In 2009, on the final day of the football season, David Spiegelhalter, a Statistician, attempted to predict the outcome of the Premier League fixtures using a mathematical model. He correctly identified the outcome (win, lose or draw) of 9 matches and the correct scoreline of 2 games. As a comparison, Mark Lawrenson, a football pundit, correctly predicted 7 outcomes and 1 scoreline using football ‘knowledge’. There had been ten fixtures in total.
I have since replicated the model and adjusted it so that people can use it to predict any Premier League fixture of their choice. Please be aware that it may take a few minutes for any changes or selections to appear on this page – so please be patient with any ‘refreshing’.
The ten most recent games that people have requested to be predicted are listed below.
The percentages shown are the chances of obtaining that result. Sometimes a team may, for example, be ‘expected’ to win, while the single most likely result could be a 1-1 draw.
A breakdown of how the model works for the first (top) listed game is detailed below.
However, there are a number of limitations.
- David Spiegelhalter details the limitations of much of the maths within his article.
- The model becomes more accurate as the season progresses – as the amount of data on which the model is based increases (it is particularly inaccurate at the start of the season).
- It is only as accurate as the information that has been entered – if I don’t update the results, the accuracy of the model will not update accordingly. See the table, below, to check how up-to-date any results are.
- Changes and selections take a few mintues for data to appear on this page.
- Only ten predictions are displayed at a time, which could be problematic if a few people want to simultaneously view predictions for a few matches.
Results upon which the model is based.