# A2: outsider sells on 0xcc/H3000 native + bars check (10 Oct 2026, #370 #382)
Script it30/outs.py (r6nat.db to 10 Oct, 203 native v4 fee3000 0xcc pools >=20 swaps).
## Outsider sells at >=5x (#370 part 1)
- Operator = top-4 senders (0x88767899 alone 298k swaps). Everyone else bucketed by price multiple vs first price (ETH): buys <1.5x 316, 1.5-5x 221, 5-20x 321, >=20x 88; sells 430 / 218 / 286 / 80. So unlike the 9 Oct WETH H3000 table (outsider sells 0.6 ETH at >=20x), native 0xcc pools show outsider-side sells at height.
- But most of that is mid-size repeat senders (hundreds of swaps: bots/routers). Strictest cut, senders with <=30 swaps in total: 77 distinct senders sold at >=5x, 148 sells, 11.3 ETH (largest single 4.8 ETH, then 2.2 ETH). So non-operator, low-activity senders CAN sell high; sells are not blocked at hook level for them.
- Caveats: "sender" is the PoolManager caller (may be a shared router, not the wallet); token-side sign inferred from amounts; no tx.from. Does not prove retail exit at +600 s (the plateau), only that exits happen at >=5x. Ben's $5 live sell test remains the clean proof; this lowers the odds of a hard sell block on native 0xcc (WETH 0xa80 kit not re-checked).
## #382 bars
- bars.db covers the same 13 Sep-10 Oct tape as r6weth/r6nat: S2-macro rerun not useful (my 15 min result n=26 native/5 WETH stands). Real test needs the #350 long-history backfill; still open at dataeng2. Proxima fee check: still open there.
## USDG/SUSD v4 rule (#370 part 2)
- No new pre-entry rule found; baseline 1.35x, groups not separable (a2-miss-10oct.md). Verdict: PARK (stage 1), not kill: 10/10 above-5M runners there justify one more look when ppdata refreshes the tape with depth.