# Audit v2q (auditor2, 14 Oct 2026) - UNVERIFIED, keep running, re-audit at arm A n>=30 or G n>=15
Forward (trades.csv, 24 rows): arm A n21 mean 1.48x (SE 0.38, 80% lower ~1.0), median 0.71x, 13/21 <1x. Without the top trade 1.31x. Four tp5x wins (4.6-4.8x) carry it.
LP pulls: 12/21 (57%) of A exits, 6 trades <=0.06x. Replay had 3 zeros in 26 (12%) and mean 3.59x -> replay is 2.4x above forward and badly underestimates pulls: replay is not a valid band; it does not support VERIFIED.
Arm G (sells_pre>=3): n3 (2 tp5x, 1 pull) mean 3.20x - no evidence at n3; the gate cannot be judged.
Checks: sellability/tax NOT done (my UR sell sim reverts empty on WETH-quote pools, known tool limit; token sell tax unverified) ; outside sellers not tested; look-ahead: entry is 2nd Sync after trigger, lag 1-8s, no look-ahead seen; hand rebuild: not done.
Kill rule not hit (A mean 1.48 >= 1.0). Fast-track rule (mean>=1.3 and lower80>=1.0) borderline: lower80 ~1.0 at n21, not met cleanly.
Verdict: NOT VERIFIED. Open: (1) G n15 or A n>=40 with lower80>=1.0; (2) a real/ forked buy->sell on a fresh 0x8bce pair (post to ben only if exec cannot fork-sim); (3) rebuild replay with real pull rate (~57%).
