Methodology
AI Football's pitch is simple: every claim on the screen is backed by a number from a citable source. This page lists those sources, the formulas we publish, and what the model cannot see.
Primary data sources
- FIFA rankings — fifa.com/fifa-world-ranking/men. April 2026 snapshot used for the published Group X-Rays.
- Elo ratings — eloratings.net. Cited where Elo gives a better strength signal than the FIFA points system.
- Club statistics (2025-26 season) — FBref. Goals, expected goals, progressive carries, shot-creating actions, pass completion under pressure.
- Player profiles & transfer values — Wikipedia + Transfermarkt.
- Match results & qualification paths — Wikipedia per-tournament pages.
- Photos & flags — Wikimedia Commons (CC BY-SA, attributed in each video's
attributions.json) and flagcdn.com for flags.
Stat-freeze dates
Sports data is alive — a stat-freeze date tells you what the model could see when the video shipped. Every published deep dive declares its freeze date in development.md.
- Group X-Rays (A–H) — frozen at the December 5, 2025 draw plus the most recent FIFA monthly ranking before publication.
- Golden Boot Race 2026 — frozen at the end of March 2026 international break.
- Best XI from Pure Data (0007) — frozen at end of Big Five domestic season + completion of all confederation qualifiers (May 31, 2026).
- $11B World Cup (0001 Economic Impact) — frozen at May 7, 2026 FIFA pricing update.
Published composite formulas
Best XI from Pure Data (0007)
Each position scored on a 100-point composite weighted as follows:
Eligibility: in a 2026 squad, 1,500+ club minutes in 2025-26, 5+ international appearances, not retired internationally.
Group strength claims
Where a group description on /groups/ says "widest rank spread", "highest combined rank", or "toughest by average rank", the underlying formula is:
Each card declares which metric drives its label. April 2026 FIFA rankings are the basis for every published Group X-Ray.
R32 prediction model (Round-of-16 forecast)
A simple Elo-based win-probability model with home-advantage adjustments for matches in the three host countries. Methodology is open — Elo input, +30 for hosts in their own city, +15 for hosts in any host country.
What the model cannot see
Every published video discloses its biases on-screen. Sitewide, the recurring caveats are:
- Big-Five league bias. Advanced statistics (npxG, shot-creating actions, progressive carries) are sparse for the Saudi Pro League, MLS, Brasileirão, and African / Asian leagues. Players in those leagues are systematically underweighted in composite scores.
- Goalkeeper context. Keepers behind dominant defences face fewer high-quality shots and can score worse on post-shot expected goals, even when they are individually elite.
- No "leadership" or "tournament experience" input. All composites are pure 2025-26 output. The model does not credit captaincy, finals played, or international tenure.
- Minutes threshold. A 1,500-minute club minimum excludes players whose 2025-26 season was injury-shortened — even when their per-90 numbers are elite.
- Stat-freeze cutoff. Final 26-man WC rosters are announced in early June. Anything after the freeze date listed for a given video is not yet visible to that video's analysis.
Errors and corrections
If a published number is wrong or out of date, file it in our bug tracker (repo issues) or email contact@aiftbl.com. We track corrections in the per-video development.md.
Last updated: 18 May 2026
