Complex Networks for the Study of a Soccer Team's Behavior - Final Thesis Project

This work explores the hypothesis of whether complex networks are a suitable tool for defining the behavior of a football team. To this end, we have defined an approach based on the passes made before a goal, considering each goal scored as a score difference to analyze. This process is based on an initial phase of data collection, processing, and filtering, followed by a second phase of statistical and topological analysis. The teams selected for experimentation are two opposing teams during the 2015/2016 season of La Liga, with FC Barcelona as the champion and Rayo Vallecano as the last-placed team. After modeling 340 complex networks corresponding to the score differences of the two teams, we analyze topological metrics such as in-degree and out-degree, clustering coefficient, betweenness centrality, and closeness centrality. For each team, we analyze the aggregated behavior of the metrics in positive, negative, or tied score differences. Finally, we compare topological metrics of players from both teams, and for example, we observe that Javier Mascherano, a central defender for FC Barcelona, has a much more significant role in ball progression compared to José Ruiz, the right-back for Rayo Vallecano. This indicates a tendency for the Catalan team to rely on their defenders to initiate possessions. Additionally, the more defensive profile of Rayo Vallecano is highlighted in comparison to FC Barcelona, due to the positions held by each of their more complete midfielders: Roberto Trashorras, the defensive pivot of the Madrid team, and Andrés Iniesta, the midfielder of the Catalan team. Regarding attack, Javi Guerra, the center forward of Rayo Vallecano, exhibits a more individualistic style of play than Leo Messi, the right winger of FC Barcelona, due to their respective roles: the former seeks to score goals, while the latter generates plays that end in goals. We believe that the application of complex networks to this domain provides a different perspective for analysis by introducing new indicators that could eventually be associated with future key performance indicators.