Failed patterns as data
A pattern that breaks the wrong way is information — if you logged the invalidation before the candle closed.
Most journals only keep winners. That trains the eye to remember beautiful textbook shapes and forget how often real markets fake the breakout.
We ask trainees to log every named pattern with three fields: boundary, target hypothesis, and kill level. When the kill level hits first, the entry goes in the failed column with a one-line cause — late entry, ignored volume, wrong degree of swing.
After a month the failed column usually teaches more than the wins. That is the point of pattern training: recognition with discipline, not pattern worship.