Probabilistic match importance in professional sports

Quantifying the importance of a match in professional sports can be beneficial in a variety of circumstances, including in the statistical modelling of match outcomes and attendance figures. Current literature on probabilistic match importance measures have overlooked key information, including the significance of a draw outcome in football (soccer), and the multiple end-of-season outcomes that a team can achieve in a season. The aim of this research was to develop a probabilistic measure of match importance that accounts for different end-of-season outcomes and is adaptable to both a two-result (win/loss) and a three-result (win/draw/loss) sport. By first furthering an existing probabilistic measure of match importance, a new model was developed that calculates the importance of achieving in different end-of-season outcomes using a Markov Chain model for both NBA basketball and Bundesliga football. Furthermore, the new model allows for the significance of a draw outcome to be quantified in the latter; which has not been fully achieved in past literature. The new model was compared to a complete Monte Carlo simulation model, where it was observed that the results between the two modelling techniques were similar. The results from the new model were then applied to an Elo ratings system and a stepwise regression model, where an effect on each model’s predictive performance was observed when matches were classified by their level of importance to the competing teams. The completion of this research helps further the knowledge on calculating and applying match importance within sports modelling