Austria vs. Jordan - Consensus / Ensemble Analysis
2026 World Cup, Group J | June 16, 2026 | Levi's Stadium, Santa Clara, CA
Kickoff: 9:00 PM PT (06:00 CEST Jun 17) | Referee: Dahane Beida (Mauritania)
What this is: A meta-analysis averaging four independent predictions — one human-authored (A = original TASK-27) and three model-authored (MiniMax-M3, grok-4.3, glm-5.1), all synthesized from a clean, self-contained source directory (
sources/matches/aut-vs-jor/). All source analyses linked at the bottom.
TL;DR
Austria clear favorite: ~58% / ~27% / ~15%. Top scoreline: AUT 2-0 (~15%). But widest model disagreement yet (11-pt spread on Austria).
- The split is one question: does losing Baumgartner "gut" Austria's attack? g4 says YES (→ 50%); the other 3 say NO (→ 58-63%). Consensus sides with the majority
- g4 misapplied Finding 9: Baumgartner is ONE player (#10), not creator+striker. Should be -3 to -5pt, not -25
- Gathering layer caught a big miss: Jordan's top scorer Al-Naimat is OUT (ACL, 8G/5A in qualifying) — the A-version missed this entirely
- Conditions ideal for the favorite: 19°C night, no heat flag, Austria's gegenpress unleashed
- Beida penalty-prone: 0.46 pens/game in major tournaments → ~25-30% pen chance favoring AUT
Watch: Jordan's striker. If Olwan starts (rusty, 4mo out) the counter has a focal point; if not, the 5-4-1 block deepens and draw rises toward mm3's 32%.
1. The Experiment
| ID | Author | Model | Notes |
|---|---|---|---|
| A | Human (lead analyst) | — | Original TASK-27. Did NOT know Al-Naimat was OUT (ACL). |
| mm3 | match-writer agent | MiniMax-M3 | Highest draw (32%). Correctly treated Baumgartner as -3 to -5pt. |
| g4 | match-writer-g4 agent | grok-4.3 | Major outlier: applied full 25-30pt "attack gutted" for Baumgartner → Austria 50%. |
| g5 | match-writer-g5 agent | glm-5.1 | Best Finding-11 reasoning ("conditions HELP the favorite"). Most bullish on Austria (61%). |
All model runs used the same self-contained aut-vs-jor/ directory. No stale weather file existed (fresh pipeline run).
2. Win Probability Consensus — the widest spread yet
All four versions side-by-side
| Version | Author | Austria | Draw | Jordan |
|---|---|---|---|---|
| A | Human | 63% | 22% | 15% |
| mm3 | MiniMax-M3 | 58% | 32% | 10% |
| g4 | grok-4.3 | 50% | 28% | 22% |
| g5 | glm-5.1 | 61% | 26% | 13% |
Consensus calculations
| Ensemble | Austria | Draw | Jordan | Spread on Austria |
|---|---|---|---|---|
| Machine only (mm3+g4+g5) | 56% | 29% | 15% | 50–61% (11-pt spread!) |
| All four averaged | ~58% | ~27% | ~15% | 50–63% |
| Excluding g4 outlier | ~61% | ~27% | ~13% | 58–63% |
Headline consensus: Austria ~58% / Draw ~27% / Jordan ~15%
This is the widest model disagreement of the tournament. The 11-point spread on Austria (50–61%) is driven entirely by one analytical question: how badly does Baumgartner's absence hurt? The consensus includes g4's caution; excluding it lands at ~61% — which is the more defensible number.
3. The Baumgartner Question — Finding 9 Misapplication
This is the single most important analytical disagreement across all four matches benchmarked:
Finding 9 (Attack Gutted) says: if a team loses its main creator + a legendary striker, downgrade the favorite by 25-30 percentage points (e.g. Uruguay losing De Arrascaeta + Suárez).
| Model | Baumgartner treatment | Austria % | Correct? |
|---|---|---|---|
| A (human) | Not flagged as critical | 63% | ✅ |
| mm3 | -3 to -5pt (creative loss, depth covers) | 58% | ✅ |
| g4 | Full 25-28pt "attack gutted" downgrade | 50% | ❌ misapplied |
| g5 | -2pt (creative loss, Sabitzer covers) | 61% | ✅ |
g4 misapplied Finding 9. Baumgartner is ONE player (the #10/creator). The "attack gutted" rule requires losing the creator AND the striker. Austria still has Arnautović (47 intl goals), Sabitzer (shifted to creator role), Schmid, Gregoritsch, Kalajdžić. This is a manageable creative reshuffle, not a gutting. g4 treated a single-player absence as a 25-30pt downgrade — the rule's trigger condition wasn't met.
Consensus verdict: Baumgartner OUT is worth ~-3 to -5 points, not -25. The correct Austria figure is ~61% (excluding g4's misapplication). The headline consensus (58%) includes g4 as a hedge, but 61% is the analytically sound number.
4. Jordan's Attack IS Genuinely Gutted (the real Finding 9 case)
The gathering layer caught what the A-version missed: Jordan's top scorer Yazan Al-Naimat is OUT (ACL rupture, 8G/5A in qualifying). Coach Sellami: "He cannot be replaced."
- Al-Naimat: 8 goals, 5 assists in WCQ — the focal point of Jordan's entire attack
- His replacement, Ali Olwan, hasn't played competitively since February 5, 2026 (4+ months out, Achilles)
- This forces Jordan into a more defensive 5-4-1 shape — harder to break down, but scores less
- The ironic effect: Jordan becomes more draw-capable (defensive solidity) but less win-capable (no finisher)
This is why Jordan's win probability is consistently low (10-15%) but the draw stays elevated (26-32%): the block is intact, but the counter-attack has lost its cutting edge.
5. Scoreline Consensus
| Score | Consensus % | Notes |
|---|---|---|
| Austria 2-0 | ~15% | Modal pick for 3 of 4 versions (g4 broke the pattern with 2-1) |
| 1-1 Draw | ~14% | The draw scoreline (mm3 weighted heaviest at 16%) |
| Austria 1-0 | ~14% | Penalty or set-piece decider (Beida factor) |
| Austria 2-1 | ~13% | g4's #1 pick; Austria win but Tamari counters |
| Jordan 1-0 | ~7% | The upset: Tamari brilliance + Abulaila heroics |
Top-3 consensus: Austria 2-0, 1-1 Draw, Austria 1-0. First match where the unanimous 2-0 pattern broke (g4 picked 2-1) — another consequence of its Baumgartner over-weighting.
6. Who Scores Consensus
🇦🇹 Austria — unanimous agreement
| Rank | Player | Consensus | Notes |
|---|---|---|---|
| #1 | Marko Arnautović | ~17% | 47 intl goals, penalty taker (Beida's whistle helps). Sub risk >30% by 60-70' (37yo, managed minutes) |
| #2 | Marcel Sabitzer | ~12% | Shifted to creator role (Baumgartner replacement). Star-as-creator risk (Finding 12) |
| #3 | Sasa Kalajdžić / Gregoritsch | ~8% | In-form target options off the bench |
🇯🇴 Jordan — unanimous agreement
| Rank | Player | Consensus | Notes |
|---|---|---|---|
| #1 | Musa Al-Tamari | ~14% | 6G/9A in Ligue 1 this season. Only credible Jordan threat. 24G in 91 caps |
| #2 | Ali Olwan | ~8% | Starting but RUSTY (4mo out). High sub risk. Hat-trick hero vs Oman in qualifiers |
| #3 | Nizar Al-Rashdan | ~5% | Midfield, arrives late |
Strong agreement on both #1s. Arnautović and Al-Tamari are unanimous — the first match with zero scorer disagreement among the top picks.
7. Where Everyone Agrees
- Austria are clear favorites (~58%). Even g4's outlier (50%) still has them favored.
- Finding 11 HELPS the favorite. 19°C night kickoff, no heat flag — g5's key insight: "unlike the depleted-favorite pattern, Austria's press is unleashed, not degraded." Ideal pressing conditions.
- Jordan are NOT minnows (Finding 5). 2023 Asian Cup finalists (beat South Korea). No version treated them as a no-hoper.
- Beida's penalty rate (0.46/game) favors Austria. ~25-30% pen chance — Arnautović is the PK taker. Every version flagged this.
- Al-Tamari is Jordan's only real threat. Unanimous #1 scorer. 6G/9A in Ligue 1.
- Jordan's acclimatization gap is real — 1-2 days in Portland vs Austria's 11 days in coastal CA. All versions applied a Finding 11 travel penalty to Jordan.
8. Where They Disagree — the Baumgartner Question
g4 vs everyone else (covered in §3)
g4 applied the full Finding 9 "attack gutted" downgrade (25-28pts) for Baumgartner's absence. The other three correctly recognized this as a single-player creative loss (-3 to -5pts). This is a Finding 9 misapplication by g4 — the rule requires losing creator + striker, not just one creator.
Impact on the consensus: g4's outlier pulls the machine-only average down to 56% Austria. Excluding it (as a miscalibration), the defensible number is ~61%. The headline consensus (58%) includes g4 as a hedge, but I note that 61% is the analytically correct figure.
mm3's elevated draw (32%)
mm3 pushed the draw higher than anyone (32%), weighting the tournament's 54% draw rate + Jordan's low-block + Abulaila's GK most aggressively. This is defensible — Jordan's gutted attack makes them MORE defensive, which paradoxically increases draw probability.
9. Key Intel the Gathering Layer Surfaced (vs the A-version)
| Intel | A-version | Gathering layer | Impact |
|---|---|---|---|
| Al-Naimat OUT (ACL) | ❌ Not mentioned | ✅ Found: 8G/5A in qualifying, "cannot be replaced" | Model-changing — guts Jordan's attack |
| Baumgartner OUT (thigh surgery) | ❌ Not mentioned | ✅ Found: ruled out of tournament, Ljubicic replaces | Creative loss for Austria (-3 to -5pt) |
| Olwan rusty (4mo out) | ❌ Not mentioned | ✅ Found: hasn't played since Feb 5, Achilles | Jordan's replacement striker is compromised |
| Beida's penalty rate | "unknown, likely moderate" | ✅ 0.46/game in major tournaments (HIGH) | Referee is penalty-prone, favors Austria |
| Jordan acclimatization gap | "slightly further" base camp | ✅ Only 1-2 days in Portland vs Austria's 11 days | Real Finding 11 travel penalty |
| Alaba fitness | "back fit" | ✅ "muscle tightness, MRI clean, expected to start" | More nuanced — 33yo, minutes managed |
The Al-Naimat miss is the biggest single intelligence gap in the A-version. Jordan's top scorer being OUT changes their entire tactical shape (from 3-4-3 counter to 5-4-1 pure block). This is exactly what the injury-sweeper agent is for.
10. The Referee Factor
Dahane Beida (Mauritania) — WC debut, first Mauritanian to officiate at a World Cup:
| Stat | Value | Note |
|---|---|---|
| Career Y/game | 3.78 | Moderate |
| Tournament Y/game | 4.38 | Card-happy at top level |
| Tournament-discounted | ~3.0-3.3 | (25% discount) |
| Penalty rate (major tournaments) | 0.46/game | HIGH — well above the 0.10-15 WC norm |
| Red cards/game | 0.11 | Low red risk |
Penalty probability for this match: ~25-30%, favoring Austria. Beida awarded a VAR-confirmed penalty to Spain at the 2024 Olympics and has controversy history (2025 AFCON QF, Cameroon complained about denied pens). The elevated penalty rate favors Austria's box-attacking style and Arnautović as PK taker.
11. Final Consensus Call
| Outcome | Probability | Confidence |
|---|---|---|
| Austria win | ~58% (excl. g4 outlier: ~61%) | Moderate agreement (50–63%) — widest spread |
| Draw | ~27% | Moderate agreement (22–32%) |
| Jordan win | ~15% | Moderate agreement (10–22%) — g4 outlier inflates |
Top scoreline: Austria 2-0 (~15%). Who Scores #1: Arnautović (~17%); Al-Tamari (~14%).
The single thing to watch: Jordan's confirmed striker. If Olwan starts (expected but rusty), Jordan's counter has a focal point. If he can't go, the 5-4-1 block deepens and the draw probability rises toward mm3's 32%.
12. Source Analyses (linked)
- A — Human, original pre-match (TASK-27): aut-vs-jor-2026-pure-football-analysis
- mm3 — MiniMax-M3 synthesis: aut-vs-jor-2026-pure-football-analysis-mm3
- g4 — grok-4.3 synthesis: aut-vs-jor-2026-pure-football-analysis-g4
- g5 — glm-5.1 synthesis: aut-vs-jor-2026-pure-football-analysis-g5
13. Pipeline Notes
Fourth full pipeline run. The most analytically interesting match yet — produced the widest model disagreement (11-pt spread on Austria) AND the first confirmed Finding 9 misapplication (g4 over-applied the "attack gutted" rule for a single-player absence).
New cross-match findings
- g4 over-applies Finding 9. It treated Baumgartner's solo absence as a full attack-gutting (25-28pt downgrade). The rule requires creator + striker loss. Calibration note for doc-1: Finding 9 needs a clarifier — "attack gutted" means losing 2+ of {main creator, main striker, GK}. A single #10 absence is -3 to -5pt, not -25.
- The injury-sweeper caught the A-version's biggest miss. Al-Naimat OUT (ACL) was not in the A-version at all. This changes Jordan's entire tactical shape. The injury-sweep process continues to be the highest-ROI gathering step.
- First match where the unanimous 2-0 pattern broke. g4 picked AUT 2-1 as #1 — a downstream effect of its Baumgartner over-weighting.
- Finding 11 can HELP the favorite. g5's key insight across IRA-NOR and AUT-JOR: comfortable conditions (no heat flag) unleash the pressing favorite, rather than degrade them. The rule cuts both ways.
Cross-match model patterns (4 matches now)
| Pattern | FRA-SEN | IRA-NOR | ARG-ALG | AUT-JOR |
|---|---|---|---|---|
| g5 = most underdog-sympathetic | ✅ (25% draw) | ✅ (25% draw) | ✅ (27% draw) | ✅ (but 26% draw — mm3 was higher) |
| g4 = scorer outlier | ✅ Mbappé 32% | ✅ Haaland 28% | ✅ Lautaro 28% | ⚠️ (no explicit %, but applied Finding 9 wrongly) |
| Unanimous fav-2-0 as #1 | ✅ | ✅ | ✅ | ❌ (g4 broke it) |
Consensus meta-analysis. No betting odds used. Aggregates four independent predictions. Source archive: sources/matches/aut-vs-jor/