Using a hybrid Elo + Betminer signal model, we simulated the entire 48-team tournament 10,000 times. Group stages, knockouts, extra time, penalties — every match, every time. The result: a probability for every team's path through the tournament, from group stage to final.
One team won more simulations than any other. But the field is closer than you'd think — and the gap between the favourites and the dark horses tells its own story.
Click any team to see their full path through the tournament — group stage finish, knockout chances, expected goals scored and conceded.
Built from 10,000 full tournament simulations. Each slot is filled by the team most likely to reach that position, with knockout rounds derived from the matchups below them.
Group standings are derived from R32 slot occupancy across simulations; knockout winners are the team most often advancing from each matchup. A real tournament will diverge from this path — that's where the upset probabilities live.
For the first time, the World Cup runs on a 12-group format with 32 teams advancing — top 2 from each group plus the 8 best third-placed teams. Here's who our model predicts will make it through.
Our model combines a defensible academic foundation — international football Elo ratings, the same methodology used by FiveThirtyEight and FIFA's own ranking system — with proprietary signals from the Betminer Algorithm. Form delta, margin distribution, and defensive solidity all feed into the simulation alongside base Elo probabilities.
Every match is sampled from a probability distribution rather than picked deterministically. Run the same model 10,000 times and you get 10,000 different tournaments — exactly as it should be. The results above are aggregated across all of them.
This project was produced in partnership with BetClever and Football Park. BetClever is an odds comparison and betting tools platform covering global football markets. Football Park is a football news, predictions and analysis publisher with a fast-growing international audience.