All projects
Active
Featured
Football-Predictor-v5.0
A Python football prediction pipeline using XGBoost, Poisson models, ensemble probabilities, and backtesting.
- Role
- Data pipeline and model developer
- Timeline
- Since May 2026
Python
XGBoost
Poisson
Pandas
Backtesting
Football-Predictor-v5.0
Football-Predictor-v5.0 is a full football prediction and backtesting system. It combines data pipelines, feature engineering, model training, probability ensembling, and evaluation across major European leagues.
Highlights
- XGBoost classifier plus Poisson regression models
- Win/draw/loss, expected goals, and over/under probability outputs
- Coverage across five major European leagues and twelve seasons
- Football-Data.co.uk and Understat data integration
- Backtesting, calibration checks, and robustness analysis
Stack
Python, pandas, NumPy, XGBoost, SciPy, scikit-learn, Parquet data pipelines.