World Cup 2026 Group B presents a clash between two teams of similar weight but very different profiles. Canada arrives with home advantage and the best expected offensive generation of the group. Switzerland brings European solidity and tactical discipline to fight for the top spot until the final matchday. Qatar faces the double challenge of improving on its 2022 performance without the host advantage. Bosnia and Herzegovina returns to the World Cup twelve years later with a young squad and an uphill mission. In this entry we break down the MatchScan Ensemble v1 probabilities, the six group fixtures and the most statistically likely qualification scenarios.
The Ensemble v1 (DIBP + Dual SPI + Elo + Bayesian Poisson, calibrated with temperature T=0.85 over 364 validation matches, post-calibration ECE 3.3%) produces probability predictions for each of the 103 World Cup fixtures. The table below shows the average probability of winning each individual match for every team within the group (not the probability of finishing 1st β that's covered in section 4):
| Team | Match win prob (avg) | xG for | xG against | Ξ xG |
|---|---|---|---|---|
| π¨π¦ Canada | 54.8% | 2.09 | 1.08 | +1.01 |
| π¨π Switzerland | 48.8% | 1.89 | 1.17 | +0.72 |
| πΆπ¦ Qatar | 25.6% | 1.16 | 1.75 | β0.59 |
| π§π¦ Bosnia and Herzegovina | 16.5% | 0.97 | 2.11 | β1.14 |
The Ξ xG gap between first and second (Canada and Switzerland) is 0.29 expected goals per match β a much tighter margin than in other groups where the host runs away with it. This reflects that Switzerland is competitive in any scenario that doesn't involve playing Canada in Vancouver. The real structural gap lies further down: between Switzerland and Qatar there's a 1.31-point Ξ xG difference, confirming a clear divide between the leading duo and the two chasers.
Canada arrives as co-host with two of its three matches played at home (Toronto and Vancouver). The model assigns it the best Ξ xG in the group (+1.01) thanks to expected offensive generation of 2.09 goals per match, the highest among the four teams. The post-2022 process has matured: the squad blends established names with European experience and a new generation born in MLS β a mix the model identifies as a sustained competitive edge.
Model strengths β Home advantage with two domestic venues (Toronto on MD1, Vancouver on MD2 and MD3). Consistent offensive generation: 2.09 xG is the highest in the group and among the 10 best in the World Cup. Wing speed that particularly exploits synthetic surfaces like the one at BC Place in Vancouver.
Risks β The match against Switzerland in Vancouver (MD3) is where the model only assigns 33% win probability β a genuinely open game. If Canada arrives needing to win (because Switzerland is already qualified), the psychological pressure of a final matchday at home can weigh.
Best scenario β An opening win vs Bosnia (70% prob.) in Toronto followed by a victory over Qatar in Vancouver (61%) leaves the math practically closed. The double chance (X2) against Switzerland on MD3 then locks in first place.
The Helvetian side combines the best of European national-team football: tactical hierarchy, defensive discipline and a squad with continental experience. It owns the second-best offensive generation in the group (1.89 xG) and the second-best defense (1.17 xG against). Its ability to win the group is real β the model gives it a 37% probability of finishing first, well above the 28% South Korea received in Group A. Switzerland isn't the second seed: it's the other half of the duopoly.
Strengths β Compact defensive block which the model identifies as the best "xG-against / opponent-possession" ratio in the group. Midfield with technique and Champions League experience. Ability to manage games with a lead β something structurally valuable in four-team groups.
Risks β Two Balkan/European opponents (Bosnia) require tactical care due to physicality. Qatar can be surprisingly competitive in tight games β the 24% Qatari win probability on MD1 isn't trivial. If Switzerland drops points on MD1, pressure on the final matchday in Vancouver increases.
Realistic path β Beat Qatar on MD1 (50%), beat Bosnia on MD2 (67%), and draw with Canada on MD3 (37%) adds up to 7 points β enough to lead the group in any scenario.
Qatar faces its second consecutive World Cup without the host advantage it had in 2022. The model records an improvement over the 2022 profile β the team has worked on its physical aspect and transition play β but the qualitative leap isn't enough to compete with Canada or Switzerland. Its realistic hope runs through the MD3 cross with Bosnia, where the model assigns 39% win probability, in what would be a direct duel for third place in the group.
Strengths β Squad with a stable core since 2022, mutual familiarity with collective movements. Tactical discipline that the model picks up in matches where Qatar doesn't need to generate β its profile is more comfortable sitting back than dominating.
Risks β Negative Ξ xG (β0.59) warns that structurally it concedes more than it creates. The defense records 1.75 xG against on average β the third worst in the World Cup among the 48 participating teams. The absence of a home match is a factor the model quantifies at roughly 40 Elo points of loss.
Realistic path β Steal points against Switzerland in the opener (50% of non-loss) and beat Bosnia on MD3 (39%) adds a route toward 4 points. Insufficient to qualify but competitive.
Bosnia returns to the World Cup twelve years after its 2014 debut. The current squad doesn't have the firepower of that Dzeko-Pjanic team, and the model records the structural deficit in numbers: β1.14 Ξ xG and 16.5% average win probability. Its only competitive path runs through a victory in the direct cross with Qatar on MD3 β a match where the model only assigns 29% win probability, reflecting the realistic difficulty of its mission.
Strengths β Physicality and aerial game typical of Balkan teams. Defensive set-piece play is a component the model rates slightly above average. Tactical discipline of the coaching staff.
Risks β The 0.97 xG generated is the lowest figure in the group and among the 5 lowest in the World Cup. The defense concedes 2.11 xG per match against group rivals β numbers that don't allow a realistic qualification math except via sustained upset.
Realistic path β Honestly, the model doesn't produce a realistic qualification scenario. Its mission is to compete with dignity, leave a good impression against Canada and Switzerland, and win the direct duel with Qatar to avoid finishing last.
The six Group B fixtures are split between Toronto (1), Santa Clara (1), Inglewood/Los Angeles (1), Vancouver (2) and Seattle (1). Probabilities reflect the model's 1X2 distribution for each match, expressed as percentages.
Up to this point we've looked at the average match-win probability (how strong each team is in each individual fixture). But the question "who finishes 1st in the group?" is different: it requires winning ENOUGH matches AND collecting MORE points than the other 3 at the end of all 3 matchdays. To answer this we simulate the entire group 100,000 times and count how many times each team finishes 1st, 2nd, 3rd or 4th.
For example: Qatar has a 25.6% probability of winning its typical match (average), but only 9% probability of finishing 1st in the group. The difference is because to finish 1st it has to collect more points than Canada (the 50% favorite) and Switzerland (37%), which is statistically difficult even when winning its matches.
Simulating the six group fixtures 100,000 times with the model's probability distributions, we obtain the following probabilities of finishing the group in each position. These numbers are the most relevant probabilistic information for making informed Battle Royale decisions.
| Team | Prob. 1st | Prob. 2nd | Prob. qualify | Prob. eliminated |
|---|---|---|---|---|
| π¨π¦ Canada | 50% | 33% | 83% | 17% |
| π¨π Switzerland | 37% | 37% | 75% | 25% |
| πΆπ¦ Qatar | 9% | 20% | 28% | 72% |
| π§π¦ Bosnia and Herzegovina | 4% | 11% | 14% | 86% |
The immediate read: Canada and Switzerland form a real duopoly. Canada qualifies in 8 out of every 10 simulations, Switzerland in 3 out of every 4. The fight for the top spot, however, is genuinely open β only 13 percentage points separate them (50% vs 37%), confirming that the Vancouver MD3 match is the most balanced of the entire World Cup group stage. Further down, Qatar has almost double Bosnia's qualification probability (28% vs 14%) β a structural difference derived from its better offensive xG and greater recent World Cup experience.
Another powerful metric: the number of points each team is expected to collect across the three group matches, weighted by the model's probabilities. Results:
| Team | XP | Expected goals (for) | Expected goals (against) |
|---|---|---|---|
| π¨π¦ Canada | 5.75 | 6.28 | 3.25 |
| π¨π Switzerland | 5.24 | 5.69 | 3.52 |
| πΆπ¦ Qatar | 3.15 | 3.48 | 5.26 |
| π§π¦ Bosnia and Herzegovina | 2.24 | 2.90 | 6.32 |
A team typically needs 5 points to secure advancement from a four-team group under the FIFA format. Canada (5.75 XP) and Switzerland (5.24 XP) both clear the threshold β a scenario different from Group A, where only Mexico reached that bar with margin. This reflects the relative strength of the Group B duopoly: two teams above the theoretical qualification cut-off, both with real margin over the two chasers. Qatar (3.15 XP) needs 1.85 points above its expected to qualify β reachable but statistically difficult. Bosnia (2.24 XP) requires beating its expectation by 2.76 points β the least likely scenario of the four.
All the group's math orbits around the matchday-3 encounter in Vancouver between the two duopoly teams. The model sees it as 30% Switzerland, 37% draw, 33% Canada β the most balanced distribution the Ensemble produces across the entire World Cup group stage. The Ξ xG is practically zero (+0.14 for Canada) β a statistically insignificant margin that confirms the genuinely open nature of the match.
The math of what each team needs at the start of this match is the following (assuming the model holds in the previous fixtures):
For Battle Royale bettors: if your MD3 pick was Switzerland, this is exactly the match where the implied market odds typically undervalue the model's real probability. Bookmakers tend to overrate Canadian home advantage by assigning it ~45% implied probability β the model says 33%. The pre-tournament Polymarket draw odds will hover between 3.30 and 3.50; the model gives 37% (implied value at odds higher than 2.70).
The model distributes upset probability mainly across two scenarios:
Scenario A β Switzerland leads the group (37% probability). If Switzerland wins its first two matches (vs Qatar and Bosnia, probabilities 50% and 67%) and at least draws with Canada in Vancouver (64% combined non-loss), it finishes group leader. This isn't a minor upset: it represents 37% of the probability space, almost three out of every eight simulations. For bettors: bookmakers usually set Canada as group-winner favorite at short odds β betting on Switzerland at odds higher than 2.70 has positive mathematical value according to the model.
Scenario B β Qatar steals third place with a point against Switzerland on MD1 (24% non-loss probability for Qatar). If Qatar draws or beats Switzerland in San Francisco, it positions itself as a real threat for third place in the group. Combined with victory in the direct duel against Bosnia on MD3 (39%), Qatar can reach 5 points β a scenario that requires Switzerland to slip at least one more time in its next two matches. The probability of this complete path is roughly 6%, but the individual scenario against Switzerland offers structural value in Asian handicap markets.
The MatchScan Ensemble v1 combines four independent predictive models weighted by their historical accuracy on the consolidated 87,000+ matches database:
DIBP (37.5%) β Dynamic Inter-Team Bayesian Poisson. Goal-generation model with inter-team rating adjustment.
Dual SPI (25%) β Soccer Power Index with separate attack/defense adjustment.
Elo v0 (37.5%) β Dynamic Elo system with opponent-strength adjustment.
Bayesian Poisson (0% in final blend) β Bayesian model that shows instability in asymmetric matches.
The resulting probability set is calibrated via scalar temperature T=0.85 (Platt scaling) with Bayesian correction b_H=-0.24 and b_D=+0.38. Calibration is verified across 364 cross-validation matches, obtaining pre-calibration ECE of 4.68% and post-calibration ECE of 3.26% β very low expected calibration error, indicating that published probabilities are reliable as frequentist expectations.
World Cup 2026 Battle Royale at MatchScan. Pick 8 teams for the tournament. One per matchday. If your team loses or draws, you're eliminated. Last survivor takes β¬100 + 1 year VIP MatchScan. Free entry, no card, no spam.
Sign up β