Evaluation of off-the-ball actions in soccer
Keywords:big data, machine learning, model validation, player tracking data
Whereas there is no shortage of statistics that have been proposed and reported for invasion sports, almost all of the widely reported statistics are based on actions involving the ball. Yet, in football (soccer), it is well-known that players typically possess the ball for less than three minutes during a 90-minute match. In this paper, we develop automatic methods that analyze the activities of players that are ``off-the-ball'' in soccer. Specifically, a metric is introduced which measures defensive anticipation in soccer. The approach is conceptually simple: Using roughly four million spatio-temporal instances, we use machine learning techniques to predict the velocity (two-dimensional directional vector and speed) of a defensive player in a given situation. A metric is then developed which compares the player's actual velocity with the predicted velocity of a typical player in this situation. The interpretation of the defensive anticipation metric is based on the tenet that fast is better than slow. The analysis is facilitated through the availability of player tracking data which records the position of players at frequent and regular intervals throughout matches. The metric is calculated for players based on a season of soccer data, where validity and reliability are demonstrated. The metric also conforms to common sense where it is expected and observed that there is a reduction in defensive anticipation as players tire. The proposed approach is applicable and can be tailored to all invasion sports where player tracking data are available.
How to Cite
Copyright (c) 2023 Statistica Applicata - Italian Journal of Applied Statistics
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.