Toward individualized learning pathway selection in martial arts tricking using predictive models

Introduction Meeting relevant prerequisites may improve movement readiness and skill development. In Martial Arts Tricking (MAT), diverse movement categories complicate individualized learning pathway selection. This proof-of-concept study investigates whether supervised machine learning can support personalized training by predicting trick performance and identifying influential athlete-specific features. Methods Synthetic data representing plausible feature relationships were used to demonstrate the framework, not to infer real-world athlete performance. Multiple models were evaluated for predicting successful trick landings and identifying influential features, including body weight, age, experience and others. Results Logistic regression was selected for its interpretability, achieving 81.8% mean accuracy in nested cross-validation and 84.3% accuracy in a separate stratified hold-out evaluation. Predicted probabilities illustrate how success likelihoods across trick categories could inform individualized learning pathway selection by highlighting athlete-specific strengths and weaknesses. Discussion The study demonstrates the feasibility of an interpretable machine-learning framework rather than a validated coaching system. Real-world athlete data are required to establish practical effectiveness. The approach may also transfer to other sports.