This study aimed to evaluate volleyball players' performance using machine and deep learning methods. It sought to predict performance, identify influential factors, and examine gender and position-based differences to inform management decisions, talent identification, and sports marketing.

This quantitative, applied, and descriptive-analytical research studied male and female players in the 2023-2024 Volleyball Nations League (VNL) and the 2020 & 2024 Olympics. Stratified sampling was used. Performance data was extracted from FIVB's VIS system. Data was split 70/30 for training/testing. ANN, CNN, Random Forest, and Gaussian Process Regression models in Mathematica were used for prediction, evaluated using R, RMSE, Taylor diagrams, PDF, and spectral scatter plots. Panel analysis in EViews with tests like LLC, F-Limer, LM, and Hausman examined explanatory variables.

Results showed Boskovic (men's Olympics) and T. Stern (women's VNL) and Leon (women's Olympics) were top performers. The CNN model was most accurate (high R, RMSE=0.713, 10s runtime). Panel analysis indicated error (β=0.74) and prior success (β=0.49) most impacted men, while technical accuracy most impacted women (R²=0.89). Gender (men: 34.59; women: 34.24) and position differences (setter passing accuracy, attacker spike efficiency) were confirmed. Combining methods provides a comprehensive view, with CNN superiority emphasized. Findings aid personalized training, error reduction, and team strategy.