Belgium vs. USA β€” Post-Event Review

2026 World Cup, Round of 16 | Mon Jul 7, 2026, 02:00 CEST | Lumen Field, Seattle WA (open-air, 27Β°C)

ELIMINATION Β· UPSET (43% β†’ 4-1) | Predicted vs. Actual | Calibration: ~25% (F)

TL;DR

Result: Belgium 4-1 USA (HT 1-1; De Ketelaere 9', Vanaken 57' sub, Lukaku 90'+3' sub / Tillman 31'). BEL won in 90' and advance to QF. Garcia's deliberate benching of De Bruyne + Doku was a tactical UPGRADE, not "creative gutting" β€” the conservative 3-destroyer XI blew out the home co-host on the counter.

Calibration Breakdown: Winner 0/25 Β· Scoreline 0/25 Β· Who Scores 10/20 Β· Findings 13/20 Β· Model-breaking 5/10 Β· βˆ’5 (Vanaken, unpredicted sub scorer, F19) = 23/100 β†’ ~25% (F band)

Honesty flag (NED-MAR rule): this is a clean miss, not a "B." The modal pick (USA advance ~57%) lost, and lost badly (4-1 in 90'). The two unavoidable zeroes (winner + scoreline) floor the score. The one genuine silver lining is that g5 most accurately diagnosed Garcia's tactic (sit-and-counter + USA's exploitable high line + bench cavalry) yet still concluded USA advance 51% β€” i.e. the pipeline read the mechanism but failed to follow it through to the probability. That is the learning.


1–5. Summary

6. Match Stats (FINAL)

Stat USA BEL
Possession 56% 44%
xG 0.63 2.01
SoT ~2* 7
Fouls 11 9
Reds 0 0
Pens 0 0

Stat line partial in scout brief (ESPN Matchcenter gameId 760507, summary capture). SoT: BEL = 7 confirmed; USA derived (Courtois 1 save + Tillman goal on target β‰ˆ 2). Total shots / yellows not broken out in brief β€” left blank rather than fabricated.

7. Key Learnings

1. REFINE Finding 22's "tournament-team depth absorbs creative gutting" rider β€” extend it from FORCED absences to DELIBERATE ROTATIONS, and add Belgium to the team set. The rider (TUN-JPN: JPN won 4-0 minus 4 of top-5 attackers) lists JPN/NED/GER/BRA/ARG/FRA/ESP/ENG/POR but omits BEL, and only covers injury/forced gutting. The pipeline applied βˆ’8 to βˆ’10pt creative-gutting for the KDB/Doku benching; reality was a 4-1 win. New rule: distinguish forced gutting (injury/suspension β†’ keep the βˆ’8 to βˆ’12pt band) from deliberate rotation (manager benches an underperforming star β†’ neutral or UPGRADE). Signal: the benched starter was sterile in the prior match (De Bruyne: 0-0 vs IRN, 23 shots, 0 goals β€” Finding 13) and the manager swaps in destroyers + a loaded bench = a deliberate sit-counter SYSTEM upgrade, not a void. This is the single biggest pipeline miss of the tournament and the highest-priority rule update.

2. VALIDATE Finding 19 (sub-scorer pattern) β€” strongest validation yet; upgrade the weight. Lukaku's 90'+3' was his 3rd bench goal of the tournament (NZL 86', SEN 86' equalizer, USA 90'+3'); Vanaken added a 57' bench goal. The TASK-141 sub-impact extractor (14.3% bench scoring rate, top-3 in tournament) was RIGHT. The pipeline flagged Lukaku as a sub x-factor (#4, 12%) β€” correct direction, but UNDER-ranked. Refinement: for any team with β‰₯2 proven bench goals in the tournament, elevate the top sub-scorer to a #2/#3 Who Scores pick, not #4, and add +3-5% to that team's 60'+ goal probability.

3. REFINE Finding 8 / R32-5c (possession flip) β€” the reactive-counter shape can produce BLOWOUTS, not just capped 1-0 edges. Previous flips were tight (MEX-ECU 43%, USA-BIH 48%, SUI-ALG 44%). BEL-USA (flip #5) was a 3-goal blowout at 44% possession. Refinement: when the counter-team has a proven bench finisher + an elite distributor waiting (KDB/Lukaku), the possession flip can compound past 60' into a multi-goal margin. Do not cap the counter-team's margin just because they hold <50% ball β€” weight 3+ margin on the reactive side whenever the press-fatigue + bench-cavalry conditions hold.

4. REFINE Finding 22 (blowout) β€” blowout risk attaches to whoever owns the clinical counter + bench finisher, NOT to the pre-match favorite. The pipeline applied the blowout EXCLUSION to BEL (assuming a "gutted" XI couldn't blow USA out) and never weighted a BEL-side blowout. Reality: the "gutted" team blew out the favorite. Refinement: the creative-gutting penalty and the blowout scenario are INDEPENDENT β€” do not let a βˆ’8 to βˆ’10pt gutting penalty also zero out the blowout band. Always run the blowout band on the team with (a) the counter pace, (b) the proven bench scorer, regardless of who the WDM favorite is.

5. VALIDATE Finding 7 (Who Scores) β€” the top-5 list is the deliverable; the single #1 is the least-reliable cell. Both #1 picks missed (Trossard, Pulisic), but the #3 picks hit on BOTH teams (De Ketelaere, Tillman) plus #4 Lukaku. Across R16 the #1 hit rate is weak while secondary picks repeatedly score. Operational rule: keep weighting a #1, but report CONFIDENCE on the top-3 cluster, and never let a single #1 miss read as a total Who-Scores failure.

8. Knockout Progression (R16)

| Stage | Result |
|---|---|---|
| R16 | BEL 4-1 USA (90') β€” BEL advance; USA eliminated ❌ |
| QF | BEL β†’ QF vs ARG-EGY winner (per bracket; ARG-EGY pending β€” do not state unplayed result) |

BEL advance to the QF against the winner of the ARG-EGY R16 tie (source-grounded in the pre-match bracket note). ARG-EGY had not been played at capture time β€” left pending, not fabricated.


12-Category Validation Checklist (compulsory)

# Category Verdict
1 Winner ❌ USA advance ~57% consensus; BEL won 4-1 in 90' (upset). Would-win bet lost.
2 Exact scoreline ❌ Modal 1-1 / USA 2-1; actual BEL 4-1 β€” not in any top-3, blowout on the wrong team.
3 Who Scores βš οΈβ†’βœ… alt #1s missed (Trossard, Pulisic); #3s hit both sides (De Ketelaere, Tillman) + #4 Lukaku sub.
4 Weather βœ… 27Β°C cool, dry; F11 neutral as predicted.
5 Possession ⚠️ USA 56% / BEL 44% β€” flip #5; underdog won with less ball (Finding 8).
6 xG ⚠️ USA 0.63 / BEL 2.01 β€” the "gutted" team was the clinical one; USA was sterile (0 big chances).
7 Cards / pens βœ… 0 pens (Makhadmeh); fouls USA 11 / BEL 9. No ref factor.
8 Lineup βœ… Lineup-checker CAUGHT the KDB/Doku benching (the model-breaker); ❌ interpretation = "gutting" (wrong sign).
9 GK βœ… Courtois confirmed starter (1 save); no GK change.
10 Subs βš οΈβ†’βœ… Bench scored TWICE (Vanaken 57', Lukaku 90'+3'); F19 validated; Vanaken unpredicted (βˆ’5).
11 Finding 12 βœ… 0% conservation both (knockout, need-win) β€” correct.
12 Key learnings Creative-rotation β‰  gutting (rule update); F19 upgrade; F8 blowout extension; F22 independence; F7 top-3 cluster. (See Β§7.)

Per-agent (least-wrong β†’ most-wrong)

Agent USA advance 90' USA / Draw / BEL Modal Verdict
g5 51% 38 / 34 / 28 1-1 Closest + best diagnosis β€” correctly read sit-counter intent + USA high-line vulnerability + bench cavalry; still bet USA.
g4 57% 46 / 22 / 32 USA 2-1 Middle; highest BEL 90' (32%) but USA 2-1 modal wrong.
mm3 62% 48 / 30 / 22 1-1 Most wrong (highest USA advance); drew the gutting line correctly ("not full F4a, βˆ’8pt") but still favored USA.

All 3 agents picked USA; all 3 lost the would-win bet. The 11pt spread (51–62%) correctly signalled genuine uncertainty, but β€” as with GER-ECU / USA-TUR β€” the upset lived in the tail and 0/3 called it. g5's lower USA% + accurate tactical read was the closest the pipeline got; the gap between g5's mechanism-reading and its probability-output is the core failure to fix.


Post-event for BEL-USA (R16). Category 0 PASSED: result BEL 4-1 USA from ESPN Matchcenter (gameId 760507) + scout brief, β‰₯2-source confirmed. Sources: sources/matches/bel-vs-usa/post-event/00-post-event-brief.md (1 file, summary capture β€” full stat line partial). FINAL stats only. Flags: ELIMINATION (USA out), UPSET (43%β†’4-1), STRICT rubric applied (post NED-MAR). Per g5's pre-match bracket note, BEL's QF opponent = ARG-EGY winner (pending).