A1119
Title: Refining soccer rankings: Balancing scored points, goals for and against with kendall correlations and radar charts
Authors: Valerio Ficcadenti - London South Bank University (United Kingdom) [presenting]
Raffaele Mattera - Sapienza University of Rome (Italy)
Roy Cerqueti - Sapienza University of Rome (Italy)
Abstract: Computational challenges of ranking when more than a feature is used in the ranking exercise are addressed. We introduce a novel ranking method showcasing an application in soccer championships, utilizing computational statistics derived from a combination of Kendall correlations and radar charts. The proposed method considers goals scored and conceded by teams in individual matches as additional scoring factors beyond the traditional win-draw-lose system in soccer. This approach aims to reduce biases in existing scoring rules, such as the mathematical certainty of some teams winning before the end of the competitions. The methodology involves calculating the areas of radar charts, which represent the normalized Kendall correlation values between different ranking metrics. By transforming these areas into target Kendall tau values, we identify alternative team rankings that align with the target correlations. This process provides a different view of team performance and reduces the impact of biases present in the current ranking system. We apply our methodology to the Italian Serie A championships, showcasing its capability to generate different rankings through a robust computational framework. This innovative approach demonstrates the potential for improving ranking systems in sports competitions by incorporating detailed performance metrics.
