For the last four seasons I’ve followed the pundits in their efforts to predict Premier League match outcomes. The stats for the last seasons can be viewed here. I pay particular attention to Mark Lawrenson (BBC) and Paul Merson (Sky) but there are other pundits that predict too.
Our paper on this topic discusses the interesting behavioural quirks that arise when pundits make predictions. As we suspected, the patterns observed are consistent with a boundedly rational view of behaviour, insofar as it concerns making an estimate about the likelihood of a future event.The pundits seem to employ heuristics when it comes to guessing match outcomes. They favour specific score lines, particularly those ones that are easy to remember or imagine and systematically underestimate certain score lines.
The chart below shows the performance of the pundits for this season. After 150 fixtures, the relatively stable ‘50/50 pattern’ emerges. The pundits call approximately half of the matches right.
Predicting the score line prior to any given match is a notoriously tricky task. The pundits, just like you or me, will do well to predict more than one correct score line in any given game week. This comes out in the data time and time again.
The worrying thing from a welfare perspective is that gambling firms specifically advertise score line ‘products’, taking advantage of gamblers poor probabilistic reasoning. As the data shows, just picking the outcome of a match is a difficult enough task, attempting to predict the score line too is extremely challenging and probably best avoided.