Group F · 20 June 2026 · neutral venue

Tunisia v Japan

Three views of the match

Tunisia
Draw
Japan

Modelbias-corrected

21%

71%

Sportsbook37 books, vig stripped

19%

27%

54%

Polymarket

17%

26%

56%

Market hedges more than the model

The model puts Japan 17 pp higher than the market — 71% vs 54%.

The most striking gap is on the Japan win: the model assigns 71% while the sportsbook consensus gives only 54%, an 17-percentage-point gap favoring the model's confidence. The model and market both rank Japan as favorite, but the market significantly hedges that view, raising both Tunisia's win probability to 19% (versus the model's 8%) and the draw to 27% (versus 21%). This pattern suggests the market may be pricing in information the model cannot access—such as recent form shifts, lineup changes, or injury news affecting either team's strength—that makes the matchup more competitive than Elo ratings and goal-model fundamentals alone would indicate. The dashboard highlights this divergence as a signal worth monitoring; it does not resolve whether the market's caution or the model's conviction better reflects actual match likelihood.

Preview

Japan enters as the model's clear favorite with a 71% win probability, while Tunisia's chances stand at 8%. The gap reflects a substantial disparity in the forecast; Japan is favored by a decisive margin.

The model's spread suggests a competitive but unbalanced match. A draw at 21% is far from negligible and ranks as the second-most likely outcome, indicating Tunisia can contain Japan for 90 minutes. However, Japan's win probability towers above both alternatives, pointing to a fixture where the favorite is expected to control proceedings and convert that dominance into three points.

The model and market diverge meaningfully here. The sportsbook consensus credits Tunisia with 19% win odds—more than double the model's 8%—and assigns Japan only 54%, a 17-percentage-point gap below the statistical forecast. Polymarket prices fall between the two, closer to the books. The dashboard makes no claim about which assessment is correct; the disagreement flags that Tunisia's true strength may exceed what the model's historical data suggests.

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