Measuring skill via player dynamics in football dribbling

Abstract

Although a myriad of studies have been conducted on player behavior in football, in-depth studies with structured theory are rare due to the difficulty in quantifying individual player skills and team strategies. We propose a physics-based mathematical model that describes football players’ movements during dribbling situations, parameterized by the attacker aggressiveness, the defender hesitance and the top speed of both players. These player- and situation-specific parameters are extracted by fitting the model to real player trajectories from Major League Soccer games, and enable the quantification of player dribbling attributes and decisions beyond classical statistics. We show that the model captures the essential dribbling dynamics, and analyze how differences between parameters in varying game situations provide valuable insights into players’ behavior. Lastly, we quantitatively study how changes in the player’s parameters impact dribbling performance, enabling the model to provide scientific guidance to player training, scouting and game strategy development.