World Cup 2026 Predictions vs Results: What Markets Got Right — and Wrong

Last updated: July 2026  ·  9 min read

Forecasts are cheap before a tournament. The interesting moment comes afterwards — or, in this case, midway through — when predictions meet results and it becomes possible to ask the only question that matters: were the probabilities any good?

With the 2026 World Cup at the quarter-final stage, this article audits the pre-tournament consensus — including our own coverage — against what has actually happened. Brazil is out. Portugal is out. Both the United States and Mexico fell in the round of 16. Cape Verde nearly produced the greatest upset in World Cup history. Some of this was priced. Much of it was not. Sorting one from the other is where the real forecasting lessons live.

What the Markets Got Right

France as the standing favourite

Pre-tournament markets consistently placed France at or near the top of the outright board, and that call has aged well: three group wins, controlled knockout victories over Sweden and Paraguay, and Kylian Mbappé tied for the Golden Boot lead with seven goals. France’s current price — implying roughly a 35% title probability — is the market’s strongest pre-tournament position still standing.

The general shape of knockout risk

Our pre-tournament upset predictions argued that the expanded format would concentrate variance in the early knockout rounds, where a single bad afternoon eliminates a favourite. That is essentially what happened: the round of 32 and round of 16 removed Brazil, Portugal, the USA, and Mexico, while Argentina survived extra time against Cape Verde and a late collapse-and-recovery against Egypt.

Directional scepticism about co-host ceilings

Markets never treated the USA or Mexico as genuine contenders, despite home advantage — a stance our analysis of whether the USA could win on home soil broadly shared. Both co-hosts reached the round of 16 and no further: Mexico lost 3-2 to England, and the USA was beaten 4-1 by Belgium. Painful locally, but well within what the prices implied.

What the Markets Got Wrong

Brazil’s floor

Brazil entered the tournament priced among the top contenders. The team then delivered a flat group stage and a round-of-16 exit to Norway — its earliest World Cup elimination since 1990. The failure here was not predicting a Brazil loss in one match; single-game upsets are always live. It was the systematic overweighting of squad reputation over recent underlying form, a bias that recurs at almost every major tournament.

Norway’s ceiling

Norway went 8-0 in qualifying with two wins over Italy, scoring 37 and conceding five — data that was fully public before a ball was kicked. Markets still filed Norway under “dangerous outsider” rather than “quarter-final calibre”. Our own dark horses analysis flagged the underpricing, but candidly, nobody’s central case had Norway eliminating Brazil. The forecast error was treating Erling Haaland’s finishing as a bonus on top of a mid-tier team, rather than as a structural factor that changes single-match win probabilities.

Cape Verde, and the tail nobody priced

An island nation of roughly half a million people, ranked 67th in the world, held Spain to a goalless draw, drew 2-2 with Uruguay, and then took defending champion Argentina to the 111th minute of extra time before losing 2-1. No mainstream forecast assigned meaningful probability to that sequence. The lesson is not that models should have “predicted Cape Verde” — it is that tail outcomes in expanded tournaments are systematically fatter than intuition suggests, because more entrants means more draws from the extreme end of the distribution.

The Format Question: More Chaos, Not Less

A common pre-tournament argument held that the 48-team format would protect big nations: softer groups, more qualification slots, fewer early landmines. The opposite occurred — this group stage produced more genuine surprises than any recent edition. Australia beat Turkey, many observers’ dark horse. South Korea exited despite an opening win over Czechia. Tunisia lost 5-1 to Sweden and fired its manager after a single matchday. Scotland barely registered against Morocco and Brazil.

Why did the safety-in-numbers logic fail? Partly because more teams means more matches in which something unlikely can happen, and partly because the gap between football’s established and emerging nations continues to narrow faster than reputational pricing adjusts. Our pre-tournament group stage predictions captured some of this risk — and still underestimated its magnitude.

What This Says About Forecasting Itself

It would be easy to read this tournament as an indictment of prediction markets. That reading misunderstands what probabilistic forecasts claim. A market that gave Brazil, say, a 12% title probability was not asserting Brazil would reach the semi-finals — it was asserting that in roughly one tournament in eight, a team like this exits early. Sometimes you are living in that one-in-eight tournament. The question is whether, across many events, the frequencies match the prices — the calibration question we examined in our review of prediction market accuracy.

The Scorecard, Honestly Kept

Halfway through the first 48-team World Cup, the fair summary reads: the market’s central forecasts (France strong, hosts limited, chaos in early knockouts) have held; its specific mispricings (Brazil’s durability, Norway’s ceiling, the Cape Verde tail) were real and instructive. Anyone claiming they saw all of this coming is describing hindsight, not foresight. The honest position — for markets and for analysts alike — is a running scorecard, updated while the tournament can still prove it wrong. Four quarter-finals will do exactly that within the week.