Football Player Market Value Prediction Using Machine Learning

Football player valuation has become an increasingly important component of modern sports management, influencing recruitment, transfer negotiations, squad planning, and financial decision-making. However, player market value is affected by a combination of performance, age, playing position, playing exposure, and competition context, making accurate valuation a challenging predictive problem. This study develops a supervised machine learning framework for predicting football player market value using historical football data obtained from the Transfermarkt dataset repository.The study constructs player-level valuation observations and derives performance statistics from the 365-day period preceding each valuation, including appearances, minutes played, goals, assists, and disciplinary statistics. Player characteristics and competition information are incorporated alongside these temporal performance features. Due to the strong right-skewness of market-value data, the target variable is transformed using log(1 + market value). Three regression approaches—Linear Regression, Random Forest Regression, and XGBoost Regression—are developed and compared using a chronological training-validation-test framework. Observations from 2013–2023 are used for training, 2024–2025 for validation, and 2026 for final testing.XGBoost provides the strongest validation performance, achieving an MAE of 0.7020, RMSE of 0.8953, and R² of 0.6869 in log-market-value space. On the unseen 2026 test set, the final model achieves an MAE of 0.728346, RMSE of 0.927319, and R² of 0.672143. Feature importance and SHAP analysis indicate that recent playing exposure, age, performance statistics, and competition context are important contributors to model predictions. The final model is deployed through an interactive Streamlit application, demonstrating the practical application of the proposed framework. The results indicate that machine learning can provide a useful data-driven approach to football player market-value estimation while also highlighting the limitations of publicly available valuation data.