# Market judge v3: entry after actual signal availability

V2 evaluated the first barrier touch from the source-bar close. V3 evaluates the same recorded predictions from the first public Binance USD-M BTCUSDT aggTrade strictly after their original recorded issue time. The subsequent raw trade sequence determines the first touch of symmetric one-percent barriers. A signal is never moved back to the source bar or assigned a retrospectively invented issue time.

This is an execution-price proxy, not evidence of executable spread, fill, commission, slippage or profit. The original source-bar reference may still be inspected separately. Research periods without actual issue timestamps can only have explicitly labeled latency simulations.

## Evidence continuity

Existing signal claims, salts, signatures, witnesses, full packs and V1/V2 evaluation documents remain byte-immutable. New derived V3 judgments are separate signed artifacts and are published only after the original signal is disclosed. The declared canonical chain, completeness rules and missing-opening/late-pack/timestamp penalties are unchanged. V3 recalculations of older signals are labeled retrospective V3, not the originally announced V2 evaluation. Training and private selection logic are outside this change.