Modeling Matches as Language: A Generative Transformer Approach for Counterfactual Player Valuation in Football

2026-03-16

AI Summary

This work addresses the challenge of evaluating football player transfers, which is inherently difficult due to the strong dependence of individual performance on tactical systems, teammates, and match context. Existing approaches often rely on static metrics and subjective judgments, limiting their ability to simulate unobserved scenarios. To overcome this, we propose ScoutGPT, the first framework to model sequences of in-game events as linguistic sequences. Built upon the NanoGPT architecture, ScoutGPT employs a generative Transformer trained via next-event prediction to learn dynamic match representations. Leveraging Monte Carlo sampling, it generates plausible event sequences under counterfactual team compositions to assess player value. Experiments on K League data demonstrate that ScoutGPT effectively simulates the impact of transferred players on offensive progression and goal-scoring probability, significantly outperforming baseline models and accurately capturing individual player contributions.

Technology Category

• Search and Optimization: Sampling/Simulation-based Search
• Natural Language Processing: Generation
• Game Theory and Economic Paradigms: Adversarial Learning

Application Category

• Economics, Online Markets and Human Computation: Trust and reliance of crowd workers and data experts on GenAI
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Evaluating football player transfers is challenging because player actions depend strongly on tactical systems, teammates, and match context. Despite this complexity, recruitment decisions often rely on static statistics and subjective expert judgment, which do not fully account for these contextual factors. This limitation stems largely from the absence of counterfactual simulation mechanisms capable of predicting outcomes in hypothetical scenarios. To address these challenges, we propose ScoutGPT, a generative model that treats football match events as sequential tokens within a language modeling framework. Utilizing a NanoGPT-based Transformer architecture trained on next-token prediction, ScoutGPT learns the dynamics of match event sequences to simulate event sequences under hypothetical lineups, demonstrating superior predictive performance compared to existing baseline models. Leveraging this capability, the model employs Monte Carlo sampling to enable counterfactual simulation, allowing for the assessment of unobserved scenarios. Experiments on K League data show that simulated player transfers lead to measurable changes in offensive progression and goal probabilities, indicating that ScoutGPT captures player-specific impact beyond traditional static metrics.

Problem

• counterfactual simulation
• player valuation
• football transfers
• contextual factors
• hypothetical scenarios

Innovation

• counterfactual simulation
• generative Transformer
• football analytics
• match event modeling
• player valuation