EDRanIc: spatiotemporal analysis of event distribution consistency in association football

Authors:
Publisher website
Google Scholar
Abstract
Purpose: Unpredictable ball movements and consistent utilization of space are crucial in association football (soccer) for outsmarting opponents and maintaining control throughout a match. Given the sport's low-scoring and unpredictable nature, consistent team performance from start to end is vital, as a single goal can often determine the outcome. In the author's prior work, event distribution randomness (EDRan) was explored for its positive contribution to match-winning performances (Bandara, 2024). This study further extends that investigation by examining the relationship between in-game variations (inconsistency) in event distribution randomness across different phases of a match and a team's chances of success. To support this analysis, the study introduces a temporal performance evaluation metric, EDRanIc (EDRan Inconsistency), designed to capture variations in the randomness of ball movement over time. Design/methodology/approach: This analysis utilized a publicly available event-log dataset containing spatiotemporal data from top-tier club competitions and international tournaments. The proposed approach involved generating “region-based cumulative possession” matrices for defined time periods within a game to obtain probability distributions of event distributions. The spatial event distribution randomness of each team was compared across successive time periods using relative entropy measures, which were then used as temporal features in developing a match-winner prediction model. Findings: A machine learning model trained on the extracted temporal features achieved 79% accuracy in predicting match winners and demonstrated a negative correlation between match-winning performances and inconsistency in EDRan between time periods (EDRanIc). This underscores the significance of maintaining consistency in event distribution randomness throughout a game for achieving match-winning outcomes. The analysis further revealed that winning teams maintained greater EDRan consistency during the early stages of the game. As the game progressed, this consistency declined, reflecting a shift toward more defensive strategies commonly adopted by leading teams in the latter stages of matches. Originality/value: In summary, this study highlights the importance of maintaining consistency in event distribution randomness for match-winning performances and proposes an approach to evaluate this EDRan consistency performance using relative entropy measures. The findings provide valuable insights into the tactical aspects of football, demonstrating how maintaining consistency in event distribution randomness contributes to match-winning performances. Additionally, the proposed team performance evaluation metric (EDRanIc) can assist team management in assessing and identifying inconsistencies in their team's ball movement.

This publication has 29 references indexed in Scilit: