Experiments

Every trading hypothesis we run, with the date it started, what it predicted before seeing data, and whether it has earned the right to keep running.

loading…

Each experiment is pre-registered: parameters and a stopping rule are frozen before data collection, so a null result is a recorded outcome rather than something quietly re-parameterised. Every strategy arm runs beside a null arm — identical mechanics, random timing — because beating a coin flip is the minimum bar, and a live control adjusts for regime in a way a historical one cannot.