Unveiling Hidden Pivotal Players with GoalNet: A GNN-Based Soccer Player Evaluation System
Jacky Hao Jiang, Jerry Cai, Anastasios Kyrillidis
TL;DR
The paper tackles the bias of traditional soccer analytics toward attackers by introducing GoalNet, a GNN-based framework that assigns credit for changes in $xT$ through event-centric graphs that fuse spatial and temporal features. It develops three variants—Basic GoalNet, GATGoalNet, and TransGoalNet—to model local and global interactions and to distribute $\Delta xT$ to involved players via learned embeddings. By integrating both node representations and edge interactions, along with temporal context, the approach highlights pivotal non-scoring contributions from midfielders and defenders and demonstrates robustness on real match data. The results show that graph-based models better capture build-up, defensive organization, and transitions, with Graph Transformers offering advantages for long-range dependencies, underscoring the method's value for comprehensive, actionable soccer analytics.
Abstract
Soccer analysis tools emphasize metrics such as expected goals, leading to an overrepresentation of attacking players' contributions and overlooking players who facilitate ball control and link attacks. Examples include Rodri from Manchester City and Palhinha who just transferred to Bayern Munich. To address this bias, we aim to identify players with pivotal roles in a soccer team, incorporating both spatial and temporal features. In this work, we introduce a GNN-based framework that assigns individual credit for changes in expected threat (xT), thus capturing overlooked yet vital contributions in soccer. Our pipeline encodes both spatial and temporal features in event-centric graphs, enabling fair attribution of non-scoring actions such as defensive or transitional plays. We incorporate centrality measures into the learned player embeddings, ensuring that ball-retaining defenders and defensive midfielders receive due recognition for their overall impact. Furthermore, we explore diverse GNN variants-including Graph Attention Networks and Transformer-based models-to handle long-range dependencies and evolving match contexts, discussing their relative performance and computational complexity. Experiments on real match data confirm the robustness of our approach in highlighting pivotal roles that traditional attacking metrics typically miss, underscoring the model's utility for more comprehensive soccer analytics.