This study was conducted with the aim of discovering distinct player roles in football according to the performance metrics that represented those groups the most. Technical performance data, obtained from Opta, was composed of a total of 1452 outfield players from 5 top-tier Europeans leagues from the 2018/19 season, providing that individuals had played a minimum of 720 minutes during the course of the season. Principal Component Analysis (PCA), a machine learning technique, was utilised in order to identify the main roles and stratify players into their appropriate subgroups. In order to collect several subgroups, PCA was performed on 5 different datasets which represented 5 different position categories: Centre Back (CB), Full Back (FB), Centre Midfielder (CM), Wide Midfielder (WM) and Centre Forward (CF). Upon applying the PCA, a total of 16 different roles (subgroups) were discovered and defined: Stopper (CB), Aerial Dominator (CB), Ball Playing Defender (CB), Defensive Full Back (FB), Nomadic Full Back (FB), Playmaking Full Back (FB), Defensive Midfielder (CM), Box-to-box Midfielder (CM), Advanced Playmaker (CM), Wide Playmaker (WM), Inside Forward (WM), 1v1 Winger (WM), Direct Chance Creator (WM), Target Man (CF), Defensive Forward (CF) and Mobile Forward (CF). Once roles were discovered, representative profiles were built to demonstrate which players contributed the most to each subgroup. Notable international players such as Virgil Van Dijk, Thiago Alcantara and Kylian Mbappe were amongst the top contributing players in the Aerial Dominator, Box-to-box Midfielder and Inside Forward subgroups, respectively.
The analysis performed in this study would be suitable for use in a professional environment, with regard to both coaching and recruitment practices. Coaches could use the profiles to ensure that a player is capable of meeting the requirements of a particular role. Recruitment analysts and scouts could also use this model to identify players who contribute to a role that suits their club’s style of play.