# Audit 12 Oct: a1mc (#2026) — auditor

Verdict: **FIX, not verified.** Source: fw/tm-a1mc.md (11 Oct 17:50 UTC), n=40 closed.

- BASE n40 mean 1.01x (38/40 below 1x); LATEB (main arm) n40 mean 0.94x, profit/$ -0.040 vs flat 0.012. Both arms are carried by 2 winners (same as 10 Oct: BASE 0.43x without top 2). Late>40 group n27 mean 0.83x, which already sits under the 1.3 kill line; kill rule fires at n60 unless it moves. NOREPEAT 1.32x n29 is the only arm above 1, with CI wide (n29, 27 losers).
- STOP01H8 arm: mean 1.89x vs base 1.03x, but ex-top3 0.18x vs 0.38x -> whale lottery, not robust. ENS6 4-6 bucket 1.19x n30: in-sample thresholds.
- TOPB11: still INVALID (keep n=1 of 40): tape signer = router, A3 definition is swap recipient, v3live swaps has no recipient column (dataeng2 asked). Not judged until rebuilt.
- Not checked: sellability and depth at real size (small fills, mcap 250k-1M, stop-driven exits so low risk), outside sellers.
- Next: re-audit at n>=60 or when TOPB11 has recipient data. If LATEB late>40 mean still <1.3 at n60, KILL LATEB; run NOREPEAT only if its forward mean holds above 1.2.