# v3cb sell-depth + outside sellers at exits (analyst2, 18 Oct 2026) - #3216 open item (1)
Script vc1.py (reads /opt/sw/v3live.db + /root/tm/v3cb/trades.csv). Covers every non-pull C15/M60 trade >=2x: 168 rows (156 replay, 12 fwd; 99 C15 + 69 M60 arms). Exit replayed with the lane's rule (0.6x stop / 50% trail), $10 position sized at exit value, +-120 s window.
- Pool ETH at exit swap: median 12 ETH (range 1.9-36; 3 rows 0, see below). Exit position $21-221 (~0.01-0.07 ETH): price impact median 0.1%, max 1.8% (0x87bb1c8c, 130 usd on 2.9 ETH), only that one near the modelled 2%.
- Outside (non-creator) sellers within +-120 s: median 42 distinct wallets, min 7 (0x22314bdb, 0.73 ETH sold, thin). 0 rows under 5 wallets; median sells 7 ETH = 100x our size. Public v3 pools, no hook: strangers sell at these prices.
- Flag: 3 of 168 (0x691b8b4f 3.09x, 0x2d309efd 2.72x, 0x371239eb 2.37x) show pool_eth_after 0 at exit = drained at/after the exit swap, likely optimistic. Impact on C15 mean ~ -0.02x (8.2x of ~640x total over n463). Forward rows: all 12 have pool 10-27 ETH, 21-48 outside sellers.
Verdict on (1): PASSES. Sell depth is not the weakness; same as v3vearly. Real weak spots stay as the auditor wrote: tail concentration (ex-top3 1.21) and 62% pulls -> items (2) n40 and (4) rug-gate shadow.
Item (3) look-ahead check: v3cb.py breadth() counts signers/buy ETH only from swaps with (b,li) <= trigger swap, entry is the first swap >=15 s after the trigger: no look-ahead. The trigger swap itself is included (known at trigger time). Pull test before trigger flags only pre-trigger removes. Clean.