Betting markets for English Premier League results and scorelines: evaluating a forecasting model

Citation

Reade J, Singleton C & Vaughan Williams L (2020) Betting markets for English Premier League results and scorelines: evaluating a forecasting model. Economic Issues, 25 (1), pp. 87-106. https://www.economicissues.org.uk/Vol25.html#a4

Abstract
Using betting odds from two recent seasons of English Premier League football matches, we valuate probability and point forecasts generated from a standard statistical model of goal scoring. The bookmaker odds show significant evidence of the favourite-longshot bias for exact scorelines, which is not generally present for match results. We find evidence that the scoreline probability forecasts from the model are better than what the odds of bookmakers imply, based on forecast encompassingregressions. However, when we apply a simple betting strategy using point forecasts from the model, there are no substantial or consistent financial returns to be made over the two seasons. In other words, there is no evidence from this particular statistical model that the result, scoreline, margin of victory or total goals betting markets are on average inefficient.

Keywords
Forecasting; Statistical modelling; Regression models; Prediction models

Journal
Economic Issues: Volume 25, Issue 1