What we are testing next
Whether higher grades come before better outcomes. Closed bars only, rules frozen before the data is read, and no results yet.
Read the note: What we are testing nextResearch log
The research log.Not the claim. The test.
Here is me measuring whether my checklist means anything: how each signal is checked, how a test is run, and what comes out. Written up as it happens.
Posts, newest first. Calculation checks and testing notes.
The indicator we would not copy
Its source is hidden, and we do not copy hidden code by fitting to its output. So we stopped, and built from the public formula instead.
Read the post: The indicator we would not copyA live bar has no correct value
Until a candle closes, an indicator's reading keeps moving. What that means for anyone trading off one mid-candle.
Read the post: A live bar has no correct valueA paid indicator, matched to four decimal places — and the one bar that was not
Rebuilt from the public formula and compared bar by bar: every closed bar agreed. The one that did not was still forming.
Read the post: A paid indicator, matched to four decimal placesHow a test is run
Rules first, data second.
Every test is written down before it runs, so the answer cannot be bent to fit.
-
Closed bars only
A live bar has no single correct value, so it is never used.
-
Rules frozen first
What counts, and how it is measured, is fixed before the data.
-
Unseen data last
Each rule is checked last on months it was never shaped on.
-
Every result written up
When a test says no, that is published too.