Modelling Football Scorelines: Where Poisson Falls Short, and How Dixon-Coles Fixes It…by Gecko Edge
If you have ever tried to model football results, you have almost certainly started with the Poisson distribution. It is commonly cited, and a natural first step — and for good reason. The Poisson process works best when modelling discrete events occurring on a consistent basis (e.g. goals). But anyone who has used it seriously…
