# Audit: p204 - auditor3, 11 Oct 2026
**Verdict: KILL (not sellable; paper fills are not achievable)**. 12 closed 5x exits (all tp, fill 4.5-4.96x).

1. **Sellability: FAIL.** Free eth_call at latest: transfer of a fresh-address token balance (state override) to the pool reverts with "salg spaerret" for 12/12 tokens. This is the same custom sale lock as the killed d0fd tokens (#847). All 12 pools share code hash 0a493d1af3d0 and launcher 0x73991a25.
2. **Outside sellers: FAIL.** From the swap logs of the 12 pools: sells per pool 61-362, each pool has exactly ONE distinct sell recipient (1/12 each), while buys come from 21-28 recipients. No outside seller ever sold. The "sell side" is one scripted address; the 5x price path is its wash flow plus buys that can never exit.
3. Look-ahead: entry rule (first swap +66 s, Mint ticks +/-204200) uses only past data. OK.
4. Fill rebuild: not done by hand; moot, since the exit cannot execute by a normal holder. Every pool also ends in a pull (pull_t) 0.7-9 h after our paper exit.
5. Replay vs live: no live trades; not applicable.

Cause of the paper 4.7x: bt.py sells through the pool as if the token were freely transferable. For lock tokens that assumption is false. Buy-then-sell proof (Ben $5) is not needed to kill; it would only matter if someone finds that the lock lifts for buyers (d0fd precedent: whitelist set in the deploy tx).
Action: testmgr3 stop p204; analyst2 (idea #47) drop the code-0a493d1af3d0 launcher family, or add a sell-gate filter (fresh-address transfer to pool must pass) before any further test.