USA vs Belgium: World Cup 1/8 Final Preview

USA and Belgium meet at Lumen Field in a World Cup 1/8 final that looks almost perfectly balanced in the betting markets but tilts slightly towards Belgium in the model-based predictions. Group-stage standings show USA topping Group D with 6 points from 3 matches (2-0-1, 8 goals for, 4 against, form from standings: WLWW), while Belgium led Group G with 5 points (1-2-0, 6 for, 2 against, form from standings: WWDD). Both arrive as group winners, but by different paths: USA more explosive in attack, Belgium more controlled and unbeaten.

Form deep-dive, using the prediction league blocks, underlines how close this is. USA’s recent World Cup form string is “WWLW” across 4 fixtures, with 3 wins and 1 loss. They have scored 10 goals in those 4 matches, averaging 2.5 per game, and conceded 4 (1.0 per game). Their attack profile is front-loaded: 4 of those 10 goals (50.00%) came between minutes 31-45, and another 2 (25.00%) between 76-90, suggesting strong spells before and after half-time. Defensively, they are vulnerable early: 3 of 4 goals conceded (50.00%) came in the 0-15 interval. Clean-sheet data (2 in 4) and “failed to score” at 0 indicate a high-event team.

Belgium’s prediction-league form is “DDWW”, also over 4 fixtures, with 2 wins and 2 draws and no defeats. They have 9 goals in that span (2.3 per match) and, like USA, 4 conceded (1.0 per match). Their scoring pattern is more back-loaded: 4 of 9 goals (44.44%) arrived between minutes 76-90 and another goal (11.11%) in 106-120, pointing to late-game impact and depth. Defensively, they tend to concede in the 16-30 and 46-75 windows rather than at the very start. Both sides therefore average exactly 1.0 goal against per game, but Belgium’s unbeaten run and late scoring edge suggest slightly better game management.

Injuries and suspensions matter here. USA are without F. Balogun (3 World Cup goals, rating 7.23) due to a red card, plus M. McKenzie (bruised foot) and C. Roldan (muscle bruise). Belgium miss defender Z. Debast with a leg injury. Losing Balogun significantly dents USA’s attacking ceiling and makes their earlier 2.5 goals-per-game average a bit misleading going into this knockout.

Head-to-head data outside friendlies is limited but relevant. In the 2014 World Cup Round of 16 on 2014-07-01 at Itaipava Arena Fonte Nova (Salvador, Bahia), Belgium were the home team and beat USA 2-1 after extra time (0-0 in regular time, 2-1 in extra time). More recently, on 2026-03-28 in a Friendly International at Mercedes-Benz Stadium in Atlanta, USA hosted Belgium and lost 5-2 in 90 minutes. Those two competitive and friendly fixtures, while from different contexts, both show Belgium finding ways to outscore USA, including a very recent five-goal away performance.

The official prediction model gives USA only a 10% chance to win in regular time, with the draw at 45% and Belgium also at 45%. That is a strong endorsement of a tight game where Belgium are at least as likely as USA to progress, and where extra time is a live possibility. Comparison indices back this up: the overall strength index is 47.6 vs 52.4 in Belgium’s favour, and the Poisson index is 52 vs 48, again marginally leaning to the Europeans rather than implying a clear superiority.

Bookmaker odds for the match winner market are extremely compressed. Across major books, USA are between 2.56 and 2.81, Belgium between 2.50 and 2.70, and the draw between 3.25 and 3.50. This aligns closely with the model’s 45/45 draw-split and underlines that the market does not see a clear favourite.

Given the prediction advice “Double chance : draw or Belgium” and the 90% combined probability on draw-or-Belgium from the model, the value-aligned play is to follow that angle. With both teams averaging over 2 goals scored and 1 conceded per match, and Belgium demonstrating strong late scoring, a cautious but data-consistent betting approach is:

• Primary bet: Double chance – Draw or Belgium.
• Leaning on outcome: Belgium to qualify (in extra time or penalties), even if regular time ends level.