Probabilistic Simulator for Race Event Forecasting and Based on Telemetry and Historical Data

In this thesis, a telemetry-informed race simulator for Formula One is developed to investigate the impact of strategic decisions and performance variability across full race distances. Lap time evolution is modelled as the sum of deterministic and stochastic components, calibrated using historical data and telemetry-derived parameters for fuel consumption, tire wear, driver lap time consistency and pit stop losses. Monte Carlo methods are employed to simulate multiple race iterations, reflecting the randomness inherent in motorsport events such as Full Course Yellow phase, mechanical failures and overtaking maneuvers. This probabilistic modelling enables the study of not just expected results but also the confidence intervals and failure risks associated with different strategies. Through this approach, the simulator provides a framework for evaluating the robustness of race strategies and the performance envelope of each driver under varying conditions. The system lays the groundwork for data-driven predictive tools capable of supporting engineering and strategic decision-making in high variability racing environments.