Spain vs. Belgium — Consensus Ensemble (QF, World Cup 2026)
Fri July 10, 2026 · 21:00 CEST (19:00 UTC / 12:00 PDT) · SoFi Stadium, Inglewood CA (fixed ETFE canopy)
TL;DR
🇪🇸 Spain advance ~55% (ESP win 90' ~41% / Draw ~26% / BEL ~33%). Modal ESP 1-0. ESP's 0 GA in 5 games (only team yet to concede) vs BEL's bench cavalry (Lukaku 14.3% sub-impact) + KDB/Doku returning. The tightest QF — 6pt spread is honest uncertainty.
(3-agent consensus. g4 = Kimi K2.7 Code — first match, replacing glm-5-turbo. g4 most bullish on ESP (59% advance) with tightest scoreline concentration (ESP 1-0 at 18% — classic scoreline-precision behavior, its validated strength). mm3 highest draw band (30%). All 3 agree ESP is the moderate favorite, ~0.60 fav_share.)
- 90' win: ESP ~41% consensus (mm3 38% / g4 46% / g5 40%) — moderate favorite, coin-flip territory.
- Draw band: ~26% consensus (mm3 30% / g4 24% / g5 23%) — mm3 runs highest (draw-hedge strength). Tight QF + Courtois pen pedigree keeps ET live.
- Advance: ESP ~55% (mm3 55% / g4 59% / g5 53%). BEL ~45%. Spread 6pt — honest disagreement.
- Modal scoreline: ESP 1-0 (g4 18% / g5 15%). Runner-up: 1-1 (mm3 13%).
- Who Scores: Yamal/Oyarzabal #1 ESP (~24-26%). De Bruyne #1 BEL (~20-24%).
- Watch: 60' sub window — BEL's bench cavalry (Lukaku 3 bench goals, 14.3% rate) vs ESP's sterile-domination risk. If level at 70', Lukaku + Courtois pen pedigree tilt BEL.
Ensemble Reconciliation
| Agent | Model | 90' ESP | 90' Draw | 90' BEL | Advance ESP | Modal 90' |
|---|---|---|---|---|---|---|
| mm3 | minimax/MiniMax-M3 | 38% | 30% | 32% | 55% | 1-1 (13%) → 2-1 ET |
| g4 | ollama/kimi-k2.7-code:cloud | 46% | 24% | 30% | 59% | ESP 1-0 (18%) |
| g5 | zai/glm-5.2 (xhigh) | 40% | 23% | 37% | 53% | ESP 1-0 (15%) |
| CONSENSUS | — | ~41% | ~26% | ~33% | ~55% | ESP 1-0 |
Agent-strength weighting applied. g4 (Kimi K2.7, first match) shows its scoreline-precision strength — tightest modal concentration (ESP 1-0 at 18%), most bullish on ESP 90' win (46%). This matches the prior g4 (glm-5-turbo) profile: best on clear favorites + scoreline precision. Promising first signal — needs re-validation. mm3's draw band (30%, highest) is its coin-flip strength. g5's BEL 90' win (37%, highest) flags the bench-cavalry edge most aggressively.
Key factors (all 3 agree):
- ESP 0 GA in 5 games — only team yet to concede. The load-bearing +5pt upgrade from WDM anchor.
- BEL bench cavalry (14.3%) — Lukaku 3 bench goals, 3.4× ESP's rate. BEL's advance (45%) exceeds their 90' win% (33%) by 12pt — the bench extends their threat into ET.
- KDB + Doku returning — BEL at full Plan A strength (WDM already priced them in). 4-3-3 → 4-2-3-1 shape change.
- Courtois hot-GK diluted (R32-5f) — confirmed-form but in open 4-2-3-1 → ×0.25-0.3 weight, not ×0.5-0.6. Pen pedigree real → BEL ~55% if pens reached.
- Oliver not high-pen (10-15% pen, 3-4 yellows) — neutral.
- F11 neutralized (22.8°C, SoFi canopy). F12 = 0% (QF).
- WDM fav_share ~0.58-0.60 — moderate favorite, not a blowout. +5pt deviation cap (R32-1) applies.
Where they disagree:
- Draw band: mm3 30% vs g4/g5 23-24%. mm3 sees the QF as more draw-prone (Courtois + tight tactical matchup). g4/g5 see ESP's 0 GA defense as more likely to produce a clean winner.
- BEL 90' win: g5 37% (highest) vs mm3 32% / g4 30%. g5 weights the bench cavalry + KDB return more aggressively in regulation.
- Scoreline modal: mm3's 1-1 (draw) vs g4/g5's ESP 1-0. The classic mm3 draw-hedge vs g4/g5 favorite-precision split.
Consensus reconciled by lead (3-agent ensemble — g4 = Kimi K2.7 Code debut). Spread 6pt (honest uncertainty). Modal: ESP 1-0. Watch: 60' sub window + Lukaku bench cavalry + Courtois pen pedigree if level.