hot
1h ATR >= 2% of priceWide enough that the exchange fee is a small share of the stop.
the rule, the reason, the record
Every tag on a card is a one-line rule over numbers the engine archives each cycle. Under each rule is what followed that state on our own labelled book: mean net R after the round-trip fee, hit rate, and how many days it was positive. The losing tags are here too. That is the point.
book: 71,692 labelled candidates over 37 days · rebuilt 2026-08-25T17:24:22Z · numbers withheld below n=100
1h ATR >= 2% of priceWide enough that the exchange fee is a small share of the stop.
1h ATR < 1% of priceThe fee alone is 0.11R+ of any 1-ATR stop here; the book lost -0.23R/trade in this bucket.
funding > +0.03% per 8hLongs are paying to hold. Crowded on the long side.
funding < -0.01% per 8hShorts are paying. Perp trades below index; selling is perp-led.
volume ratio >= 1.8x its baselineParticipation is above normal for this coin.
volume ratio < 0.6x baselineNobody is here. Fills are worse than the chart suggests.
range position > 85% of the recent rangePressing the ceiling.
range position < 15%Pressing the floor.
RSI(1h) > 70 and RSI(4h) > 65Both timeframes stretched the same way.
RSI(1h) < 30 and RSI(4h) < 35Both timeframes stretched the same way.
ADX > 30Directional movement is strong regardless of side.
ADX < 18No trend to ride; breakouts here have failed more often.
|24h change| > 5%Already extended on the day.
Each record is the mean net R (after the round-trip fee) of every labelled candidate that carried the tag, day-blocked. It is a description of what followed that state on this book, not a forecast. Numbers are withheld below n=100. The best composite we ever found netted -0.057R over two years.
See the tags live on the 50 cards, which fired today in the digest, and how the book is labelled in the methodology.