TELEMETRY
CHAOS
Machine learning meets 75 years of Formula 1. Race predictions, telemetry analysis, and cross-era simulations.
Try the Race PredictorRace Winner Predictor
Optuna-tuned XGBoost model predicting race winners from grid position, form, constructor strength, weather, and track history. Auto-updates weekly during the season.
Virtual Race Simulator
Monte Carlo engine that pits legends against each other across eras. Senna vs. Verstappen? Fangio vs. Hamilton? Era-normalized profiles, lap-by-lap simulation with tyre strategy, weather, and safety cars.
Data Foundation
Built on the most comprehensive open F1 dataset available
How It Works
Ingest
Pull race results, qualifying, lap times, pit stops, and per-lap telemetry from FastF1, Jolpica-F1, and open datasets. 1950–present.
Engineer
Build 21 features per driver per race: rolling form, practice pace, tyre degradation, pit stop speed, safety car probability, weather.
Train
XGBoost + LightGBM ensemble tuned with Optuna (200 trials, GPU). Time-series split to prevent leakage. Evaluated on 2024–2025.
Predict
Win probabilities for every driver in every race. Auto-retrains weekly during the season with fresh data.