TEXT

Abstract

Despite the rapid growth of analytics in most major sports, advances in soccer have been more limited. In particular, while certain metrics such as expected goals and key passes have gained popularity, the use of network analysis in soccer remains mostly unexplored. This thesis aims to investigate the use of network analysis, with a particular focus on possession, to quantify and evaluate player performance in soccer. We find that various properties of possession, such as total passes and density, are significantly correlated with team performance. Furthermore, we find that network measures such as closeness centrality offer tremendous promise in identifying key players for a team. Overall, we determine that network analysis, applied to possessions, offers the potential for novel insights into understanding both the performance and play styles of soccer players and their positions.