Modelling Test Matches Outcome Using Bradley Terry Model

Abstract

In cricket, the ranking of players or teams has a significant impact on determining the top player or team. The goal of this study is to forecast Test Cricket Match Outcomes and then create a team rating based on these predictions. In general, in any sport, the team with the most rating points is regarded as the greatest, and the teams are ranked based on their ratings. The Bradley-Terry model is used to forecast match outcomes using data from January 3rd, 2010 through August 25th, 2020, for a total of 444 matches. The teams of the top ten cricket-playing nations have been ranked using model rating.  The rating of the underlying model is also compared to the updated 25th August 2020 official ICC ranking, which yields roughly similar findings. To improve model accuracy, the home ground effect is added, which has a substantial impact on team performance. After the home factor is added as a potential covariate, the model ranking is compared to the ICC ranking, which yields more closed results. Later, when these two results are compared with or without the home ground component, there is a significant improvement between them. To assess team performance, winning probabilities and confidence intervals are produced for each participating nation using model estimate coefficients. Three performance indicators are computed for the top ten playing teams, taking away and home ground factors into account. When these real data were compared, the home ground factor produced more precise results than removing the specified covariate. The outcomes of the suggested approach are also compared to predetermined odds that differ marginally.