Optimal football team building : a data-driven approach to build the best team

Details

Supervisors

Delvenne, Jean-Charles ; Taminiaux, Martin

Faculty

Ecole polytechnique de Louvain

Degree label

Master [120] : ingénieur civil en mathématiques appliquées, à finalité spécialisée

Abstract

In this study, we introduce a graph-based representation of a football team, where players and their interactions are modeled as nodes and edges, respectively. This representation allows us to quantify both individual player contributions and the chemistry between players. Then the goal is to find the arrangement of players, which player at which position, that maximizes the overall team performance which is the sum of the individual player scores and the edge scores. By incorporating edge scores to represent player interactions, we enhance the accuracy and relevance of the team selection process. We introduce also a validation method to assess the relative importance of the chemistry between players compared to the individual qualities of the players. Finally, we demonstrate the potential applications of our model, including player recruitment, in-game management, and match outcome prediction. We also discuss the limitations and possible improvements of the model.