Lionel Messi scored both goals against Austria. That is 18 career World Cup goals, the most by any player in the history of the tournament. France played through a thunderstorm in Philadelphia, the first weather delay of these finals, and beat Iraq 3-0. Norway edged Senegal in a five-goal match. Three groups are now settled at the top.
June 22: the scorecard
| Match | Result | P(result) | Brier |
|---|---|---|---|
| France vs Iraq | 3-0 | 79.3% | 0.073 |
| Senegal vs Norway | 2-3 | 39.8% | 0.546 |
| Argentina vs Austria | 2-0 | 64.5% | 0.198 |
| Algeria vs Jordan | 2-1 | 59.3% | 0.249 |
Mean Brier: 0.267. Three correct directional calls, one miss. Running mean through 44 matches.
Argentina 2-0 Austria: the record
The model had Argentina at 64.5%. They won 2-0. Messi scored both.
The second goal took his career World Cup tally to 18, the most by any player in the history of the men's tournament. He played in 2006, 2010, 2014, 2018, 2022, and now 2026. The previous record, held for years, fell on a Monday night in a group match against Austria.
Austria were not passive. They sat on 3 points from their Matchday 1 win over Jordan and arrived needing a result to stay top of Group J. The model rated the draw at 28.0%, the highest draw probability on the card, precisely because both teams had shown they could defend. Austria held for long stretches. Messi decided it anyway.
Brier: 0.198. Argentina are confirmed as Group J winners with a match to spare. The model read the quality gap correctly and the margin landed close to projection.
France 3-0 Iraq: through the storm
France had the largest Elo gap on the card and the model gave them 79.3%. They delivered 3-0.
The match was delayed by thunderstorms in Philadelphia, the first weather stoppage of the tournament. When play resumed, Kylian Mbappe scored twice, the second a weak-footed effort from distance that beat the goalkeeper at his near post. Ousmane Dembele added the third.
Iraq, ranked far below France, could not absorb the quality difference even with the interruption breaking France's rhythm. The model's expected-goals projection had France comfortably ahead, and the result matched it.
Brier: 0.073. The model's best call of the day. France qualify for the knockout round. Iraq, now 0-2, are eliminated from contention at the top.
Senegal 2-3 Norway: the miss
This was the one the model could not separate. Norway 39.8%, Senegal 32.8%, draw in between. Just over six points of probability split the two teams. Norway won 3-2.
Marcus Pedersen opened the scoring at 43 minutes. Erling Haaland scored twice, a brace from the striker who carries Norway's expected-goals weight. Ismaila Sarr answered with both Senegal goals, one of them in stoppage time, but Senegal could not find the equaliser.
Brier: 0.546. The model rated Norway as the marginal favourite, and Norway won, so the direction was technically right. But the confidence was low, and a five-goal match between two closely-matched sides is exactly the kind of result a near-even rating cannot reward. The penalty reflects a genuine coin-flip, not a blown call.
Norway qualify for the knockouts alongside France. Senegal sit on zero points from two matches and must beat Iraq on the final matchday to have any hope.
Algeria 2-1 Jordan: the comeback
The model had Algeria at 59.3%. Jordan led first. Algeria came back to win 2-1.
Al-Rashdan put Jordan ahead in the first half, and for a stretch the model's projection looked exposed. Then Benbouali equalised in the 68th minute and Gouiri scored the winner in the 82nd. Algeria took all three points from a match they had to win.
Brier: 0.249. The direction held despite the early deficit. Jordan, now 0-2, are eliminated. Algeria keep their knockout hopes alive into the final matchday.
The day's pattern: favourites hold, the coin-flip flips
Three of four matches went to the favourite, and the three the model rated with real confidence (France, Argentina, Algeria) all landed. The single miss was the one the model never claimed to know: Norway vs Senegal, rated barely above even, which produced a five-goal swing.
The contrast with the draw groups is sharp. Groups I and J have now completed, and across eight matches they produced exactly one draw. Decisive results, clear winners, multiple goals. The model's Elo framework is built for precisely this kind of group, where quality gaps translate into scorelines, and the day's mean Brier of 0.267 reflects that fit.
The lesson holds from earlier in the tournament: the model is strong when the gap is wide (France, England-to-come) and honest when it is narrow (Norway-Senegal). A 0.546 on a near-even match is not a failure of the model. It is the model telling you, before kickoff, that it did not know, and being graded on a match nobody could call.
June 23: the chase groups
Groups K and L return for Matchday 2. Both groups produced split results on the opening day, so today decides who controls qualification.
| Matchday 1 | Group | Result |
|---|---|---|
| Portugal vs DR Congo | K | 1-1 |
| Colombia vs Uzbekistan | K | 3-1 |
| England vs Croatia | L | 4-2 |
| Ghana vs Panama | L | 1-0 |
Group K is open: Colombia lead on 3 points, Portugal and DR Congo share the draw, Uzbekistan are bottom. Group L is split between two winners, England and Ghana on 3 points each, with Croatia and Panama both still searching for a point.
| Match | Group | Model H/D/A |
|---|---|---|
| Portugal vs Uzbekistan | K | 62.2 / 25.1 / 12.6 |
| DR Congo vs Colombia | K | 11.6 / 26.4 / 62.0 |
| England vs Ghana | L | 69.7 / 22.5 / 7.8 |
| Croatia vs Panama | L | 60.0 / 24.3 / 15.7 |
Portugal vs Uzbekistan: the response
Portugal drew DR Congo 1-1 on Matchday 1, conceding the equaliser at 95 minutes through Wissa. They sit on a single point and need a win to keep pace with Colombia.
The model gives Portugal 62.2%, Uzbekistan 12.6%, draw 25.1%. The Elo gap is 257 points, with Cristiano Ronaldo leading the line. Uzbekistan arrive on zero points after losing 3-1 to Colombia.
The draw probability at 25.1% carries weight here. Portugal were held once already by a lower-ranked side, and a second slip would leave their qualification in real doubt. The model favours them clearly, but the margin for error is thin.
DR Congo vs Colombia: the favourite away
DR Congo held Portugal to a draw and sit on a point. Colombia beat Uzbekistan 3-1 and lead Group K on 3. The model reads this as 62.0% Colombia, 11.6% DR Congo, 26.4% draw.
The Elo gap is 320 points. James Rodriguez is Colombia's creative axis and the source of much of their expected-goals output. DR Congo, who travelled 13,818km into this fixture, showed against Portugal that they can frustrate a stronger side, but the model sees Colombia's quality as the deciding factor.
A Colombia win would put them on the brink of qualification. A DR Congo result would blow Group K wide open heading into the final matchday.
England vs Ghana: the widest gap on the card
England (the strongest side on today's schedule) against Ghana produces a 517-point Elo gap, the widest of the day. The model gives England 69.7%, Ghana 7.8%, draw 22.5%.
England beat Croatia 4-2 on Matchday 1, Harry Kane scoring twice with Jude Bellingham and Marcus Rashford adding the others. Ghana beat Panama 1-0 with a 95th-minute winner. Both sit on 3 points, but the quality difference is the largest the model has priced today.
Kane carries a 12.5% anytime-scorer probability, the highest individual figure on the card. England need a win to take command of Group L. The model sees this as the most decisive fixture of the day.
Croatia vs Panama: the must-win
Croatia lost 4-2 to England. Panama lost 1-0 to Ghana, conceding at 95 minutes. Both teams enter Matchday 2 with zero points and a defeat to overturn. The model gives Croatia 60.0%, Panama 15.7%, draw 24.3%.
Zlatko Dalic's Croatia were World Cup finalists in 2018 and semifinalists in 2022. A second defeat here would all but end that pedigree at the group stage. Panama, in their first World Cup appearance of this scale, have the same zero points and the same urgency.
The loser of this match is almost certainly out. For Croatia, a side that has reached the last four in consecutive tournaments, the prospect of a group-stage exit makes this the most loaded fixture on the schedule, whatever the model's 60% suggests.
Calibration through 44 matches
The June 22 day average of 0.267 is one of the lowest single-day means of the tournament, pulling the running mean down. Three confident calls landed and the only penalty came from a match the model rated near even.
The story through 44 matches is consistent: the model is sharp when the Elo gap is wide and the result is decisive, and it pays its largest penalties on draws and coin-flip matches between closely-ranked sides. Groups I and J, now complete, were the model's natural territory: decisive results, clear quality gaps, one draw in eight matches. Groups K and L offer a mix, with England vs Ghana the kind of wide-gap fixture the model rates most confidently and Portugal vs Uzbekistan the kind of follow-up where a second draw would sting.
All probabilities are frozen pre-match model outputs, locked before kickoff. The model publishes probabilities, not recommendations. Methodology: /docs/methodology/. Track record: /accuracy/. Full Terms of Use.