football-predictor

What problem does it solve?

Football analytics for match prediction and betting decisions often require coordinating data ingestion, model development, and ROI-focused recommendations. This skill provides a structured pipeline that combines public data sources, multiple predictive models, and clear EV-based guidance for actionable bets across leagues and seasons.

Core Features & Use Cases

• Data ingestion from public football APIs and sources
• Dixon-Coles Poisson model with Bayesian enhancements and ELO-based predictions
• Model comparison using metrics like AUC, Brier, and RPS
• EV calculation and Kelly criterion-based stake sizing
• Outputs for ML pipelines and dashboards (CSV/JSON, predictions, and datasets)

Quick Start

Run the data collector to fetch today’s matches and generate the ML-ready dataset, then train and compare models to generate actionable predictions