How We Ran the 2026 World Cup 10,000 Times Before a Ball Was Kicked
Most World Cup predictions you read this summer will come down to a hunch dressed up in three paragraphs of build-up. Someone fancies Spain. Someone else thinks France have the squad. The Argentina pieces will write themselves on the back of Messi being Messi.
We wanted to do something a bit more useful than that.
For the past few months we have been building a simulation engine for the 2026 World Cup, hosted at https://worldcup.betminer.co.uk. The model takes the full 48-team field, runs every group game, every knockout tie, extra time, penalties, the lot, and then it does it again. And again. Ten thousand times, end to end. Every team's run through the tournament gets simulated from a probability distribution rather than picked by a person with a favourite shirt.
What you end up with is not a prediction. It is ten thousand predictions stacked on top of each other, and a probability for every team's path through the tournament.
Here is what came out.
The model thinks the field is more open than the bookies do.
Spain finished first across the 10,000 simulations with a 12.9% chance of winning the tournament. That is the highest probability of any team in the field. It is also a really long way short of the kind of headline number you would expect from a "favourite".
That 12.9% is the story.
The World Cup has just expanded to 48 teams. There are now 12 groups instead of eight, a Round of 32 ahead of the Round of 16, and 104 matches in the schedule instead of 64. Every extra knockout round is another coin flip the favourite has to call right. By the time you compound those probabilities across the whole bracket, even the team the algorithm rates highest only lifts the trophy in about one tournament in eight.
Put differently: in 87% of our simulated tournaments, somebody other than Spain wins the World Cup. That is not the model hedging, that is the maths of a 48-team knockout being honest with you.
France came in second on 10.1%. Argentina third on 8.4%. After that the drop-off gets steep, fast. Morocco are fourth on 5.0%, Japan fifth on 4.9%, and England sixth on 3.6%. By the time you reach the bottom half of the seedings the numbers are in fractions of one percent.
How the algorithm actually works
We have written the full methodology up over on the methodology page, but the short version goes like this.
The base layer is international football Elo. It is the same methodology FiveThirtyEight used to use, and it is broadly what FIFA's own ranking system was redesigned around back in 2018.
Every team starts with a rating, every match shifts those ratings up or down based on the result and the strength of the opposition, and the rating gap between two teams gives you a base probability for any matchup. That foundation is well-tested, peer-reviewed in the academic literature, and unsexy in the way good foundations usually are.
What we add on top of that is the Betminer signal layer. Form delta, margin distribution, defensive solidity, recent shape against tactically similar opposition. The Betminer Algorithm has been refined over years of in-season football prediction work and the signals it surfaces are the same ones that drive the daily picks on Football Park and SoccerTips.
The hybrid model takes the Elo probability as its anchor and shifts it based on what the recent signals say. A team that is structurally strong but currently misfiring gets pulled down. A team Elo undervalues because they have been playing weaker opposition gets pulled up. The shifts are bounded, so the model cannot go completely off the rails on the back of a hot run of friendlies, and the output for any given match is a probability distribution rather than a single number.
Then we simulate. Every match in the bracket gets sampled from its distribution, the group tables resolve themselves, the knockout draw plays out, and the tournament has a winner. Run the same setup 10,000 times and you get 10,000 different World Cups. The team that lifts the trophy in the most of them is the model's pick.
What the bracket says
The most likely path through the bracket has Spain beating Argentina in the final. France lose to Spain in one semi, Morocco lose to Argentina in the other. The Morocco run is one of the more interesting subplots in the model: they top a tough Group C ahead of Brazil and Scotland, and the bracket gives them a winnable Round of 16 against England before a quarter-final clash that the model expects them to win.
Yes, that means our model has England losing to Morocco in the quarters. We will come back to that.
Argentina's path is the other one that catches the eye. The defending champions are the third-most-likely winners overall, but the model has them losing the final. Statistically that is not a meaningful drop from second to third (10.1% vs 8.4%), but narratively it sets up a clean storyline of Messi reaching one more final and not getting over the line.
Brazil are the conspicuous absence. The five-time champions came out 9th in the rankings on 1.8%. The bracket has them out in the Round of 16, beaten by Spain.
The bits that are worth a bet
Now to the bit that matters if you actually back football with money rather than just argue about it on the bus.
Two genuine value plays jump out of the model, and one looks like a trap.
Japan to reach the Quarter-Finals at 7/2 is the one we would lead with. Our model has them as the most likely winners of Group F (above Netherlands), through a winnable R32 tie, and into the Round of 16 in a strong majority of simulations. Getting them to the quarters from there is one extra knockout game against the kind of opposition the bracket sends them. Bookmakers pricing 7/2 are giving you a 22% implied probability on that happening. Our model has Japan at 4.9% to win the whole tournament, which means they are reaching the quarter-finals in something north of a quarter of all simulations. The number on the board is short of where our maths puts the line.
Morocco at 150/1 outright is the second one we would look at, on a small stake. They are not lifting the trophy in our base case, but they are reaching the semis in 12.3% of simulations and the final in 6.9%. A 150/1 outright on a team the model has at roughly 1-in-20 to win it is the kind of overlay that exists because the market is still pricing African nations off historical reputation rather than current form.
Brazil at 9/1 is the one to leave alone. The 9/1 implies the market thinks Brazil have roughly a 10% chance of winning. Our model has them at 1.8%. That gap is five times. Either we are catastrophically wrong about Brazil's current state, or every bookmaker in the country is still pricing the yellow shirt by reputation. Our money is on the second one. There may be sharper value in a Brazil exit market than there is in backing them outright.
What this is, and what it isn't
A few things worth saying plainly before anyone gets too excited.
This is a probabilistic model. It is built on the best data we could find, calibrated against historical tournament results, and stress-tested against the recent runs of every team in the field. It is not a crystal ball. The whole point of running 10,000 simulations rather than just picking a winner is that the model is telling you it is uncertain, and showing you the shape of that uncertainty.
Spain at 12.9% means Spain at 12.9%. Not Spain to win. Read it as "if we ran this tournament 100 times, Spain lifts the trophy in about 13 of them." That is a meaningfully different statement to "Spain are going to win the World Cup."
This is also not betting advice. A model output is one input to a betting decision. Bankroll, liquidity, your own read on injuries and form between now and June, all of that sits on top of what we are showing you here.
If you want to dig into the full picture, the predictions page carries the rankings for all 48 teams, the predicted bracket shows the consensus path, and the groups page breaks down every group's most likely finishing order. Worth a proper look ahead of June.
We will be updating the model as squads get announced and the final round of friendlies plays out. The probability shifts get smaller as the tournament gets closer, but they do still move.
For now, the headline reads like this: Spain are the model's favourites, the field is wider than the bookies are pricing it, Japan look like the smartest tournament-stage bet on the board, and someone other than the five-time champions is going to be lifting that trophy in New Jersey on July 19th. Probably.