PUBLIC CALIBRATION LEDGER

The misses stay visible.

Signals, predictions, avoid calls, and observed fills use separate denominatorsThe eligible claim set used to calculate a hit rate. Different claim types remain separate.. Signal hit rate is not live-trading performance.

The current public snapshot predates policy v2. Its old long/short bucket is withheld until refresh.

Directional predictions
100%
Resolved
1/3
Hits
1
Avoid decisions
100%
Resolved
6/7
Hits
6
Conditional claims
Resolved
0/1
Hits
0
SCORING CONTRACT

Same claim type. Same rule.

signal

A long/short idea without machine-readable fill evidence. It is not trading performance.

prediction

Direction at maturity, with a neutral band around flat returns.

avoid

An avoidance decision scored separately because its success condition differs from entry signals.

paper_trade / live_trade

Shown only when paper or live fills are recorded in machine-readable form.

conditional

Manual review until the trigger itself is explicitly observed.