The model rated Switzerland's probability of victory at 73.2%. Qatar's was 8.4%, with the draw at 18.4%. This was the widest disparity on the June 13 card. The final score from Levi's Stadium was 1-1.
What happened at Levi's Stadium
Switzerland's goal came early, a Breel Embolo penalty in the 17th minute after Qatar goalkeeper Mahmud Abunada was booked for a foul in the box a minute earlier. From that point, the match statistics map out a story of almost total Swiss control. They held 68% of possession and took 26 shots to Qatar's 6. They registered 42 touches in the Qatari penalty area; Qatar managed only 8 in the Swiss box.
The dominance in chance creation was even more pronounced. Switzerland generated six big chances and converted one (the penalty). Five were missed. Abunada made five saves in total, and Qatar's defenders produced nine blocks and 31 clearances across 90 minutes of sustained pressure.
The equalizer arrived in the fourth minute of stoppage time. Miro Muheim, who had been substituted on in the 89th minute, headed the ball into his own net while attempting to clear a cross. It was Qatar's most significant offensive moment of the match, and it came from a Swiss player.
The model was right about the process
Our pre-match projection for expected goals was Qatar 0.72 and Switzerland 2.54. The actual on-pitch performance, based on the location and quality of shots taken, produced an xG of 0.60 for Qatar and 3.20 for Switzerland. Switzerland did not just meet the model's expectation of offensive output, they exceeded it by a wide margin. The process that the model anticipated, a lopsided match in terms of chance creation, unfolded even more extremely than the projection suggested.
The gap between generating 3.20 expected goals and scoring one goal (plus conceding an own goal) comes down to finishing variance and a single low-probability event in stoppage time. Our model projects the likelihood of processes playing out over thousands of simulations, and the simulation that produced a 3.20 xG match ending 1-1 would have required something close to what actually happened: a team creating chance after chance without converting, and then an error at the death. An 18.4% draw probability, roughly one time in five, accounts for exactly this kind of scenario.
The per-model breakdown
| Model | Qatar | Draw | Switzerland |
|---|---|---|---|
| Elo | 0.0% | 17.5% | 82.5% |
| Dixon-Coles | 8.0% | 15.4% | 76.6% |
| Historical Poisson | 16.8% | 22.0% | 61.1% |
| Stacking | 4.2% | 12.0% | 83.9% |
| Ensemble | 8.4% | 18.4% | 73.2% |
The Historical Poisson model was the most favorable to an upset, giving Qatar a 16.8% win probability and the draw a 22.0% likelihood. Elo and Stacking were far more confident in a Swiss victory, both above 82%. Elo gave Qatar a literal 0.0% win probability, the most extreme call on any June 13 match. At the bottom of its rating range, Elo compresses small differences into near-zero probabilities that don't reflect the actual tail risk of a football match. The other three models all gave Qatar a nonzero chance, which turned out to be the more defensible position.
Group B from here
Qatar secures a point that few projections would have given them. Switzerland drops two points from a match they dominated by every statistical measure, which opens the group up for Canada and Bosnia and Herzegovina. A Swiss win would have established a clear group hierarchy, but the draw leaves Group B much more competitive heading into the second round of matches.
All probabilities are frozen pre-match model outputs locked before kickoff. Post-match statistics sourced from FotMob. The model publishes probabilities, not recommendations. Methodology: /docs/methodology/. Full Terms of Use.