World Cup 2026 · Forecast
Who wins
the World Cup?
A history-plus-squad forecast of all 48 teams, from 1,100,000 Monte-Carlo simulations of the official bracket. The field is tight — the top five sit within a few points, and the model matches the betting market rather than beating it.
Locked, pre-tournament. Headline: Argentina 16.5% · Spain 16.0% · Brazil 9.0% · England 9.0% · France 8.9%.
Fig. V1 Top 8 · with Monte-Carlo error
Champion probability
Argentina and Spain lead, but the gap to the chasing pack is small — and the whiskers (simulation error) overlap.
§ 01
By the numbers
The data and computation behind the forecast, at a glance.
International matches modelled
1872–2026
Tournament simulations
100k locked + 1.0M power re-draws
WC2026 squad players
45 × 26 + 3 × 25
Countries of club data
de-biased squad layer
Engineered features / team
2026 panel width (35 features + team key; 15 historical)
Out-of-sample test matches
across 3 held-out tournaments
Internal consistency
The published probabilities are internally consistent: Σ P(champion) = 1.00 across all 48 teams, and the expected number of teams reaching the Round of 32 is 32 — exactly the 32 knockout slots. Both identities are checked against the locked simulation output at build time.
§ 02
The bracket at a glance
The official 48-team draw, split into two halves and four quarters. Each quarter lists its leading contenders, shaded by champion probability — so you can see where the strength is stacked.
Fig. V5 Knockout draw · shaded by P(champion)
Where the contenders sit
Top half🇪🇸 Spain most likely to reach the Final (26%)
Quarter 1 → Final 53%
Quarter 2 → Final 44%
Quarter 3 → Final 47%
Quarter 4 → Final 56%
Bottom half🇦🇷 Argentina most likely to reach the Final (26%)
Argentina and Spain — the two title favourites — are kept in opposite halves, so they could only meet in the Final, yet each can still meet a fellow top side as early as Round of 32.
§ 03
What sets this forecast apart
The differentiators — read each as a short essay.
We made the squad layer global
Top-five-Europe-only club data biased the model toward European squads. Adding 68 countries of club football removed that bias — and flipped the squad model from worst to best-non-market.
The data story → 02Power vs the draw
Two rankings: raw strength (Power) and fixture-aware reality. The gap is draw-luck — who got an easy bracket and who got a brutal one.
The two rankings → 03Validation
Out-of-sample RPS, calibration, conformal coverage and a subgroup audit — with the limit stated plainly: at n = 3 tournaments, no model difference survives multiplicity adjustment.
The validation →§ 04
Group favourites
Twelve groups. Each card shows the two strongest sides and the dark horse, ranked by P(reach the Round of 16).
- Mexico 54%
- Czechia 36%
Dark horse Korea Republic · 32%
- Switzerland 61%
- Canada 44%
Dark horse Bosnia and Herzegovina · 19%
- Brazil 67%
- Morocco 45%
Dark horse Scotland · 26%
- Türkiye 44%
- Paraguay 36%
Dark horse USA · 29%
- Germany 62%
- Ecuador 51%
Dark horse Côte d'Ivoire · 32%
- Netherlands 52%
- Japan 38%
Dark horse Sweden · 18%
- Belgium 62%
- IR Iran 40%
Dark horse Egypt · 28%
- Spain 71%
- Uruguay 37%
Dark horse Saudi Arabia · 6%
- France 68%
- Norway 51%
Dark horse Senegal · 30%
- Argentina 69%
- Austria 29%
Dark horse Algeria · 25%
- Portugal 64%
- Colombia 58%
Dark horse Congo DR · 11%
- England 69%
- Croatia 47%
Dark horse Ghana · 7%
Source · Oxford Football Forecasting model — P = probability of reaching the Round of 16