Predicting FIFA Ultimate Team Player Market Prices: A Regression-Based Analysis Using XGBoost Algorithms from FIFA 16-21 Dataset

This study investigates the use of XGBoost, a machine learning algorithm, for predicting player prices in FIFA Ultimate Team (FUT) from FIFA 16 to FIFA 21. Virtual economies in gaming, particularly in FUT, have grown substantially, with in-game asset prices influenced by a variety of factors such as player attributes, performance metrics, and market dynamics. The objective of this research is to enhance the accuracy of price predictions in FUT through advanced machine learning techniques. The dataset comprises historical player data, including attributes such as rating, skills, and in-game statistics. XGBoost was employed due to its ability to handle large, complex datasets and capture non-linear relationships effectively. The model achieved an R-squared value of 0.8911, indicating that it explains 89% of the variance in player prices, while the RMSE value of 30368.85 reveals the model's precision in estimating prices. Feature importance analysis showed that attributes such as WorkRate and Rating significantly influenced price predictions. Compared to traditional methods like linear regression, XGBoost provided superior accuracy and computational efficiency, making it a valuable tool for understanding player price dynamics in virtual gaming markets. The findings suggest that accurate price predictions can improve gaming strategies for players and provide valuable insights for game developers in optimizing virtual economies. This research also highlights the potential for further exploration using advanced machine learning algorithms to predict price fluctuations in gaming environments.