The working cloud preserved Canada scoring too aggressively.
Project Turtle / settled analysis / July 4, 2026
Canada 0-3 Morocco
This file is settled. Turtle read Morocco direction and the low-total support correctly, but the public exact-score cloud missed because the clean-sheet spread family was not protected before exact-score selection.
Four-score cardMiss
ContainmentMiss
Support marketHit
Carry-forward ruleOpponent-threshold spread first
Pregame and postgame analysis
The file stays honest: the prediction missed, the correction is now written.
Pregame analysisThe late-lock cloud expected Morocco pressure but did not preserve the Morocco clean-sheet spread thread.
The working pregame card leaned toward Morocco without enough discipline about the exact shape of that advantage. The four-score card stayed around one-goal and BTTS-adjacent states instead of proving the clean-sheet ladder first.
The relevant missed family was not a random outlier: 0-1, 0-2, and 0-3 all describe Morocco control with Canada failing to score. The old gate overvalued Canada's raw scoring rate and undervalued opponent class.
Postgame analysisMorocco won 3-0; Ounahi scored twice, Rahimi added the stoppage-time extension.
At 90+5 the match was already Canada 0-2 Morocco. Final settlement was Canada 0-3 Morocco after the late third goal. That makes the four-score exact-card a miss and the Under 3.5 support a hit.
The failure class is now named: clean-sheet spread underweight. Turtle saw Morocco as the stronger result bucket, but it did not ask the opponent-threshold question strongly enough: how does Canada score against a Morocco-level control side, and how often does Morocco convert that matchup into a clean-sheet win?
Score audit
What was public, what landed, what changes.
Score 4 also missed.
The missed score was a clean-sheet spread extension.
The low-total support thesis settled correctly.
Carry-forward correction
The next model must build the spread thread before exact scores.
- 01Opponent threshold first
Canada rows against weak or peer opponents cannot be used flatly against Morocco. The model must filter Canada by Morocco-realm opponents before deciding Canada scoring is live.
- 02Spread family before score cell
If Morocco win is the strongest result bucket, Turtle must compare Morocco by 1, Morocco by 2+, draw, and Canada upset before choosing exact scores.
- 03Clean sheet cannot be deleted by one gate
A clean-sheet score inside S10, or a rank-11 extension in the same preserved family, needs a second independent veto before removal.
- 04Postgame does not rewrite pregame
The public record remains a miss. The correction changes the next file, not this result.
Public files for this analysis.