Scotland vs. Morocco - Consensus / Ensemble Analysis
2026 World Cup, Group C Matchday 2 | June 19/20, 2026 | Boston Stadium (Gillette), Foxborough, MA
Kickoff: 6:00 PM ET (00:00 CEST Sat) | Referee: Ilgiz Tantashev (UZB)
What this is: Meta-analysis of 3 model-authored predictions (mm3, GLM-5 Turbo, glm-5.2). First match with GLM-5 Turbo replacing grok-4.3. All 22 calibration rules byte-identical via build.mjs. TOML metrics blocks included.
TL;DR
Morocco favored but the DRAW is elevated: MAR ~53% / DRAW ~30% / SCO ~18%. Top scoreline: 1-1 (~15-18%). The headline: TIER-INVERSION — Morocco (Tier 2, must-win) attacks; Scotland (Tier 3, 3pts already) plays for the draw.
- Group C: SCO 3pts (+1), MAR 1pt, BRA 1pt, HAI 0pts. SCO can all-but-qualify with a draw (4pts). MAR must win (desperation).
- TIER-INVERSION is the key analytical call: MAR (Tier 2, AFCON champ) MUST attack. SCO (Tier 3) can defend for the point. The favorite-in-form (SCO on 3pts) conserves; the favorite-by-tier (MAR) goes for it. Finding 12 conservation = 0 on both sides (MAR desperate, SCO Tier 3).
- SCO switched to 3-4-2-1 (defensive posture — Tierney at LCB). Gilmour OUT (knee, tournament-ending).
- MAR attack unchanged (Saibari MD1 scorer + Diaz). Defence experimental (Mazraoui shifted to RCB — exploitable).
- MAR manager = OUAHBI (not Regragui — resigned Mar 2026).
- Venue Gillette: OPEN-AIR, ~27°C (no heat penalty). Both acclimatized.
- Ref Tantashev (UZB): first WC, ~32% pen band (high), 2-4Y expected.
Watch: MAR's experimental CB pairing (Mazraoui at RCB) vs Adams' pace. If it cracks, SCO could steal an upset.
1. Win Probability Consensus
| Model | Morocco | Draw | Scotland |
|---|---|---|---|
| mm3 | 55% | 31% | 14% |
| GLM-5 Turbo (g4) | 52% | 28% | 20% |
| glm-5.2 (g5) | 51% | 30% | 19% |
| Consensus | ~53% | ~30% | ~18% |
Tight — MAR within 4pts. Draw 28-31% (elevated, all agree). GLM-5 Turbo slightly higher on SCO (20%) — the old g4 would have been lower.
g4 (GLM-5 Turbo) vs old grok-4.3
GLM-5 Turbo is more disciplined — SCO at 20% (not over-discounting the underdog), draw respected at 28%, MAR not over-bullied (52% not 58%+). No outlier behavior. Good first showing.
2. Scoreline Consensus
| Score | mm3 | g4 | g5 | Consensus |
|---|---|---|---|---|
| 1-1 Draw | 15% (#1) | 14% (#1) | 18% (#1) | ~16% (modal) |
| 0-1 MAR | — | 14% (#1) | — | ~10-12% |
| 0-0 | — | — | — | ~8% |
| 1-2 MAR | — | — | — | ~8% |
All 3 agree on 1-1 as modal — the tier-inversion (SCO defending, MAR profligate vs the block) produces the classic draw signature.
3. Who Scores Consensus
| MAR #1 | SCO #1 | |
|---|---|---|
| mm3 | Saibari (38%) | McGinn (25%) |
| GLM-5 Turbo | Saibari (22%) | McGinn (16%) |
| glm-5.2 | Saibari (22%) | McGinn (16%) |
| Consensus | Saibari | McGinn |
Saibari unanimous (MD1 scorer, in-form). McGinn unanimous for SCO (MD1 scorer).
4. The Single Thing to Watch
MAR's experimental CB pairing (Mazraoui at RCB, Diop/Riad) vs Adams' pace. If the defence cracks, SCO steals a goal against the run of play → 1-1 or even a SCO upset. If it holds, MAR's possession + Saibari eventually breaks through → 0-1 or 1-2.
Source Analyses (per-model)
Consensus ensemble for TASK-63. Sources: match dir + canonical profiles. 22 calibration rules. Lineups CONFIRMED. TOML metrics blocks included. First match with GLM-5 Turbo (replacing grok-4.3). Kickoff 00:00 CEST Sat.