Prediction of soccer clubs' league rankings by machine learning methods: The case of Turkish Super League.

This article focuses on predicting the final league rankings of the Turkish Super League (TSL) using machine learning (ML) models based on 23 technical and physical parameters collected via the SENTIO Sports optical tracking system over three seasons (2015–2018). Three ML methods—Artificial Neural Networks (ANN), Radial Basis Function Networks (RBFN), and Multiple Linear Regression (MLR)—were applied, with ANN achieving the highest predictive accuracy (R² = 0.60) compared to MLR (R² = 0.46) and RBFN (R² = 0.26). Key variables influencing rankings included average goals per game, goals scored from throw-ins, and key passes per match. The study suggests that integrating physical and technical data enhances league ranking predictions and can assist coaches and trainers in optimizing team performance and training strategies.