Methodology

Methodology & Data Sources

Updated: 2026-06-15 · Applies to the dashboard, analysis hub, and blog pages

This site is a model-assisted analytics tool. It does not guarantee outcomes, does not provide betting advice, and avoids claims such as guaranteed wins.

Model inputs

The dashboard uses fixtures, teams, groups, locations, model probabilities, expected goals, score distributions, completed scores, standings context, and public match-center data.

Score and goal predictions

Score forecasts use team expected goals and a Dixon-Coles style score distribution to estimate common scorelines, over/under probabilities, and both-teams-to-score signals.

AI synthesis

The AI module compares the platform model, xG, score distribution, group context, and backtesting signals. It does not claim access to unprovided live odds, injuries, or private information.

Completed-match backtesting

Backtesting compares actual results with model leans, average goal error, over/under direction, BTTS performance, and upset recognition.

Data freshness

Prediction files are generated locally, while scores and lineup information are refreshed from public match-center data when available.