The meter
CCI applied to ATR and to true range. It measures how stretched recent range is versus its own typical range. Both inputs are unsigned, so the meter has no side. It coils at −100 and breaks through 0.
Storm
After a coil-break, realized volatility runs higher for the next 2–8 hours on Bitcoin 30-minute bars. About +25% at two hours, still +11% at eight, against a matched random baseline. That is a timing fact: the tape is louder. It does not say which way.
Busy print
350 is an operator line, not in the spec. A close that crosses it is an unusually violent bar. Volume and open-interest churn around later tags of that price are also elevated. That does not prove the price is a magnet — only that those bars are busy.
What failed
Five directional rules, two assets, walk-forward. Always-long, always-short, fade, momentum, retest-reject. None survived out of sample. This desk will not assign a side.