Japan vs Sweden
Group F · Thursday, June 25, 2026, 23:00 UTC · Dallas (Arlington)
The model's read
Japan hold a narrow edge at 41%, with Sweden at 26% and 33% for the draw. Goals project around 1.6–1.2 (both teams to score 53%, over 2.5 52%). The biggest single factor is Overall strength, favouring Japan. Confidence sits at 47/100 but upset risk is high — variance and form keep the underdog live.
Auto-updated every hour as ratings, odds and news change.
Match outcome
oracle-v1.0.0
Japan win
41.1%
Draw
33.0%
Sweden win
25.9%
Expected goals
1.56 – 1.19
Most likely exact score
1-0
top exact 1-0 · 12.1%
Confidence
47/100
Result lean
Draw-leaning
Score band
Likely one-goal game
Goal environment
Moderate
If it opens up — high-scoring alternative
2-2 at 5.5% — most likely scoreline with 3+ goals total.
top exact score carries 12.1% probability
Risk & uncertainty
Upset risk
46/100
Data quality
81/100
Uncertainty band
33 – 63%
Most likely scorelines
Top 12 from a Dixon-Coles adjusted Poisson matrix.
Top 12 most likely scorelines. Heat is relative to the most likely outcome. Japan vs Sweden.
Team comparison
Rating profile across the key dimensions.
Ensemble breakdown
Final probabilities blend the technical model with every available source layer. Persisted 2026-07-11 01:01:06 UTC.
Final (blended) home
41.1%
Final draw
33.0%
Final away
25.9%
Source weights applied
- technical47.1%
- learned16.1%
- market19.3%
- crowd7.5%
- external0.0%
- intel0.0%
- fundamental7.0%
- sentimentMomentum3.0%
Source presence
- technical1 signal
- learned1 signal
- market32 signals
- crowd1 signal
- externalnot used
- intelnot used
- fundamental27 signals
Why this prediction
Source conflict detected — model confidence has been reduced.
Technical
73/100
Fundamentals
50/100
Betting
71/100
Prediction market
94/100
Sentiment
0/100
Top positive drivers
- · technical signal favours home (14.6 pp above uniform)
- · learned signal favours home (13.0 pp above uniform)
- · market signal favours home (12.3 pp above uniform)
Top risk drivers
- · Source conflict: technical vs crowd disagree (conflict score 100)
- · crowd signal is against home (-33.3 pp below uniform)
Missing: external prediction sites · NLP intel (news-derived signals)
Learned model
The learned model contributes to the public ensemble blend, alongside ratings, market signals and news intelligence. Performance is validated against an Elo baseline.
Japan win
46.3%
Draw
39.5%
Sweden win
14.2%
Total goals (λ)
2.50
Model confidence
30/100
Logistic regression model — version wdl-v1, trained on 19k historical internationals.
Article intelligence
Recent news, injury alerts and sentiment extracted by the NLP layer.
Japan
28 articles / 14d
Goals markets
BTTS yes
53.0%
Over 1.5
74.0%
Over 2.5
52.0%
Over 3.5
29.8%
Under 2.5
48.0%
Market intelligence
Model probability
41.1%
Market probability
45.6%
Gap: +2.4 percentage points.
Market confidence: 98/100 across 32 bookmakers.
Odds shown for analysis only — CupCastLab does not place bets, list bookmakers, or recommend wagers.
Explanation
Why the model sees the match this way.
Model prediction: Japan 48% to win, 31% for Sweden, 21% draw. Expected goals: Japan 1.56 vs Sweden 1.19. Most likely exact score is 1-0 at 12.1%, but no exact score is dominant — the score cluster is 1-0 / 0-1 / 1-1. Goal environment moderate (BTTS 53%, Over 2.5 52%). Likely one-goal game. Primary factor: overall strength difference is the dominant driver. Confidence: 47/100. This is a low-confidence call — uncertainty is wide.
Factor contributions
Risk factors
- No notable risk factors identified.
Uncertainty factors
- No major uncertainty flags.
Tactical read
No standout tactical mismatch — the stylistic edge is modest.
Upset risk
High upset risk — variance, tactical fit, or recent form suggests the underdog is live.
Market divergence
Model and market are in close agreement.
These predictions are probabilistic and not guarantees. Not betting advice.