Last-50 tendency rates
These rates explain why the first read chased goals. The postgame analysis says they cannot be consumed without strength adjustment.
Project Turtle / miss postgame analysis / July 3, 2026
This was the painful miss that made Turtle stricter. The raw ledgers made BTTS and high-total branches tempting, but the model did not sufficiently filter Algeria scoring evidence by opponent strength and Switzerland control.
Pregame / postgame analysis
The raw ledger gave a seductive argument: Algeria had scored in 84% of the last 50, Switzerland had conceded in 70%, Algeria BTTS was near half, and Switzerland BTTS was 60%. That made both-team-scoring branches look natural.
The weakness was portability. Turtle treated Algeria scoring frequency as if it transferred directly to this opponent class. It also treated Switzerland concession rate as broad evidence instead of asking whether Switzerland could control a weaker attacking profile in a knockout setting.
The pregame cloud therefore leaned too high: 2-1, 1-2, 2-2, and 3-2. The missing candidate family was Switzerland clean sheet: 1-0, 2-0, and 3-0.
Switzerland won 2-0. That result did not contradict the existence of Algeria's scoring history; it showed that the scoring history was not sufficiently opponent-adjusted.
The new rule is deterministic: before forcing BTTS or removing clean-sheet scores, Turtle must pass a strength gate. The gate checks opponent class, current tournament form, unit quality, and whether the weaker team's scoring record came against comparable opponents.
Evidence figures / last 50 rows
These rates explain why the first read chased goals. The postgame analysis says they cannot be consumed without strength adjustment.
The actual 2-0 branch existed in Algeria's loss family. Turtle failed to elevate it because BTTS evidence was allowed to dominate too early.
A team that scores often against weaker opposition does not automatically make BTTS live against a stronger control-side opponent. Clean-sheet branches need their own gate.