Dynamic bayesian forecasting models of football match outcomes

Dynamic Generalized Linear Models (DGLMs) are essentially generalised linear models with parameters that are stochastic. They are Bayesian in flavour and are particularly suited to forecasting applications. This paper outlines a practical implementation of a Poisson DGLM model that can easily be deployed using the freely available software WinBUGS. Using match results data from the Scottish Premier League (SPL) between 2003/2004 to 2005/2006, the DGLM approach is shown to provide more improved predictive probabilities of future match outcomes, compared to the non-dynamic form of the model.