The English Premier League is arguably the most popular national football competition in the world. There are a number of interesting analyses available on this particular competition such as

• the division hierarchy
• home-field advantage
• a long historical overview of match results
• the shift in league domination

Here I attempt to model the match results of the Premier League using Bayesian estimation of a Poisson model. The model I use is taken from the paper by Baio & Biangiardo (2010). For now, I'll focus on one season (2014/15), which is similar to the approach of Baio & Biangiardo (BB), who focus on the Serie A. In contrast, Bååth (2015), focussing on La Liga, uses all available data. An earlier version of this work on the Premier League used his model, but in the interest of computation time, I decided to take another route. Moreover, the average tenure of a Premier League manager is just over 1 year.

I use match data from Football-Data which covers the Premier League from the 1993/94 season onwards. Below some figures that summarise the results. Hopefully in the near future I will also focus on match-prediction, rather than only the aggregates, and improve the model.