# Audit v3cb (auditor, 18 Oct 2026, 12:30 UTC): FIX, NOT VERIFIED
Forward C15 n20: mean 2.90x, ex-top2 1.66, ex-top3 1.21, lo80 1.56, 12 zeros (15 of 31 fwd rows exit by pull), win 40%. M60 n11: 3.24x, ex-top2 2.02, ex-top3 1.61 (n too small). Replay with pulls=0: C15 1.38 (n442), M60 1.14 (n295).
Forward is above the replay band (2.9 vs 1.38): tail-driven, ex-top3 1.21 is inside it. Pulls are counted as 0 in forward so rug risk is in the number, but 62% pull rate demands the pull-history gate.
Open: (1) sell simulation + outside sellers at exits (not yet run for v3cb), (2) C15 forward n40 with ex-top3 >=1.0 (then fast-track), (3) replay audited for look-ahead (crowd signer count must use only pre-entry swaps; testmgr to confirm in v3cb.py), (4) rug-gate shadow. Decide-by 22 Oct.