Methodology

This section explains transparently how Calciometrica builds its composite indicators and how the automated editorial process works. The goal is to let you judge the reliability of every number you see on the site, and, if you want to integrate our data into your own analysis, to know exactly what each metric represents.

Pages

• How we compute the AI Score: inputs, model, confidence band, limits.
• How we compute form (momentum): exponentially weighted moving average over the recent matches, role-specific weights.
• How editorial fact-checking works: generation pipeline, claim verification, publication thresholds.

Versioning

Each page records its current methodology version at the bottom. When we change a formula non-trivially (e.g. moving from v1 to v2), we flag it here and in the release notes. This helps anyone tracking rankings over time understand any drift.

General limits

Our indicators are synthetic measures based on the data available at compute time: they are not predictions of future performance, nor recommendations for betting or operational decisions. We recompute after every matchday; the model and parameters may evolve.

Methodology version: v1, last updated April 2026.