Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

ScoutGPT: Capturing Player Impact from Team Action Sequences Using GPT-Based Framework

Miru Hong, Minho Lee, Geonhee Jo, Jae-Hee So, Pascal Bauer, Sang-Ki Ko

TL;DR

The paper tackles the challenge of predicting transfer success by modeling football match play as a GPT-style sequence conditioned on individual players. It introduces ScoutGPT, a decoder-only transformer that jointly predicts next-event attributes and residual on-ball value (rOBV), with counterfactual substitution to simulate how a player would perform in a different tactical environment. Empirical results show improved next-event prediction and spatial accuracy over adapted baselines, and the learned embeddings reveal interpretable, non-position-specific role structure. The transfer-analysis capabilities, demonstrated via hypothetical substitutions and case studies, provide a principled framework for assessing fit and guiding recruitment decisions in context-specific settings.

Abstract

Transfers play a pivotal role in shaping a football club's success, yet forecasting whether a transfer will succeed remains difficult due to the strong context-dependence of on-field performance. Existing evaluation practices often rely on static summary statistics or post-hoc value models, which fail to capture how a player's contribution adapts to a new tactical environment or different teammates. To address this gap, we introduce EventGPT, a player-conditioned, value-aware next-event prediction model built on a GPT-style autoregressive transformer. Our model treats match play as a sequence of discrete tokens, jointly learning to predict the next on-ball action's type, location, timing, and its estimated residual On-Ball Value (rOBV) based on the preceding context and player identity. A key contribution of this framework is the ability to perform counterfactual simulations. By substituting learned player embeddings into new event sequences, we can simulate how a player's behavioral distribution and value profile would change when placed in a different team or tactical structure. Evaluated on five seasons of Premier League event data, EventGPT outperforms existing sequence-based baselines in next-event prediction accuracy and spatial precision. Furthermore, we demonstrate the model's practical utility for transfer analysis through case studies-such as comparing striker performance across different systems and identifying stylistic replacements for specific roles-showing that our approach provides a principled method for evaluating transfer fit.

ScoutGPT: Capturing Player Impact from Team Action Sequences Using GPT-Based Framework

TL;DR

The paper tackles the challenge of predicting transfer success by modeling football match play as a GPT-style sequence conditioned on individual players. It introduces ScoutGPT, a decoder-only transformer that jointly predicts next-event attributes and residual on-ball value (rOBV), with counterfactual substitution to simulate how a player would perform in a different tactical environment. Empirical results show improved next-event prediction and spatial accuracy over adapted baselines, and the learned embeddings reveal interpretable, non-position-specific role structure. The transfer-analysis capabilities, demonstrated via hypothetical substitutions and case studies, provide a principled framework for assessing fit and guiding recruitment decisions in context-specific settings.

Abstract

Transfers play a pivotal role in shaping a football club's success, yet forecasting whether a transfer will succeed remains difficult due to the strong context-dependence of on-field performance. Existing evaluation practices often rely on static summary statistics or post-hoc value models, which fail to capture how a player's contribution adapts to a new tactical environment or different teammates. To address this gap, we introduce EventGPT, a player-conditioned, value-aware next-event prediction model built on a GPT-style autoregressive transformer. Our model treats match play as a sequence of discrete tokens, jointly learning to predict the next on-ball action's type, location, timing, and its estimated residual On-Ball Value (rOBV) based on the preceding context and player identity. A key contribution of this framework is the ability to perform counterfactual simulations. By substituting learned player embeddings into new event sequences, we can simulate how a player's behavioral distribution and value profile would change when placed in a different team or tactical structure. Evaluated on five seasons of Premier League event data, EventGPT outperforms existing sequence-based baselines in next-event prediction accuracy and spatial precision. Furthermore, we demonstrate the model's practical utility for transfer analysis through case studies-such as comparing striker performance across different systems and identifying stylistic replacements for specific roles-showing that our approach provides a principled method for evaluating transfer fit.

Paper Structure

This paper contains 15 sections, 5 equations, 2 figures, 7 tables.

Table of Contents

  1. Introduction
  2. Methodology
  3. Event Representation and Tokenization
  4. Autoregressive Next-Event Prediction
  5. Model Architecture
  6. Hypothetical Player Substitution for Transfer Simulation
  7. Experiments
  8. Dataset
  9. Baselines
  10. Evaluation Metrics
  11. Main Results
  12. Applications
  13. Player Embedding for Similar Player Retrieval
  14. Hypothetical Transfer Simulation
  15. Conclusions

Figures (2)

  • Figure 1: Overview of the ScoutGPT framework. Our nanoGPT-based Transformer model autoregressively predicts event tokens, enabling counterfactual 'what-if' simulations. For instance, replacing Darwin Núñez with Alexander Isak could alter actions (e.g., pass/shot) or modify the same action with a different location, outcome, or on-ball value (OBV). A video example of this sequence can be seen at the 23-second mark in the https://www.youtube.com/watch?v=y9wH5s0C33w.
  • Figure 2: Player Embeddings by Position Category. This visualization demonstrates how functional tendencies are encoded through sequential event patterns and learned player-conditioned residual on-ball value (rOBVt) dynamics, even without using positional labels during training.