Per-Agent Calibration Scores: Argentina vs Egypt

Result: 3-2 (90') → 3-2 (ARG win in regulation, advance to QF vs Spain)
Match flow: EGY led 1-0 HT (Ibrahim 15'), 2-0 at 67' (Zico); ARG scored 3 in 11 min (Romero 79', Messi 83', Enzo 90'+2'). Block held 78 min, then collapsed.
Auto-scored baseline: winner_90 15 + advance 10 + scoreline 15 = 40/50 each | who_scores: g4 12, g5 12, mm3 10 (separate)

mm3

Findings: 18/20

#	Finding staked	Direction	Magnitude	Score	Note
1	F11 NEUTRALIZED — roof CLOSED, indoor 20-22°C, "the 33°C CPV struggle will NOT repeat" (Step 3 + TL;DR)	✅	✅	5	Roof closed confirmed in brief; zero heat penalty. ARG press undegraded. Clean.
2	F12 = 0% both sides — ARG defending champ + aggressive Scaloni + Messi milestone → no conservation; EGY historic occasion → max effort (Step 2)	✅	✅	5	ARG came from 0-2 down = unambiguous full effort; no conservation on either side.
3	F4/R32-5f hot-GK Shoubir confirmed-form (×0.5-0.6) + 4-4-2 block = draw-forcer (+3 draw, "EGY 1-1 BEL/IRN/AUS through 90'" template, Step 6 #2)	✅	❌	3	Block held 78 min + draw was live (2-2 at 83', live to 90'+2') — direction right. But Shoubir conceded 3 in 11 min; GK-heroic leg failed. Magnitude over-weighted.
4	Marmoush BENCHED = tactical reshuffle, NOT gutting (−2 EGY, "Zico+Lashin reshuffle is tactical, not gutting — vs −25-30 for gutting", Step 6 #6)	✅	✅	5	Zico (the replacement) scored 67'; brief: "lineup change produced a goal, not a void." Magnitude exactly right — correctly resisted the gutting framing.

Messi milestone (R32-5g → goals, #1 pick) also staked + validated (83', 20 WC goals) — not scored here to avoid double-count with who_scores (auto 10/20, #1 Messi hit).

Model-breaking: 10/10

All material model-breakers caught: GK ✅ match both sides (Dibu + Shoubir, flagged in TL;DR); EGY tactical shock — Marmoush BENCHED + 4-2-3-1→4-4-2 shift — explicitly flagged ("EGY tactical shock"); ARG XI changes (Tagliafico/Paredes/Álvarez promoted) noted; weather (roof closed, F11 neutralized); Letexier ref profile covered. ARG's 0-2→3-2 comeback is outcome variance, not a missed ref/GK/lineup/weather factor. No model-breaker missed.

TOTAL: 40 (auto) + 18 (findings) + 10 (MB) = 68 / 80 → 85% → Band B

g4

Findings: 19/20

#	Finding staked	Direction	Magnitude	Score	Note
1	F11 FULLY NEUTRALIZED — roof closed, climate-controlled, no heat tax (Step 3)	✅	✅	5	Roof closed confirmed. Clean.
2	F12 = 0% both sides — defending champ + aggressive Scaloni + Messi milestone; EGY Tier-3 max effort (Step 2)	✅	✅	5	ARG full effort confirmed (0-2 comeback).
3	F4 hot-GK Shoubir + block compound, DOWN-WEIGHTED (×0.4) — "4-4-2 weaker block than CPV 5-4-1, Shoubir 26 caps not >50, ARG should find it EASIER to score vs EGY's 4-4-2 than CPV's 5-4-1" (Step 4)	✅	✅(mostly)	4	Direction right (block made it close, draw live at 2-2). g4's distinctive down-weighting validated: ARG scored 3 vs only 1-in-90' vs CPV. Sharpest calibration of the three. Not 5/5 because GK still conceded 3 (heroics didn't fully hold).
4	Marmoush benched = reduced counter pace, Rule #13 NOT triggered ("single absences = −3-5pt total, not gutting", Step 6)	✅	✅	5	Zico (replacement) scored; "not gutting" thesis validated.

Messi milestone (R32-5g → goals, #1) staked + validated (83', 20 goals). g4 also correctly flagged ARG R32 fatigue carryover (−3) and Salah 60-65' cap (Salah blanked).

Model-breaking: 10/10

All material model-breakers caught: GK locked (ESPN-confirmed, no change flagged); Marmoush bench + 4-4-2 formation shift caught; ARG fatigue carryover explicitly modeled (−3%); weather (roof closed); ref profile. ARG's comeback is outcome variance, not a missed model-breaker. Clean.

TOTAL: 40 (auto) + 19 (findings) + 10 (MB) = 69 / 80 → 86% → Band B

g5

Findings: 16/20

#	Finding staked	Direction	Magnitude	Score	Note
1	F11 NEUTRALIZED — roof CLOSED, indoor, F11 "does not apply" (Step 3)	✅	✅	5	Roof closed confirmed. Clean.
2	F12 = 0% both sides — milestone = no conservation; EGY max effort (Step 2)	✅	✅	5	ARG full effort (0-2 comeback); no conservation.
3	F4/R32-5f hot-GK confirmed-form (×0.5-0.6) + block = draw-forcer (+4 draw, HIGHEST of the three, Step 6)	✅	❌	3	Block held 78' + draw live (2-2 at 83') — direction right, and g5's higher draw band (25%) was the best-calibrated (draw genuinely ~25% live). But Shoubir conceded 3 in 11 min — the GK-heroic magnitude g5 weighted most heavily was the most over-priced.
4	"EGY attack GUTTED" −8 — Marmoush benched + Salah hamstrung + defensive shift = gutting delta (Step 6)	⚠️	❌	3	Direction partly right (Salah blanked, hamstring confirmed — main threat neutralized). But magnitude wrong: EGY scored 2, Zico (Marmoush's replacement) scored 67', and the brief explicitly states "the lineup change produced a goal, not a void. Hassan's tactical shift wasn't a downgrade for scoring." −8 gutting framing over-stated the downgrade.

Messi milestone (R32-5g → goals, #1 pick 32%) staked + validated (83', 20 goals). g5 also correctly staked Salah "farewell = minutes not goals" (Salah blanked) and had Zico as EGY #2 scorer (Zico scored 67'). These are noted but the 4th findings slot scores the "gutting" magnitude thesis — the contestable differentiator.

Model-breaking: 10/10

All material model-breakers caught (g5 = model-breaker specialist): GK locked; EGY 4-4-2 shift + Marmoush bench caught; g5 even correctly projected Zico as the Marmoush-replacement beneficiary (Zico scored 67'); ARG fatigue carryover (−2) flagged; weather (roof closed); ref profile. Salah hamstring cap correctly flagged (Salah blanked). No material model-breaker missed.

TOTAL: 40 (auto) + 16 (findings) + 10 (MB) = 66 / 80 → 83% → Band B

Cross-agent notes

All three Band B (83-86%). Auto-score identical (40/50): winner_90 right (ARG), advance right (ARG in 90'), scoreline within-margin (15/25 — modal 2-0/1-0 vs actual 3-2). The differentiation is entirely in the qualitative findings.
The two honest differentiators:
1. EGY-reshuffle framing (gutting vs tactical): mm3 (−2) and g4 (−2-3) correctly resisted the "gutting" framing → Zico scored, brief says "not a void" → both 5/5. g5 (−8 "gutted") over-applied it → 3/5. This matches the brief's own learning #5.
2. Hot-GK magnitude: g4 down-weighted Shoubir hardest (×0.4, "weaker 4-4-2, ARG scores easier") and that thesis was validated (ARG 3 vs 1-in-90' v CPV) → 4/5. mm3/g5 weighted it more (+3/+4 draw) → 3/5 (block held but GK conceded 3).
Common correct calls: F11 neutralized (roof closed) and F12 = 0% were clean 5/5 for all three — and both materially validated (no heat penalty; ARG came from 0-2 = zero conservation).
Messi milestone (R32-5g) now 7/7 in KO — all three staked it, all three right, deliberately not double-counted with who_scores.
Brief gaps: pen status ("need to verify if any pens") — no pens in the scorer list (all open play / set-piece), so Letexier pen band (15-25%) likely went 0/1; not material to findings scoring. Possession (ARG 64%) and xG (ARG 2.90) came in HIGHER than all three projected (58-60% / 1.9-2.1) — ARG's volume told, but they were profligate (missed 4 big chances), giving mild support to the sterile-domination flags.

Scores Summary (machine-readable — DO NOT EDIT)

{
  "match": "arg-vs-egy",
  "result": { "score90": "3-2", "final": "3-2", "outcome90": "win", "advance": "a", "route": "90" },
  "agents": {
    "mm3": { "auto": 40, "winner_90": 15, "advance": 10, "scoreline": 15, "findings": 18, "model_breaking": 10, "total": 68, "pct": 85, "band": "B", "who_scores": 10 },
    "g4":  { "auto": 40, "winner_90": 15, "advance": 10, "scoreline": 15, "findings": 19, "model_breaking": 10, "total": 69, "pct": 86, "band": "B", "who_scores": 12 },
    "g5":  { "auto": 40, "winner_90": 15, "advance": 10, "scoreline": 15, "findings": 16, "model_breaking": 10, "total": 66, "pct": 83, "band": "B", "who_scores": 12 }
  }
}