SPRING 2022, VOLUME 4

AUTHORS

Amy Liu, University of Pennsylvania (Capstone)
Sardar Cheema, University of Pennsylvania (Capstone)

ABSTRACT

The rise in sports analytics has focused on popular American games such as baseball and basketball, as well as soccer. While cricket is one of the most popular sports on a global level, there has been limited application of sport analytics reported in the literature. In this work, established sports analytics methods are applied to cricket; two cricket models are created, a Pythagorean win rate, and an Elo rating system. The cricket Pythagorean win rate was built from One Day Internationals (ODI) results; the runs scored and the runs conceded on a per-year basis for each team, for data spanning the years 2000 to 2021. A limitation of a Pythagorean win rate is that it does not account for the quality of the competing team, and hence an ELO model was also created. The ELO model was built from International Cricket Council (ICC) full members results in ODIs. The Elo model was initialized with ratings = 1500 at the start of the first year, the model then calculated the ratings for all 27 teams in the dataset. Plotting the change over time made it obvious that we should only look at the full members and remove most of the associate members since there are very few data points for the latter. While to date cricket has not followed the movement towards data analytics seen in other major sports, our work shows there are no inherent limitations in cricket data. Both models created are decently robust, and we hope that they will serve as a foundation for others to build on. We hope this works starts a conversation about analytics in cricket.