Portugal vs. DR Congo - Consensus / Ensemble Analysis
2026 World Cup, Group K | June 17, 2026 | NRG Stadium, Houston, TX
Kickoff: 12:00 PM CDT (19:00 CEST) | Referee: Abdulrahman Al-Jassim (Qatar)
First match through the full Strategic Decision Framework (doc-5)
What this is: A meta-analysis averaging three model-authored predictions (MiniMax-M3, grok-4.3, glm-5.1), all synthesized from a clean, self-contained source directory (
sources/matches/por-vs-drc/). No human A-version was written — this is the first pure-machine consensus. All source analyses linked at the bottom.
TL;DR
Portugal clear favorite: ~69% / ~21% / ~10%. Top scoreline: POR 2-0 (~19%).
- Decision graph (doc-5) first test: Finding 12 conservation kept LOW (0-3%) — Martínez is aggressive, Ronaldo is milestone-motivated (6th WC, age 41), Portugal has NEVER won a WC (hungry not complacent)
- Rúben Dias OUT (CB, confirmed by Martínez) = -4pt defensive downgrade. NOT gutted (single absence, attack untouched). Finding 4a correctly applied by all 3 models
- Climate NEUTRAL: NRG Stadium roof CLOSED + AC = climate-controlled indoor. Houston's notorious heat is neutralized for the match itself
- Al-Jassim penalty-prone: 0.40-0.45 pens/game (2× elite norm) → ~30-40% pen chance favoring Portugal's dribblers
- DR Congo NOT minnows: AFCON 2025 semifinalists, PL-quality defense (Mbemba, Wan-Bissaka, Tuanzebe). Low block keeps it respectable
- g4 is the outlier again: 78% POR / Ronaldo 38% scorer — both well above mm3 (62%) and g5 (66%). Consensus includes but hedges it
Watch: Does Al-Jassim award a penalty? And does Tomás Araújo (4 caps) hold up as Dias's replacement — if he's exposed, DR Congo's counter (Wissa/Bakambu) could force a tighter game.
✅ LINEUP CONFIRMATION — T-45 pre-kickoff (Jun 17, ~18:15 CEST)
Model VALIDATED. No ensemble re-run required (Finding 4 passes). Official XIs confirm every consensus assumption; no model-breaking surprises.
- Ronaldo STARTS ✅ — #1 scorer pick (~29%) and the milestone/no-sub logic hold (confirmed appearing at a record 6th men's WC).
- GK Diogo Costa (POR) starts ✅ — no model-breaking GK change.
- GK Mpasi (DRC) starts ✅; low block intact (Mbemba, Wan-Bissaka, Tuanzebe all start).
- Tomás Araújo at CB for Dias ✅ — the exact flagged "Watch" variable. Partner is Renato Veiga (not Inácio as some predicted) → both CBs light on caps; marginally reinforces the exposure watch (+1-2% to DRC counter/draw, negligible at headline level).
- Wissa + Bakambu START (DRC) ✅ — counter threat as modeled. Wissa still rusty (sub risk >30% by 60' remains a live in-play watch).
- Minor scorer reshuffle: Pedro Neto starts over Rafael Leão → Leão (predicted POR #3 at ~11%) drops to sub tier (~3-4%); weight shifts to Ronaldo/Bruno/Neto. Does NOT move the win%.
- Headline holds: POR ~69 / Draw ~21 / DRC ~10, top scoreline POR 2-0, #1 scorer Ronaldo (~29%).
- Sources archived:
sources/matches/por-vs-drc/lineups/01-confirmed-xis.md(Bolavip, The Standard, World Soccer Talk, The Athletic).
1. The Experiment
| ID | Author | Model | Notes |
|---|---|---|---|
| mm3 | match-writer agent | MiniMax-M3 | Most conservative on Portugal (62%). Highest draw (25%) |
| g4 | match-writer-g4 agent | grok-4.3 | Outlier: 78% Portugal, Ronaldo 38% scorer — well above the others |
| g5 | match-writer-g5 agent | glm-5.1 | Best decision-graph reasoning. Cleanest Finding 4a application |
All models used the same self-contained por-vs-drc/ directory. No human A-version was written — this is a pure 3-model machine consensus.
2. Win Probability Consensus
All three versions side-by-side
| Version | Author | Portugal | Draw | DR Congo |
|---|---|---|---|---|
| mm3 | MiniMax-M3 | 62% | 25% | 13% |
| g4 | grok-4.3 | 78% | 15% | 7% |
| g5 | glm-5.1 | 66% | 24% | 10% |
Consensus calculations
| Ensemble | Portugal | Draw | DR Congo | Spread on Portugal |
|---|---|---|---|---|
| Machine only (all 3) | ~69% | ~21% | ~10% | 62–78% (16-pt spread!) |
Headline consensus: Portugal ~69% / Draw ~21% / DR Congo ~10%
g4 is the outlier — again. Its 78% is 12-16 points above mm3 and g5. This continues the cross-match pattern: g4 is consistently the most bullish on favorites and hottest on scorer #1 picks. Here g4's Ronaldo 38% scorer probability (vs mm3's 28% and g5's 22%) is the same pattern seen in FRA-SEN (Mbappé 32%), IRA-NOR (Haaland 28%), and ARG-ALG (Lautaro 28%).
Excluding g4 as an outlier: Portugal ~64% / Draw ~25% / DR Congo ~12%. This is closer to the analytically defensible number — g4 over-prices the favorite's margin.
3. The Decision Graph in Action (doc-5)
This is the first match analyzed with the full Strategic Decision Framework. Here's how the graph resolved:
Step 1: Group Context
- Group K: Portugal, Colombia, Uzbekistan, DR Congo
- DR Congo = weakest opponent for Portugal (FIFA #46, first WC since 1974)
- Portugal doesn't strictly NEED the win, but top-of-table control matters → they play to win
- MD1 = positioning, not desperation
Step 2: Finding 12 (Conservation) — LOW discount (0-3%)
The graph asks: should Portugal conserve? The answer runs through every modifier:
| Factor | Value | Conservation impact |
|---|---|---|
| Tier | Tier 1 contender | Normally 5-8% |
| Need-to-win? | No (but want top spot) | Slight reduction |
| Defending champion? | NO (Argentina is) | No statement-mode override... but |
| Manager personality? | Martínez = aggressive | Reduce to 0-3% |
| Milestone-chasing star? | YES — Ronaldo, 6th WC, age 41 | 0% for Ronaldo (hard rule) |
| Never won a WC? | YES — hungry, not complacent | Reduce further |
Result: 0-3% conservation discount. The decision graph correctly overrides the flat "Tier 1 → subtract 5-8%" rule. The modifiers compound: aggressive manager + milestone star + historical hunger = minimal conservation. This is exactly the kind of contextual reasoning the graph was built for.
Step 3: Weather × Style
- NRG Stadium roof CLOSED + AC = climate-controlled indoor
- Portugal is a pressing team, but the heat penalty (>28°C degrades pressing by 60') is NEUTRALIZED
- Finding 11 status: NEUTRAL (perfect conditions indoors)
Step 4: Weather × Defense Stack
- Comfortable (indoor) + DR Congo 5-3-2 low block = block eventually breaks
- DR Congo has PL-quality defenders (Mbemba, Wan-Bissaka, Tuanzebe) → the block is solid, not soft
- The combination predicts: Portugal wins, but the margin is controlled (2-0, not 5-0). The low block + Dias OUT keeps the clean sheet probability below a normal Tier 1 vs Tier 3 match.
Step 5: Temperament
- Both teams well-drilled and disciplined
- Portugal: organized, quality spine, no chaos risk
- DR Congo: Desabre pragmatic, "never lost by more than one goal" — extremely disciplined
- Upset-via-indiscipline: LOW
4. Scoreline Consensus
| Score | Consensus % | Notes |
|---|---|---|
| Portugal 2-0 | ~19% | Modal pick (g4's #1 at 28%, g5's #1 at 16%) |
| Portugal 2-1 | ~14% | mm3's #1 (16%); DR Congo scores on counter/set-piece |
| Portugal 3-0 | ~12% | g4's #2 (18%); blowout if block fractures early |
| Portugal 1-0 | ~10% | Tight win, possibly via penalty (Al-Jassim factor) |
| 1-1 Draw | ~8% | Low block holds + Dias OUT exploited |
Consensus top scoreline: Portugal 2-0 (~19%). The 2-0 is the modal pick for g4 and g5; mm3 preferred 2-1. All agree Portugal wins comfortably but not by a blowout margin.
5. Who Scores Consensus
🇵🇹 Portugal
| Rank | Player | Consensus | Notes |
|---|---|---|---|
| #1 | Cristiano Ronaldo | ~29% (mm3 28%, g4 38%, g5 22%) | Milestone-motivated (6th WC, age 41, all-time WC scorer chase). Penalty taker (Al-Jassim helps). Will NOT be subbed early (milestone rule) |
| #2 | Bruno Fernandes | ~16% | Chief creator, set-piece delivery, arrives late in box. Star-as-creator risk (Finding) |
| #3 | Rafael Leão | ~11% | Explosive 1v1, cuts inside, penalty box threat |
| #4 | Bernardo Silva | ~10% | Dribble-heavy, draws fouls, secondary scorer |
🇨🇩 DR Congo
| Rank | Player | Consensus | Notes |
|---|---|---|---|
| #1 | Cédric Bakambu / Yoane Wissa | ~12% / ~10% | Bakambu: captain, 34yo, 21 goals in 70 caps. Wissa: pace weapon but RUSTY (PCL injury, 4mo out, sub risk >30% by 60') |
| #2 | (either) | ~5% | Set-piece threat from Mbemba/Tuanzebe |
Ronaldo is the consensus #1 at ~29%, but with wide disagreement: g4's 38% is an outlier (its scorer-hot pattern again). mm3 and g5 are closer at 22-28%. The penalty factor (Al-Jassim) boosts Ronaldo's probability because he's the PK taker.
6. The Rúben Dias Factor (Finding 4a — correctly applied)
All three models correctly applied Finding 4a: Dias OUT is a single-player absence worth -3 to -5pt, NOT a 25-30pt "attack gutted" downgrade.
| Model | Dias treatment | Portugal % | Correct? |
|---|---|---|---|
| mm3 | -4pt (CB downgrade, Araújo replaces) | 62% | ✅ |
| g4 | -4pt (single CB absence, attack untouched) | 78%* | ✅ (but over-bullish elsewhere) |
| g5 | -4pt (defensive reshuffle, not gutted) | 66% | ✅ |
All three models avoided the g4 AUT-JOR mistake (where it applied -25pt for Baumgartner). The Finding 4a clarification in doc-1 is working — the agents now correctly distinguish a single absence from a gutting.
*g4's 78% comes from adding +7% back for "comfortable conditions allow block to break" — this is analytically defensible but aggressive.
7. Where Everyone Agrees
- Portugal are clear favorites (~69%). Even mm3's conservative 62% has them favored.
- Conservation discount is LOW (0-3%). The decision graph modifiers (aggressive manager, milestone star, never-won hunger) override the flat Tier 1 rule. This is the first successful application of doc-5.
- Climate is NEUTRAL. The roof-closed + AC finding is critical — Houston's heat is neutralized. Finding 11 doesn't trigger.
- Al-Jassim's penalty rate favors Portugal. ~30-40% pen chance — the highest of any match alongside ARG-ALG (Marciniak) and AUT-JOR (Beida).
- DR Congo are NOT minnows (Finding 5). AFCON 2025 semifinalists with a PL-quality defense. The low block is solid.
- Ronaldo is the consensus #1 scorer. Milestone-motivated, penalty taker, will not be subbed early.
8. Where They Disagree
g4's bullishness on Portugal (78%)
g4 added +7% back for "comfortable conditions allow block to break." This is analytically sound but aggressive — it pushes g4 well above the others (mm3 62%, g5 66%). The consensus includes g4 but the defensible number is closer to 64-66% (excluding g4's outlier).
g4's Ronaldo scorer outlier (38%)
g4's 38% for Ronaldo is 10-16 points above mm3 (28%) and g5 (22%). This is the consistent g4 scorer-hot pattern. The consensus (~29%) includes g4 but the analytically defensible number is ~25%.
Scoreline divergence
g4 picks POR 2-0 at a very high 28% (almost 1-in-3). mm3 and g5 have 2-0 at 14-16%. The consensus (~19%) splits the difference.
9. Key Intelligence from the Gathering Layer
| Intel | Details | Impact |
|---|---|---|
| Rúben Dias OUT | Confirmed by Martínez: "not 100%, not fit for the game." Araújo (4 caps) replaces | -4pt defensive downgrade (not gutting) |
| NRG Stadium roof CLOSED | FIFA decision — climate-controlled indoor. AC system | Finding 11 NEUTRALIZED |
| Al-Jassim penalty rate | 0.40-0.45 pens/game (2× elite norm) | ~30-40% pen chance favors Portugal |
| DR Congo zero-travel | Based IN Houston. Portugal from Florida | Minor freshness edge to DRC |
| DR Congo Ebola quarantine | 21-day isolation, disrupted prep cycle | Partial offset to their travel advantage |
| Wissa rusty | PCL knee injury, 4 months out, only 1 PL goal this season | Sub risk >30% by 60'. Caps DRC counter threat |
| Ronaldo milestone | 6th WC, age 41, chasing all-time WC scoring record | NO conservation, NO sub risk for him |
10. Final Consensus Call
| Outcome | Probability | Confidence |
|---|---|---|
| Portugal win | ~69% (excl. g4 outlier: ~64%) | Moderate agreement (62–78%) |
| Draw | ~21% | Moderate agreement (15–25%) |
| DR Congo win | ~10% | Moderate agreement (7–13%) |
Top scoreline: Portugal 2-0 (~19%). Who Scores #1: Ronaldo (~29%); Bakambu/Wissa (~11%).
The single thing to watch: Al-Jassim's whistle. If he awards a penalty early (30-40% chance), Ronaldo scores it and POR 2-0 is very likely. If the low block holds and Dias's replacement (Araújo) is exposed, DR Congo's counter keeps it tighter (2-1 or even 1-1).
11. Source Analyses (linked)
- mm3 — MiniMax-M3 synthesis: por-vs-drc-2026-pure-football-analysis-mm3
- g4 — grok-4.3 synthesis: por-vs-drc-2026-pure-football-analysis-g4
- g5 — glm-5.1 synthesis: por-vs-drc-2026-pure-football-analysis-g5
12. Pipeline Notes
Fifth full pipeline run. First pure-machine consensus (no human A-version). First match through the full Strategic Decision Framework (doc-5).
Decision graph validation
The doc-5 framework successfully resolved the Finding 12 question that the flat rule would have gotten wrong:
- Flat rule: "Tier 1 contender → subtract 5-8%" → would have over-discounted Portugal
- Decision graph: aggressive manager + milestone star + never-won hunger → 0-3% → correct
All three models independently arrived at a low conservation discount through the graph, confirming the framework's value.
Finding 4a working
All three models correctly treated Dias OUT as -4pt (single absence), not -25pt (gutting). The AUT-JOR clarification is preventing the g4 misapplication from recurring.
Cross-match model patterns (5 matches now)
| Pattern | FRA-SEN | IRA-NOR | ARG-ALG | AUT-JOR | POR-DRC |
|---|---|---|---|---|---|
| g4 = bullish on favorite | 55% | 55% | 62% | 50% | 78% (biggest gap yet) |
| g4 = scorer outlier | Mbappé 32% | Haaland 28% | Lautaro 28% | Arnautović | Ronaldo 38% |
| g5 = best discipline | ✅ | ✅ | ✅ | ✅ | ✅ |
Consensus meta-analysis. No betting odds used. Aggregates three independent model predictions. First match with no human A-version. Source archive: sources/matches/por-vs-drc/