moment we hear a forecast, we imagine it happening. It then becomes a believable outcome and one that is easy to call to mind in the future. The scenario that we imagine looms large in our minds; other scenarios, equally plausible, fade to the background. As a result, we can be sceptical of forecasts in general yet still hooked by a particular one.

I notice this tendency in myself whenever I hear someone opining on the stock market. As an abstract proposition I think that it’s almost impossible to predict what the stock market will do. But the moment someone starts to tell me a story about what will happen to it, I’m hypnotised.

Possibility two is that forecasts offer us a lazy way to understand a complex world. The background to the conflict in Syria is complicated. So is Chinese politics. So, too, is the evolution of the Japanese economy. Trying to understand what is going on in any of these places requires an investment of time and attention that most of us are not willing to make. Wise heads at this newspaper could explain the intricacies to you or to me for hours yet barely have begun to do the topic justice.

But a forecast? That’s different. A forecast about what will happen in Syria, China or Japan is a simple way to convey a fleeting sense of understanding. The forecast will probably be wrong. But at the instant it is consumed, it gratifies. As I say, a lot like Pringles.

Written for and first published at ft.com.

Subscribe to TimHarford.com

Enter your email address to subscribe to this blog and receive notifications of new posts by email.

Join 4,251 other subscribers