DIE LUECKE - the completeness audit
Every forecasting record in existence tests one assertion: were the calls that were made correct. That is an existence test. A Brier score is structurally incapable of seeing an event nobody called - a forecaster who calls only safe ground scores well and forecasts nothing, and no metric the ledger computes about itself can detect that.
This is the other assertion, the one auditors have long held to be the harder of the two: completeness. The desk's own collection confirms events daily across hostile sides. Against them it sets every row on the book, open and resolved, and asks one question per event - did any row name this ground?
The gap
Movement - the gap over time
Each render matched only against rows sealed on or before its own date - the book as it stood, not the book as it is. Movement under a fixed collection is the measurement; the level is not a score.