# Methodology

The project emphasizes transparent baselines before complex models. Analysts should document the target variable, feature window, evaluation split, and uncertainty.

| Component | Recommendation |
|---|---|
| Baseline | Start with simple ratings or historical averages. |
| Features | Prefer interpretable features before adding complexity. |
| Metrics | Use accuracy, log loss, Brier score, and calibration notes. |
| Limitations | Explain missing data, tournament uncertainty, and model assumptions. |