Prediction System for National Football Team Match Results

• Gustavo C. Bento Laurindo UFSCar
• Ivan Fernando R. Villalba UFSCar
• Murilo C. Naldi UFSCar

DOI: https://doi.org/10.5753/eniac.2026.30784

Resumo

This work presents a supervised learning system for predicting the outcomes of football matches between national teams, classifying the results as home win, draw, or away win. Three classifiers were evaluated: XGBoost, Logistic Regression, and an MLP Neural Network. Records from 1872 to 2026 were used to compute Elo ratings and performance indicators, with learning applied to matches from 2015 onward, enriched with overall FIFA video game player ratings aggregated by national team. Twelve configurations per model were tested using Rolling and Expanding Window validation. The results indicate that Elo-based features are the most informative predictors. XGBoost achieved the best accuracy and probabilistic calibration, whereas the balanced MLP achieved the highest macro F1-score and the best performance in drawing identification.

Referências

Baboota, R. and Kaur, H. (2019). Predictive analysis and modelling football results using machine learning approach for english premier league. International Journal of Forecasting, 35(2):741–755.

Capital, J. B. (2024). Fifa players – fifa 15 to fifa 24. Kaggle dataset. Accessed: 2026.

Dahatonde, A. (2023). Fifa world cup all dataset. Kaggle dataset. Accessed: 2026.

Gao, Z. and Kowalczyk, A. (2021). Random forest model identifies serve strength as a key predictor of tennis match outcome. Journal of Sports Analytics, 7(4):255–262.

Hvattum, L. M. and Arntzen, H. (2010). Using elo ratings for match result prediction in association football. International Journal of Forecasting, 26(3):460–470. Sports Forecasting.

Jürisoo, M. (2023). International football results from 1872 to 2017. Kaggle dataset. Accessed: 2026.

Laurindo, G. C. B., Villalba, I. F. R., and Naldi, M. C. (2026). Prediction System for National Football Team Match Results: Jupyter notebook and final dataset. [link]. Accessed: 2026.

Nyagami (2024). Ea sports fc 25 database: Ratings and stats. Available on Kaggle. Accessed: 2026.

Papageorgiou, G., Sarlis, V., and Tjortjis, C. (2024). Evaluating the effectiveness of machine learning models for performance forecasting in basketball: a comparative study. Knowledge and Information Systems, 66(7):4333–4375.

Przybyła-Kasperek, M. and Wesecki, M. (2025). A comparative study of ensemble methods and feature selection techniques for predicting english premier league match outcome. Procedia Computer Science, 270:4822–4829. 29th International Conference on Knowledge-Based and Intelligent Information Engineering Systems (KES 2025).

Rodrigues, F. and Ângelo Pinto (2022). Prediction of football match results with machine learning. Procedia Computer Science, 204:463–470. International Conference on Industry Sciences and Computer Science Innovation.

Rovnez (2025). Fc 26 / fifa 26 player data. Available on Kaggle. Accessed: 2026.

Warkiani, M. E. and Moattar, M. H. (2025). A comprehensive survey on recent feature selection methods for mixed data: Challenges, solutions and future directions. Neurocomputing, 623:129372.