[The final entry on this post can be accessed here]
On the 28th of August I began recording Sky Sports Pundit and ex-Premier League Footballer Paul Merson’s predictions for EPL fixtures. These predictions are usually published by Sky Sports the day before EPL matches. The first four entries that followed his predictions can be found here, here, here and here. Given that Paul will make 380 predictions over the course of the Premier League season (289/380 to date), we have a rare opportunity to analyse the accuracy of a football pundit and Sky Sports football expert who systematically predicts.
Merse has now predicted the outcome of 289 fixtures and only has 9 gameweeks left. He has successfully predicted the right result 117 times, the right score line 27 times and has been incorrect on 145 occasions. As per the past entries Merson's predictions are compared to the output of a random number generator. Since the last update, the results have stayed mostly the same. Merse is again approximately 15% ahead of randomness.
Also included below is Merson's updated Premier League Table which represents what the EPL would look in light of his estimations. Merson has successfully predicted the standing of the first two teams but has overestimated how many points Chelsea, Man City, Liverpool and Arsenal would have collected by this stage.
Even though he has predicted half of the outcomes incorrectly he still gets the top five correct (although the order is wrong). In terms of the Champions League race, the table would suggest that Merson thinks Liverpool and Arsenal will be 3rd and 4th respectively. Currently, Manchester United are 4th. Merson underestimates the performance of Tottenham and Southampton, suggesting that they would have 9 and 16 less points accumulated respectively. Finally in regards to relegation Merson has one of the three teams in the relegation zone correct – Burnley. They are currently 18th in reality and also from Merson’s estimates. He has underestimated the performance of West Brom.
Although it may appear Merse isn't performing too well, as I showed recently, he is ahead of the BT pundits when it comes to predicting correct outcomes and his table still looks reasonably accurate.
As John Eakins has previously discussed and I introduced some weeks back, we think the predictions of pundits can be explained through the lens of behavioural economics. Basically pundits suggest results that are quite easy to imagine. There is plenty of theoretical work on nonadditive subjective probabilities and our hunch is that these pundits may fall foul to an 'imaginability' bias - in short this goes that imagining an event may cause one to hold an increased subjective probability of it happening. We have a hunch that outcomes that are easier to imagine will be judged as more likely. How many of us go into Paddy Powers to back a 0-0 draw?!
We are also currently asking whether the market believes in this? A issue similar to this has been addressed before in sport. In 1989 Colin Camerer asked whether the basketball market believed in the 'hot hand'. By examining whether perceived hot streaks by fans affected the point spreads in betting markets, Camerer's study found that bets on teams on 'winning streaks' were more unsuccessful while teams on 'losing streaks' tended to over perform. As Camerer notes mistaken beliefs in the market such as the hot hand raise important economic questions as there presence may make the allocation of resources sub-optimal.
I will provide two more updates on this entry - before and after the final fixtures of the season. At the end of the season I will look at the Merson's goal distribution as John Eakins did here for Mark Lawrenson and I will give an update in regards to the findings from our paper where we are looking at whether pundits succumb to this bias.