Research Papers
Better Prevent than Tackle: Valuing Defense in Soccer Based on Graph Neural Networks
Hyunsung Kim, Sangwoo Seo, Hoyoung Choi, Tom Boomstra, Jinsung Yoon, Chanyoung Park
Evaluating defensive performance in soccer remains challenging, as effective defending is often expressed not through visible on-ball actions such as interceptions and tackles, but through preventing dangerous opportunities before they arise. To address this issue, this paper proposes DEFCON, a framework that quantifies player-level defensive contributions across all attacking situations in soccer. DEFCON estimates the expected value of each attacking option and each defender's responsibility for stopping it using Graph Attention Networks. Then, it assigns positive or negative credits to defenders according to how much they reduce or increase the opponent’s expected value. Experiments on Eredivisie event and tracking data demonstrate strong correlations between DEFCON’s defensive credits and player market valuations, highlighting its potential for real-world scouting and performance analysis.