After the shootout drama of Day 2, the Round of 32 settled down. Three matches, three winners inside 90 minutes, no penalties required. France were dominant. Norway were dramatic. Mexico were historic.
France 3-0 Sweden
Model: France 57.0%, Draw 26.0%, Sweden 17.0%. Result: France win. Correct.
France produced the most complete performance of the knockout round so far. Michael Olise was the architect: a short corner to Dembele, who found Mbappe at the byline for the opener in the 45th minute. Then Olise nutmegged Lagerbielke and released Barcola for the second in the 53rd. Mbappe added a bending strike from range in the 74th to complete the rout.
The headline number: Mbappe now has 9 World Cup knockout goals, the most in tournament history. He passed Leonidas and Ronaldo (both on 8) and moved level with Messi on 6 goals at this tournament in the golden boot race.
Brier score: 0.281. The model's best R32 prediction so far. France are doing what the model expected: winning comfortably. They face Paraguay in the Round of 16, where they will be heavy favourites.
Sweden exit at the Round of 32. Their group stage (drew Netherlands 2-2, drew Japan 1-1, beat Tunisia 1-0) earned them a place among the 32, but France on this form were a class apart.
Ivory Coast 1-2 Norway
Model: Ivory Coast 29.2%, Draw 29.7%, Norway 41.1%. Result: Norway win. Correct.
Norway are in the World Cup knockout rounds for the first time, and they won their first-ever knockout match the way Erling Haaland has won so many others: by scoring when it mattered most.
Antonio Nusa opened the scoring with a curling strike from the edge of the area in the 39th minute. Ivory Coast equalized through Amad Diallo in the 74th, a left-footed finish inside the box. The match looked headed for extra time.
Then Haaland. The 86th minute. Patrick Berg played a pass into the box, and Haaland converted from close range. Norway's tournament continues.
Brier score: 0.520. Norway were the model's pick (41.1%), and they delivered, though the margin was tight and the late winner means the Brier reflects the genuine uncertainty.
Norway face Brazil in the Round of 16, one of the marquee fixtures of the next round. Haaland vs Vinicius. Norway's defensive discipline vs Brazil's high press. It is Norway's reward for making history.
Mexico 2-0 Ecuador
Model: Mexico 31.8%, Draw 33.9%, Ecuador 34.3%. Result: Mexico win. Wrong directional call.
The model had this as the tightest three-way split of Day 3. Ecuador were marginally favoured (34.3%), the draw most likely at 33.9%, and Mexico third at 31.8%. In practice, the match was nothing like a coin flip.
Julian Quinones scored in the 22nd minute and Raul Jimenez doubled the lead in the 31st. Mexico were dominant in front of 80,000 at the Estadio Azteca. By half-time, the contest was effectively over.
This was Mexico's first World Cup knockout win since 1986, when they reached the quarter-finals as hosts. Forty years is a long drought, and it ended in the same stadium where it began. The Azteca crowd, the home co-host advantage, the psychological weight of the occasion: the model does not account for any of it.
Brier score: 0.698. The model's miss was modest in probability terms (31.8% vs 34.3% for Ecuador, a gap of 2.5 points) but wrong in direction. Home advantage at the Azteca, against a team that had scored just twice in three group matches, was the variable Elo could not price.
Mexico face England or DR Congo in the Round of 16.
Model scorecard: R32 after 7 matches
| Match | Prediction | 90-min result | Call | Brier |
|---|---|---|---|---|
| Canada 1-0 South Africa | Canada 50.9% | Away win | Correct | 0.368 |
| Germany 1-1 Paraguay (4-3 pens) | Germany 52.5% | Draw | Wrong | 0.840 |
| Brazil 2-1 Japan | Brazil 49.5% | Home win | Correct | 0.384 |
| Netherlands 1-1 Morocco (3-2 pens) | NED 34.8% / Draw 31.8% / MAR 33.4% | Draw | Coin flip | 0.698 |
| France 3-0 Sweden | France 57.0% | Home win | Correct | 0.281 |
| Ivory Coast 1-2 Norway | Norway 41.1% | Away win | Correct | 0.520 |
| Mexico 2-0 Ecuador | Ecuador 34.3% / Draw 33.9% / Mexico 31.8% | Home win | Wrong | 0.698 |
Average R32 Brier: 0.541. Five correct directional calls, two misses (Germany, Mexico), and one match the model correctly flagged as a coin flip (Netherlands/Morocco). The Day 3 average (0.500) was better than Day 2 (0.641), though still above the group-stage baseline (0.557).
The two misses share a pattern: both were near coin-flips where the model gave the wrong side a marginal edge. Germany was 52.5% (wrong by 2.5 points). Mexico was 31.8% vs Ecuador's 34.3% (wrong by 2.5 points). The model is not wildly miscalibrated; it is losing close calls.
Today: England vs DR Congo, Belgium vs Senegal
Two more Round of 32 matches today.
| Match | Venue | UTC | H% | D% | A% |
|---|---|---|---|---|---|
| England vs DR Congo | Mercedes-Benz Stadium, Atlanta | 22:00 | 58.2 | 30.3 | 11.5 |
| Belgium vs Senegal | Lumen Field, Seattle | 01:00 (Jul 2) | 44.8 | 28.7 | 26.5 |
England are strong favourites at 58.2%. DR Congo, who drew Portugal 1-1 in the group stage (Wissa's historic 95th-minute goal), are 11.5%. The winner faces Mexico in the Round of 16.
Belgium vs Senegal is closer. Belgium (44.8%) are favoured but not comfortably. Senegal (26.5%) qualified as one of the best third-placed teams. The winner faces Morocco or Canada.
Bracket shaping up
The Round of 16 picture is filling in:
| R16 match | Date | Venue |
|---|---|---|
| France vs Paraguay | Jul 4 | MetLife, East Rutherford |
| Brazil vs Norway | Jul 5 | MetLife, East Rutherford |
| Morocco vs Canada | Jul 4 | NRG Stadium, Houston |
| Mexico vs England/DR Congo | Jul 5 | Mercedes-Benz, Atlanta |
France vs Paraguay is the Elo mismatch of the round. Brazil vs Norway is the headline fixture: Haaland vs Vinicius, Norway's defensive resolve vs Brazil's high press. Morocco vs Canada is the first all-upset match, two teams who beat higher-seeded opponents.
Probabilities are from the model run frozen before each match's kickoff. The Mexico vs Ecuador predictions (31.8%/33.9%/34.3%) are from the June 30 model run. Draw probabilities represent the 90-minute result; knockout matches proceed to extra time and penalties if level. Brier scores grade the 90-minute outcome. The model publishes probabilities, not recommendations. Full methodology. Full Terms of Use.