Applying Analytic Network to Extract the Strategy in the Game of Football

In this thesis, we use complex system models, in particular complex network theory to identify strategies and the key players on soccer games using the Matrix of Passes Distribution (MDP) we recreate the network of the match where the nodes are the players and the weighted link are the number of passes naked between the two players during a match, analyzing the characteristics and metrics of the network to obtain the strategy. Applying the methodology to create the network and analyzing designed by Lopez Pena, et al.[1]. We recreate the MDP using Python, indicating the nodes names as the player names, and allocating the links with his respective number of passes. Once the network is generated, using the complex network theory we proceed to calculate the centrality metrics of the network using degree, closeness, betweenness analyzing and interpreting the metrics on the context of football [3]. We make a complementary analysis, creating the network visualization of every team using Gephi software to detect particular strategies and key players. Performing a more robust analysis, we complement the analysis generating an average network to obtain the general characteristics of the teams. To do that, we join the network of each match on a single one, keeping the most participative team members, then calculate the centrality metrics and analyze. More than 20 Illustrative examples for the teams of Mexico, Chile, Argentina and Germany during the 2014 World Cup in Brazil are analyzed. We replicate the methodology and complement analysis, getting similar results between the match and the expert analysis.