Passing networks in football: Selected Manchester United matches during the 2015-2016 season

The interactions between players during a football match can be recorded as a matrix of passes and visualised as a network diagram. The diagram provides semi-quantitative insight into the operation of the team as a whole, the functioning of units of the team (defence, midfield, attack) and the importance of individual players to team effort. In addition, in-depth quantitative analysis of patterns of team passing using network mathematics techniques has given insight into team dynamics (Clemente, Martins and Mendes, 2016) and some network parameters have been shown to be correlated with team performance in football (Clemente, Martins, Kalamaras, Wong and Mendes, 2015; Grund, 2012; Peña & Touchette, 2012). In the present study, passing networks were analysed using both informal and formal mathematical methods in an attempt to relate network parameters to the performance of an elite level team. Six national and international matches of Manchester United Football Club during the 2015-2016 season were annotated using Dartfish 10 software. The matches covered a range of team success in terms of match outcome. Account was taken of completed passes by players of the focus team, the pitch location of those passes, the context of passing in relation to opposition play and an enhanced set of match statistics. Data to be presented will relate informal and standard network measures (degree, centrality and modularity included) to match outcomes, unit performances and the influence of individual players, as supported in part by non-parametric statistics. The sensitivity of network parameters to errors in the recording of passes will be reported.This study highlights the potential and potential constraints of using network parameters in the analysis of performance of a football team.