Group L · 27 June 2026 · neutral venue · Full time

Croatia 2–1 Ghana

Pre-match forecast — frozen before kickoff

Croatia
Draw
Ghana

Modelbias-corrected

78%

19%

Sportsbookvig stripped

54%

29%

17%

Polymarket

54%

29%

16%

Model gave this result 78%, books 54%, Polymarket 54% — model closest.

Most likely scorelines

Model expected goals: Croatia 2.42 − Ghana 0.53 (Croatia +1.89)

2–015.3%
1–012.5%
3–012.3%

Home goals down the rows, away goals across the columns; the gold diagonal is draws. “5+” folds in all higher scores.

Scoreline probabilities come from the uncorrected model — their implied result split won't exactly match the corrected bars above.

Market hedges more than the model · written pre-match

The model puts Croatia 25 pp higher than the market — 78% vs 54%.

The model assigns Croatia a 78% win probability, but the sportsbook consensus puts it at 51%, a 27-percentage-point gap favoring the model's confidence in a Croatian victory. The market's sharply higher probability on Ghana—18.6% versus the model's 2.9%—suggests real-money flows are pricing in information the Elo-based model cannot access: possible recent lineup changes, injury news, or form swings that have shifted expectations toward Ghana. The draw also sits 12 points higher in the market view, consistent with a broader hedging stance. This dashboard surfaces the divergence as a signal of potential model-market information asymmetry, not as evidence either forecast is miscalibrated.

Pre-match preview

The model backs Croatia to win with a 78% probability, leaving Ghana with only a 3% chance of victory. This represents a substantial gap: the model views Croatia as a heavy favorite in this neutral-venue matchup.

The model's spread reflects a confident but not overwhelming outcome distribution. A draw sits at 19%, suggesting the match carries genuine competitive uncertainty in the opening stages, but Croatia's win probability dominates. The 78–19–3 split indicates the model expects Croatia to control play and secure three points more often than not, while draws remain plausible rather than unlikely.

The sportsbook consensus and Polymarket prices show marked divergence from the model. Both betting sources place Croatia's win probability near 51%, some 27 percentage points below the model's forecast. The market assigns Ghana roughly 19% win probability—nearly seven times higher than the model's estimate—and prices draws substantially higher at roughly 30%. This divergence is large enough to signal genuine disagreement between the statistical model and real-money markets. The dashboard does not adjudicate which view is correct.

Written by Claude before the match from the figures above — it has no squad, injury, or form data. See the methodology.