France vs. Morocco - Post-Event Review

2026 World Cup, Quarter-Final | Thu July 9, 2026 | Gillette Stadium, Foxborough MA (open-air)

Predicted vs. Actual | Calibration: ~85% (B) | ⚠️ 2-AGENT ENSEMBLE (g4 rate-limited)

TL;DR

Result: France 2-0 Morocco (HT 0-0; Mbappé 60' Doué assist, Dembélé 66' Mbappé assist). FRA advance to SF1 vs NOR-ENG winner. FRA's elite CB axis (3 GA in 6 now) strangled MAR's false-9 pivot — the no-finisher thesis held, and a 6-minute 60'-66' burst settled a QF that was scoreless at the hour.

• Winner: ✅✅ France (advance ~60%, 90' win ~47% — the modal favorite on BOTH the advance bet and the 90' bucket). FRA won in regulation; the would-win bet paid cleanly.
• Scoreline: ⚠️ margin off modal — modal was 1-1 (mm3 16%) / FRA 2-1 (g5 13%); actual 2-0. Winner right, margin within 1 goal of the 2-1 modal. FRA's clean sheet was slightly cleaner than predicted; the 25% draw band did NOT fire.
• Who Scores: ✅✅ #1 hit + secondary — Mbappé (#1, ~30-36%) scored (60'); Dembélé (secondary FRA pick) added the 66' second. MAR #1 Ounahi missed (MAR scored 0).
• Finding 17 (4th, no finisher): ✅ — MAR's false-9 pivot (Díaz) produced 0 goals vs the Saliba-Upamecano axis. The model-breaker fired exactly.
• R32-5f (hot-GK diluted): ✅ — Bounou's confirmed form did NOT translate in the open 4-2-3-1; he conceded 2. The block is the system.
• Finding 8 (possession flip): ✅ 6th KO validation — MAR 51.9% / FRA 48.1%. The KO-favorite won with <50% possession + clinical finishing again.
• Tello LOW-PEN: ✅ — 0 pens, 1 yellow (Diop 63'). The inverse-archetype projection held.

Calibration Breakdown: Winner 25/25 · Scoreline 15/25 · Who Scores 15/20 · Findings 20/20 · Model-breaking 10/10 = 85/100 → ~85% band (B)

⚠️ 2-AGENT ENSEMBLE CAVEAT. g4 (glm-5-turbo, xhigh) failed 3× to API rate-limit and did not produce output. The consensus is mm3 + g5 only (1pt spread — tight, flagged as shared anchoring on the MAR xG-regression thesis). g4's validated strength is scoreline precision on clear favorites (FRA fav_share 0.66 qualifies) — its absence makes the modal scoreline the least-robust component. That is exactly where the miss landed (2-0 vs modal 1-1/2-1), so the limitation was material to the scoreline line item, not the winner. The winner + who-scores + findings were anchored on structural analysis that did not require g4.

1-5. Summary

• Decisive narrative: A goalless hour snapped in 6 minutes — Mbappé 60' (Doué assist), Dembélé 66' (Mbappé assist). The QF was tighter than the consensus fav_share 0.66 implied through HT, then FRA's finishing quality told. Two goals, clean sheet, regulation.
• Key player — Mbappé: Goal + assist in the decisive window. The #1 pick over-delivered on involvement (though a single goal, not a brace). Now tied/moving on the WC goals leaderboard — generational-scorer + finishing-mechanism thesis held.
• Lineup note — Doué for Barcola: The LW swap (Barcola→Doué) produced the opener's assist within the hour. Per the brief both XIs started as confirmed; the change was logged pre-match and vindicated immediately. No model-breaking surprise.
• Tactical story — MAR's false-9 failed: Saibari dropped, Díaz false 9, Ounahi LW, Salah-Eddine CB. The restructure produced 0 goals and 1 SoT. The no-specialist-finisher downgrade (F17 4th) was the central MAR thesis and it fired exactly.
• MAR's R16 over-performance reverted: 0.79 xG → 3 goals in the R16 was a +2.2 over-performance; vs FRA it regressed to 0 goals from ~0.5 xG (5 shots, 1 SoT). The regression thesis was the backbone of the MAR downgrade and it held.
• Possession flip (6th KO validation): MAR held 51.9% possession and lost 0-2. The favorite (FRA) won with 48.1% + clinical finishing — the KO-favorite pattern is now 6/6.
• Referee — Tello quiet: 1 yellow total (Diop 63'), 0 pens. The LOW-PEN / low-card projection held; the match was not refereed into the scoreline.

6. Match Stats (FINAL)

Goals: Mbappé 60' (Doué assist), Dembélé 66' (Mbappé assist). Yellow: Diop (MAR) 63'. HT: 0-0. Source: post-event brief, FINAL stats only.

7. Key Learnings

1. ✅ VALIDATE Finding 17 (4th) — no specialist finisher = 0 goals vs elite CB axis. MAR's false-9 pivot (Díaz, with Saibari dropped) generated 0 goals and 1 SoT against FRA's Saliba-Upamecano axis (now 3 GA in 6). The no-finisher downgrade is robust in knockouts: when the underdog's only path is a synthetic/false 9 and the favorite has an elite CB pair, model 0-1 goals, not 2+. This was the central MAR thesis and it fired exactly.

2. ✅ VALIDATE R32-5f — hot-GK is a SYSTEM product, diluted without the block. Bounou's confirmed R16 form did NOT carry into the open 4-2-3-1 — he conceded 2 (both well-worked FRA goals). Confirms the standing rule: confirmed-form GK ×0.5-0.6 only WITH a low block; in an open shape, hot-GK is SWE-negative, not draw-forcing. The pre-match ×0.25 weight (vs the standard ×0.5-0.6) was directionally AND magnitude-correct here — do not restore the higher weight for open-shape opponents.

3. ✅ VALIDATE Finding 8 — possession flip is now 6/6 in KOs. FRA won with 48.1% possession. MEX-ECU, USA-BIH, SUI-ALG, ENG-MEX, BEL-USA, FRA-MAR. The rule — cap projected KO-favorite possession at ~48-52%; the favorite winning with <50% possession + clinical finishing is the default KO pattern, not the exception — is locked. MAR holding 51.9% and losing 0-2 is the cleanest validation yet (the underdog "won" possession and lost by two).

4. ✅ VALIDATE the MAR xG-regression thesis. MAR's R16 (0.79 xG → 3 goals, +2.2 over-performance) reverted to 0 goals vs FRA's elite defense. When a knockout underdog's advance was xG-borrowed, model hard regression vs a top-tier defense (≤1 expected goal). This was the backbone of the fav_share 0.66 anchor and it held.

5. 🔁 PROCESS NOTE — 2-agent ensemble is workable for winner + findings, weak for scoreline. With g4 (scoreline-precision specialist) absent, the modal scoreline was a split (1-1 / 2-1) and the actual 2-0 missed. The structural analysis (false-9, xG regression, hot-GK dilution, low-pen ref) was carried entirely by mm3 + g5 and it all fired — so the winner and findings lines were unaffected by g4's absence. Only the scoreline line was exposed. Refinement: when a scoreline-precision agent (g4) drops out on a clear-ish favorite, explicitly widen the scoreline band in the consensus TL;DR (e.g., "2-0 / 2-1 / 1-1 all live") rather than naming a single modal. Carry forward as a consensus-authoring rule.

8. Knockout Bracket — QF → SF

⚠️ FRA's SF1 opponent is the NOR-ENG winner (QF2, plays Fri Jul 10). Not yet played — do not assume. SF1: Tue Jul 14, Dallas Stadium. FRA are the first confirmed semifinalist.

Post-event review for FRA-MAR QF. Category 0 (score/sources) PASSED: 2-0 confirmed from the scout brief (≥2 sources). Sources: sources/matches/fra-vs-mar/post-event/00-post-event-brief.md. FINAL stats only. ⚠️ 2-agent ensemble (g4 rate-limited) — noted in TL;DR + Learning 5. Calibration 85/100 (~85%, band B) via STRICT rubric.