Every club watches the World Cup.
Forma watches what it tells us.
Five modules built on the open WC26 dataset: a tactical fingerprint of all 48 nations, a knockout simulator running 10,000 brackets, a Hidden Value Index of tournament risers, a club→country talent pipeline, and a daily Forma Spotter. 64 of 72 matches in. Refreshing daily.
48 nations. One map.
Possession, shot rate, foul rate, set-piece reliance, defensive load, and chaos — distilled into a tactical fingerprint and clustered into six archetypes. The dataset has no event coordinates, so shot rate stands in for verticality and foul rate stands in for pressing. Hover any dot. Filter by archetype.
10,000 brackets. A model strength rating, not a literal forecast.
Per-team xG ratings blended with an elo prior, Poisson-sampled match outcomes, run end-to-end through the remaining bracket. Bracket pairings are seeded by qualification order, not by FIFA's draw rules — these numbers are a strength rating, not a quoted title probability.
Model strength rating — top 12
Most likely finals
- 1Spain v Argentina2.5%
- 2Germany v Argentina2.3%
- 3Spain v Portugal2.1%
- 4Germany v Portugal2.0%
- 5Canada v Argentina1.8%
What this model is missing
Per-team xG-for and xG-against from completed group matches, blended with an elo-derived prior using confidence weighting (matches played / 3). The elo→xG mapping (1.0 + (elo−1500)/600, clamped 0.5–2.2) is an uncalibrated heuristic, not fit to historical data. Match outcomes are Poisson-sampled with team-strength-adjusted expected goals. Extra time uses a 30% scoring rate; penalties resolve with a coin flip. Bracket pairings are seeded by qualification order (R32[0] vs R32[1], etc.) — they are NOT FIFA's official draw, so a given team's strength rating does not correspond to its real path. Model ignores injuries, suspensions, travel, fatigue, home-crowd effects, goalkeeper identity, and the actual draw. Treat outputs as a relative strength rating, not a literal title probability.
Where the World Cup is actually trained.
Club→country minutes flow. Some nations run on the Premier League. Some on Liga MX. A few on leagues you wouldn't expect.
Clubs supplying the most WC26 minutes
- 1Arsenal FCPremier League · 13 players2,400 min
- 2FC Bayern MünchenBundesliga · 12 players2,220 min
- 3Paris Saint-GermainLigue 1 · 13 players2,160 min
- 4FC BarcelonaLa Liga · 12 players2,130 min
- 5Manchester City FCPremier League · 13 players2,070 min
- 6Real Madrid C. F.La Liga · 9 players1,950 min
- 7Liverpool FCPremier League · 8 players1,920 min
- 8Al Hilal SCSaudi Pro League · 9 players1,650 min
- 9Mamelodi Sundowns FCSouth African PSL · 6 players1,620 min
- 10Borussia DortmundBundesliga · 7 players1,410 min
- 11Lille OSCLigue 1 · 8 players1,380 min
- 12SK Slavia PrahaOther · 5 players1,350 min
Domestic share — squads who keep it at home
- ENGEngland81%
- RSASouth Africa73%
- GERGermany70%
- MEXMexico56%
- EGYEgypt44%
- KSASaudi Arabia44%
- ESPSpain39%
- BRABrazil23%
- FRAFrance20%
- AUTAustria18%
- PORPortugal17%
- SCOScotland15%
Minutes — not players — is the unit, because minutes is what wins matches. Club→league mapping covers 72% of squad-minutes; the rest sits in `Other`. Domestic-share treats a nation's own top-flight league as domestic via an exact league→nation map; partial substring matches are intentionally not used (they conflate North/South Korea and the two Congos). Club→league mapping coverage: 72%.
One card. Every match day. Through the final.
One number. One player. One tactical observation. Auto-surfaced from the data — nothing invented.
What this hub is — and isn't.
Built from the open WC26 Kaggle dataset. Aggregated team stats per match, no event-level coordinates. Group-stage sample is 3 matches per team — every model output below labels its sample size and is presented as directional, not as a quoted projection.
Sources. Built from the Kaggle: mominullptr/fifa-world-cup-2026-dataset. Last refresh: 2026-06-27. Snapshot regenerated daily on every site rebuild.
Every model output on this page is presented as directional. None of it is a quoted projection, a recommendation, or a substitute for the kind of work Forma does for our clients — which is built on far richer data and analyst review.
We build analytics that survive a real club's questions.
This hub is a public snapshot of how we think. Our real work — for clubs, on full-season data, with analyst review — goes further than any open dataset can.