Scholarlink

This study presents a novel predictive framework for individual rider outcomes in the Polish speedway PGE Ekstraliga. Using 2018–2024 data, we introduce the Expected Points (xP) metric and apply feature engineering, including Elo ratings, gate- and track-specific indicators, and rivalry stats. An XGBoost model, optimized with hyperparameter tuning, achieves over 50% accuracy across four classes. Feature importance analysis shows the critical role of long-term rider attributes, dynamic ratings, and contextual variables such as riders’ gate preferences. The validated xP sets a benchmark for speedway analytics and enables practical applications in tactical planning, rider evaluation, and talent identification. Future work should explore the integration of telemetry and cross-league data for improved predictions.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Racing Beyond the Gate: Predicting Speedway Results with Expected Points (xP)

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Piotr Kaczorek,
  • Jędrzej Słupski,
  • Franciszek Kornobis,
  • Michał Zaręba,
  • Tomasz Górecki,
  • Tomasz Piłka

摘要

This study presents a novel predictive framework for individual rider outcomes in the Polish speedway PGE Ekstraliga. Using 2018–2024 data, we introduce the Expected Points (xP) metric and apply feature engineering, including Elo ratings, gate- and track-specific indicators, and rivalry stats. An XGBoost model, optimized with hyperparameter tuning, achieves over 50% accuracy across four classes. Feature importance analysis shows the critical role of long-term rider attributes, dynamic ratings, and contextual variables such as riders’ gate preferences. The validated xP sets a benchmark for speedway analytics and enables practical applications in tactical planning, rider evaluation, and talent identification. Future work should explore the integration of telemetry and cross-league data for improved predictions.