← all fixtures
United States
vs
Belgium
Round of 16 · 7/7/2026, 12:00:00 AM
0-1
alt 1-1 · xG 0.91–1.76
5conf
actual 1-4 ✓ correct
Score-probability matrix
0
1
2
3
4
5
6
0
7
12
11
6
1
6
12
10
6
2
5
3
4
5
6
↓ United States goalsBelgium goals →
Win / draw / loss
W 19%D 24%L 57%
Market
37% / 28% / 35%

Belgium favoured (57% to win); the model's most likely scoreline is 0-1.

Confidence 5/10. Even the favoured outcome (57%) fails about 43% of the time; a single match is high variance.

How we got to 0-1
  1. 1
    Model fit
    Team strengths fit from 15,899 matches (Dixon-Coles, recency-weighted, low-score correction ρ=-0.044). Neutral venue — no home advantage.
  2. 2
    Team strengths (log scale, vs average)
    United States: attack +1.01, defence +1.01
    Belgium: attack +1.46, defence +1.30
    Higher attack = scores more; higher defence = concedes fewer.
  3. 3
    Expected goals
    United States: exp(base +0.15 + att +1.01 − opp.def +1.30) = 0.87
    Belgium: exp(base +0.15 + att +1.46 − opp.def +1.01) = 1.81
    “base” is the league scoring level the model fitted.
  4. 4
    Agent adjustments
    λ 0.87/1.81 → 0.91/1.76
    • Context home attack ×1.05 — Neutral venue with no availability issues, so adjustments are minimal. USA's strong recent form (WLWW, 2.5 scored/game) suggests slightly sharper attacking edge and tighter defence than the baseline conservative xG implies, hence small upward nudges; Belgium's steady form is already well captured by the baseline, so no change.
    • Context home defence ×1.03 — Neutral venue with no availability issues, so adjustments are minimal. USA's strong recent form (WLWW, 2.5 scored/game) suggests slightly sharper attacking edge and tighter defence than the baseline conservative xG implies, hence small upward nudges; Belgium's steady form is already well captured by the baseline, so no change.
  5. 5
    Scoreline
    A Poisson model on those expected goals (with ρ) gives the score-probability matrix above; its most likely cell is 0-1 at 12%. Win / draw / loss come from summing the matrix.
  6. 6
    Confidence
    5/10 — a sharper peak and a clearer favourite raise it; large agent moves lower it (model agreement 93%).
United States vs Belgium — predicted 0-1 · Lambda XI