Transformer-Based Framework for Versatile Analysis of Events Data in Soccer

Aviv Rovshitz, Rami Puzis

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

In sports analytics, machine learning models are used for players’ appointment analysis, strategic planning, etc. Multiple rating systems were developed to rank the player performance based on statistical measurements, machine learning, and deep learning analysis of actions and sequences. However, capturing the complex relationships between actions within open play sequences, affected by tactical game planning and team deployment, is a major challenge. Inspired by language models, we developed SoccerTransformer, a transformer-based deep learning framework tailored to soccer analytics to address this challenge. The proposed framework includes (1) self-supervised pre-training to capture generic in-game dynamics, (2) a downstream sequence classification task to predict the attack phase outcomes, and (3) a player rating system. Empirical results demonstrate that SoccerTransformer accurately captures player roles and predicts goals with F1 of 0.814 to 0.862 on previously unseen games. SoccerTranformer framework stands out by providing a greater correlation with market values than other state-of-the-art rating systems.

Original languageAmerican English
Title of host publicationMachine Learning and Data Mining for Sports Analytics - 11th International Workshop, MLSA 2024, Revised Selected Papers
EditorsUlf Brefeld, Jesse Davis, Jan Van Haaren, Albrecht Zimmermann
PublisherSpringer Science and Business Media Deutschland GmbH
Pages80-92
Number of pages13
ISBN (Print)9783031866913
DOIs
StatePublished - 1 Jan 2025
Event11th International Workshop on Machine Learning and Data Mining for Sports Analytics, MLSA 2024 - Vilnius, Lithuania
Duration: 9 Sep 20249 Sep 2024

Publication series

NameCommunications in Computer and Information Science
Volume2460 CCIS

Conference

Conference11th International Workshop on Machine Learning and Data Mining for Sports Analytics, MLSA 2024
Country/TerritoryLithuania
CityVilnius
Period9/09/249/09/24

Keywords

  • language model
  • market value
  • soccer
  • sports analytics
  • transformers

All Science Journal Classification (ASJC) codes

  • General Computer Science
  • General Mathematics

Fingerprint

Dive into the research topics of 'Transformer-Based Framework for Versatile Analysis of Events Data in Soccer'. Together they form a unique fingerprint.

Cite this