The 2026 FIFA World Cup began in North America Earlier this week. The World Cup is the showcase event of the world's most popular sport and easily the biggest sporting event on the planet--FIFA estimates five billion people tuned in at some point to watch the 2022 tournament in Qatar. This edition of the World Cup is expected to eclipse the last one. FIFA projects that the current tournament, hosted across the U.S., Canada, and Mexico with a record 48 teams, could engage up to six billion people globally.
Football/ soccer is played in over 210 countries; basically in every country and territory across the planet. There have been 22 past World Cups since 1930, with 80 different countries participating. Yet, time and time again, the same few countries manage to win the tournament. As shown below, only eight countries have World Cup: Brazil (5), Italy (4), Germany (4), Argentina (3), France (2), Uruguay (2), Spain (1), and England (1). That's it...no one else. Fairly even between Europe (12) and South America (10), but always seemingly the same few countries. A number of factors can help explain this phenomenon, including that European and South American nations benefit from higher league competition density, have deeper player pools, and longer institutional experience in elite tournaments. Is this unusual? Not really...As we noted previously, within Europe's major football leagues a just a few teams similarly dominate.
So, who will win this tournament? The prediction markets,
Polymarket and
Kalshi, provides possibly the best insights into what most people are thinking. Both have former winners Spain (16% Polymarket, 17% Kalshi) and France (17% Polymarket, 16% Kalshi) as the top two favorites (as of June 13). Surprisingly, both also have Portugal (10.8%, 10.5%, respectively) as the third choice; somewhat of an egalitarian choice. Buy hey, the
wisdom of crowd concept does argue that the collective intelligence of a large, diverse group of people can often make better decisions or predictions than any single individual or a small group of experts.
Speaking of whom, what
do the geeks say? By that, we mean Wall Street forecasters of course! And who better than
Goldman Sachs's pedigreed economics team. Goldman, and its German chief economist Jan Hatzius, apply the same approach they might use to make market calls. Goldman's prediction engine is based on nearly 20,000 international matches since 1978, using the
Elo rating system to measure team strength (based on match results and opponent quality). The model incorporates: historical performance, scoring talent, team momentum, geographical factors, and the "winner's slump" to adjust for the challenges of repeating. The bank ran 50,000 Monte Carlo simulations to estimate win probabilities. Spain emerges as the favorite in Goldman's model with a 26% probability of winning the World Cup, followed by France (at 19%), Argentina (at 14%), Brazil (at 8%), and England (at 5%). All previous winners (see above).
Likewise, a team of
European academics also built a World Cup prediction engine utilizing machine learning. Their model blends four strength signals: team abilities from historic results (a bivariate Poisson model with exponential time-weighting), a bookmaker-consensus rating from 24 bookmakers, plus-minus player ratings, and transfer market values to produce probabilistic forecasts based on 100,000 simulations. This model also has Spain as the favorite to win the World Cup at 14.5%, closely followed by England and France at 12.4% each and Germany at 11.2%.
These models (or any model for that matter) almost by definition use comparatively limited information that is a small portion of all the information that’s arguably in the possession of the millions of people that have bet online into prediction markets. But they also have less bias and recency bias, so who knows.
Then there’s AI.
Anthropic’s newest model Claude Fable 5 also predicts Spain (18%) will beat France (14%) in the final, followed by Argentina (11%), England (10%!), Brazil (8%), and…Portugal (7%).
So, whether it's prediction markets, econometric models, or AI it's still the usual suspects that mostly come up as projected winners...
After nearly three days and seven matches, some observations:
