On the 28th of August (here) 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 on a Friday before the weekend ties.
Given that Paul will make 380 predictions over the course of the Premier League season (70/380 to date), we have a rare opportunity to analyse the accuracy of a football pundit who systematically predicts.
We’re 7 gameweeks into the EPL season – how’s Merse getting on?
After 70 predictions he has called the correct score line 5 times, predicted the right result 32 times and has been incorrect with the result 38 times. Paul’s pie chart is below and shows the percentages. Quite interestingly, since I began collecting the data I've observed Merse's preferences are commonly time-inconsistent; he often reverses his Friday predictions a day later when he sits down with Geoff et al for matches on Saturday afternoon in the Sky Sports studio.
I also put the random number generator to work for all these ties, generating two numbers between 0 to 5 for each fixture to see how ‘the chimp’ gets on. The pie chart for randomness is the second chart below. Randomness is behind Merse when it comes to predicting the outcome (23 right, 47 wrong). In terms of estimating the score however the random number generator is only marginally behind Merse and has predicted 3 right score lines in comparison to Merson's 5.
When it comes to predicting the score line of ties it seems that randomly generating two numbers between 0-5 is not far behind (and maybe just as good as) relying on our own knowledge.