World Cup 2026 predictions: models, markets, and odds compared

Serious 2026 World Cup predictions come from four kinds of sources, and they currently agree on the favorite. Prediction markets (Polymarket, Kalshi) are for people who want prices set by real money — Polymarket prices Spain at roughly 17% to win as of June 2026. The Opta supercomputer is for readers who want a pure simulation — it gives Spain 16.1%. Bookmaker odds suit people who just want a quick consensus, margin included. And AI allocation tools like Prever are for people who want those market prices turned into a structured multi-stage position automatically.

The current 2026 picture, source by source

Source	Favorite (price/prob.)	Key limit	Best for
Polymarket winner market	Spain ~17%, France ~16%	You do all analysis yourself; 0.75% sports taker fee	Live, money-backed probabilities
Opta supercomputer	Spain 16.1% (25,000 sims)	Read-only; updates on its own schedule	Model-based context
Kalshi	Champion, group, and match contracts	US-focused; fee 0.07 × P × (1 − P) per contract	US-regulated event contracts
Prever (on Polymarket prices)	Allocates across 7 stages; from ~$5; 1% taker fee in the order	World Cup 2026 only	AI-structured picks without manual pricing

All figures as of June 2026.

Markets vs models: why both matter

A prediction market price is a probability that someone paid for: if Spain trades at 17¢ on a contract that pays $1 if Spain lifts the trophy, the market collectively believes Spain wins about 17% of the time. A model probability is the output of simulating the tournament thousands of times — the Opta supercomputer runs 25,000 simulations and, as of June 2026, has Spain at 16.1%, France 13.0%, England 11.2%, and Argentina 10.4%. When market and model agree, as they broadly do on Spain and France right now, the edge is small. Predictions get interesting where they disagree, and where deeper stage-by-stage markets (group winner, Round of 32 progression) are priced less efficiently than the headline Champion market.

Bookmaker odds are the third reference, but they always contain an overround — the bookmaker’s margin baked into the prices, which is why bookmaker implied probabilities sum to more than 100%. Our docs page on implied probability and overround shows how to strip the margin out before comparing.

What the new 48-team format changes

The 2026 World Cup, hosted by the USA, Mexico, and Canada from June 11 to July 19, 2026, is the first edition with 48 teams in 12 groups, where the top two from each group plus the eight best third-placed teams reach a brand new Round of 32. That format makes single “who wins it all” predictions less informative and stage-by-stage predictions more valuable: a strong team now has seven separately priced checkpoints between the opening match and the trophy. Prever’s Road to Glory builder treats those seven stages — Groups, Round of 32, Round of 16, Quarterfinal, Semifinal, Final, Champion — as one allocation problem, weighting a single stake by money-on-it, live price, and return upside.

Where history fits into 2026 predictions

Head-to-head history is context that neither markets nor simulations show you directly. Across 7,503 recorded internationals covering all 1,128 pairings of the qualified nations, Brazil holds a 13–1 lifetime record against its 2026 group opponents, and Spain is a perfect 8–0 against its own group. Notably, 27 of the 72 group-stage fixtures in 2026 are first-ever meetings between the two nations — matches where history offers no signal at all. The AI Picks page turns this dataset into per-fixture angles you can act on.

If you are ready to put a prediction on, the step-by-step walkthrough lives in how to make your 2026 World Cup picks; if you want to judge any forecaster after the tournament, score them with the methods in Brier score and log loss.

Keep reading

Head-to-head numbers: Michill WC2026 H2H Open Data, CC BY 4.0. Prediction markets carry risk. Not available in restricted jurisdictions.