This tracker was an experiment: enjoy the first 48-team World Cup with AI as the assistant — never the oracle. Every probability on the site came from a statistical model; every score, table and record came from verified data; and the AI's only job was to narrate those facts in two languages. It was never allowed to predict, embellish, or invent a number.
✓Every simulation was snapshotted to a public ledger before kickoff — no retroactive edits.
✓The AI wrote from verified fact blocks only; when a fact was missing it stayed silent.
✓Where the model lost to the betting market, the site said so on the page.
Lamine Yamal 10′ · M. Oyarzabal 21′ · M. Oyarzabal 24′
A. Baena 42′
M. Oyarzabal 36′ · P. Porro 66′ · M. Oyarzabal 89′
M. Merino 90′
F. Ruiz 30′ · M. Merino 88′
M. Oyarzabal 22′ (p) · P. Porro 58′
F. Torres 106′
The tournament in numbers
308 goals in 104 matches · 2.96 per game
Goals by stage
Group stage215 goals · 72 matches
Round of 3242 goals · 16 matches
Round of 1623 goals · 8 matches
Quarter-finals12 goals · 4 matches
Semi-finals5 goals · 2 matches
Third place10 goals · 1 matches
Final1 goals · 1 matches
The Golden Boot race
1Kylian Mbappé10 goals
2L. Messi8 goals
3J. Bellingham7 goals
4E. Haaland7 goals
5
The title odds, snapshot by snapshot
Our model's champion probability for both finalists, from the public bracket ledger — every point written before kickoff, none edited after.
O. Dembélé
6 goals
Spain100%Argentina67%
Model vs reality — the final ledger
Twice a day, the model published its simulation to an immutable ledger, then graded itself against what actually happened. This is the record it earned.
Round of 3232 of 32 called
Round of 1615 of 16 called
Quarter-finals8 of 8 called
Semi-finals4 of 4 called
Final2 of 2 called
And the honest part
Match by match, the betting market stayed sharper than our model — the ledger says so plainly, and that honesty was the point. The model's probabilities came from math alone; the AI never forecast anything.
51/80 matches called
our model
57/80 matches called
the market
Contrarian calls that landed
France – Spain · Spain — our model 45%, the market 30%
England – Argentina · Argentina — our model 45%, the market 32%
USA – Belgium · Belgium — our model 45%, the market 34%
The closing wordAI
Spain claimed the World Cup 2026 title with a 1-0 victory over Argentina in the final, decided in extra time. The champions' campaign was exceptional: across eight matches they won seven and drew one, surrendering just a single goal while posting seven clean sheets. They scored 14 goals in that run, establishing themselves as the tournament's most dominant force.
The wider tournament produced 308 goals across 104 matches, averaging 2.96 per game. Our statistical model tracked its own predictions openly throughout the tournament. In the group stage it called all 32 matches correctly, went 15 of 16 in the round of 16, and achieved perfect scores in the quarterfinals, semifinals, and final. Across eighty tracked matches, the model called fifty-one correctly. The betting market, by contrast, called fifty-seven.
Spain entered the tournament with minimal backing: our model assigned them just 2 percent probability. The math never promised certainty—only a framework for reasoning through uncertainty. That distinction proved crucial. The numbers computed every prediction, the AI only narrated the story as it unfolded.
AI-written from the tournament's verified facts — the last note of the campaign.
Cape Verde Islands0-0Atlanta
Saudi Arabia4-0Atlanta
Uruguay1-0Zapopan
Austria3-0Inglewood
Portugal1-0Dallas
Belgium2-1Los Angeles
France2-0Arlington
Argentina1-0East Rutherford
