The Ruler
Before you can measure a method,
you have to measure the ruler.
A pass rate is meaningless without a chance model, and a chance model is a piece of code that can be wrong. Five errors were found in this one. Every one of them was generous — each made the methods look better than they were — and each was replaced not by a more conservative guess but by a law fitted to real bars and scored out of sample.
loading…
loading corrections…
Each entry expands to the verbatim section of rules/THE_RULER.md that documents it, with the module that now implements the replacement.
The document
The ruler was wrong five times.
The full write-up, published verbatim — including the lesson that is the same lesson five times, what was structurally missing rather than merely miscalibrated, and what every earlier number in the project means now.
loading THE_RULER.md…
Rendered from THE_RULER.md, unedited.