Residual Hybrid Learning for Data-Limited Formula 1 Lap Time Prediction Across Multiple Circuits
Resumo
Lap time prediction in Formula 1 is a key element of race strategy analysis, as small changes in race pace can affect pit stop timing, tire compound selection, and the viability of tactics such as the undercut or overcut. This work uses publicly available telemetry data extracted with the FastF1 library and examines five Formula 1 circuits from the 2022 to 2025 seasons that fall under a single technical regulation. The methodology uses residual learning in a two-stage deep learning forecasting approach to address data limitations and improve the accuracy of drivers’ lap time predictions across multiple seasons. First, a Multiple Linear Regression (MLR) model predicts the aggregated lap time series for all drivers on each circuit. Then, a Long Short-Term Memory (LSTM) network predicts the residuals, filtered by driver and year. In addition, we propose a score that balances performance and model variation, helping identify the most suitable model. We call our approach Hybrid MLR-LSTM. It achieves the best training performance across all circuits. In testing, the hybrid model performs best on Italy and achieves performance comparable to competing models on the other circuits. To support reproducibility, the code repository is publicly available at https://github.com/emipe09/HybridLSTM-F1.
Referências
Bansal, A., Arora, A., Bhati, L., Sethia, K., Verma, I., and Kapoor, N. Advanced machine learning approaches for formula 1 race performance prediction: A comprehensive analysis of championship point forecasting, 2025. Preprint available at ResearchGate.
Bekker, J. and Lotz, W. Planning formula one race strategies using discrete-event simulation. Journal of the Operational Research Society 60 (7): 952–961, 2009.
El Haber, E., Sawaya, E., Attieh, M., Tannous, A., Ghazaly, W., and Owayjan, M. Formula 1 Race Winner Prediction Using Random Forest and SHAP Analysis. In IC2AI. IEEE, Beirut, France, pp. 1270–1274, 2025.
Fatima, S. S. W. and Johrendt, J. Deep-racing: An embedded deep neural network (ednn) model to predict the winning strategy in formula one racing. International Journal of Machine Learning 13 (3), 2023.
Fédération Internationale de l’Automobile. F1’s new era. [link], 2025. Accessed: 2026-05-03.
Géron, A. Hands-On Machine Learning with Scikit-Learn, Keras, and TensorFlow: Concepts, Tools, and Techniques to Build Intelligent Systems. O’Reilly Media, Sebastopol, CA, 2019.
Heilmeier, A., Graf, M., Betz, J., and Lienkamp, M. Application of monte carlo methods to consider probabilistic effects in a race simulation for circuit motorsport. Applied Sciences 10 (12): 4229, 2020.
Heilmeier, A., Graf, M., and Lienkamp, M. A race simulation for strategy decisions in circuit motorsports. In Proc. of the IEEE 21st ITSC. IEEE, pp. 2986–2993, 2018.
Hyndman, R. J. and Athanasopoulos, G. Forecasting: Principles and Practice. OTexts, Melb., Australia, 2021.
James Fong, S., Li, G., Dey, N., González Crespo, R., and Herrera Viedma, E. Finding an accurate early forecasting model from small dataset: A case of 2019-ncov novel coronavirus outbreak. International Journal of Interactive Multimedia and Artificial Intelligence 6 (1): 132–140, Mar., 2020.
Kara, M. F. and Küçükmanisa, A. Predicting lap times in formula 1: A deep learning approach using lstm networks. In 2025 IEEE 21st ICCP. pp. 1–5, 2025.
Kiser, A. C., Shi, J., and Bucher, B. T. An explainable long short-term memory network for surgical site infection identification. Surgery vol. 176, pp. 24–31, 2024.
Krzysztoń, S. A. and Smołka, J. Application of machine learning for predicting formula 1 race results. Journal of Computer Sciences Institute, 2026.
Mahmud, A., Noor, S., Musa, K., Hamzah, F., Yudin, Z., Kamaruddin, N., Madawana, A., and Awang Nawi, M. Hybrid arima-lstm for covid-19 forecasting: a comparative ai modeling study. PeerJ Comp. Sci. vol. 11, pp. e3195, Sep., 2025.
Michelucci, U. and Venturini, F. Best practices for machine learning experimentation in scientific applications, 2025.
Peixeiro, M. Time Series Forecasting in Python. Manning Publications, 2022.
Quiroga García, M. P. Optimization of pit stop strategies in formula 1 racing: A data-driven approach, 2024.
Sasikumar, A., Leema, A. A., and Balakrishnan, P. Data-driven pit stop decision support for formula 1 using deep learning models. Frontiers in Artificial Intelligence vol. Volume 8, 2025.
Stepien, Z. The pro-ecological evolution of powertrains and fuels in formula 1. Energies 18 (22): 6013, 2025.
Thomas, D., Jiang, J., Kori, A., Russo, A., Winkler, S., Sale, S., McMillan, J., Belardinelli, F., and Rago, A. Explainable reinforcement learning for formula one race strategy. arXiv preprint arXiv:2501.04068 , 2025.
Todd, J., Jiang, J., Russo, A., Winkler, S., Sale, S., McMillan, J., and Rago, A. Explainable time series prediction of tyre energy in formula one race strategy. In Proc. 40th SIGAPP. SAC ’25. ACM, pp. 1082–1089, 2025.
Vatsa, A., Hati, A. S., Kumar, P., Margala, M., and Chakrabarti, P. Residual lstm-based short duration forecasting of polarization current for effective assessment of transformers insulation. Sci. Rep. 14 (1369), 2024.
Zhao, Z. Deep neural network-based lap time forecasting of formula 1 racing. Applied and Computational Engineering vol. 47, pp. 61–66, 03, 2024.
