# #762 / #763 (analyst2, 10 Oct 2026) scripts /opt/crew/work/analyst2/it46/

## #763 H3000 outsiders with tx.from (swapfrom.db grp h3000, 6,626 non-operator swaps, 1,840 wallets, 129 pools)
- Buys by price multiple vs first price: <1.5x 587, 1.5-5x 3,061, 5-20x 1,780. Sells: <1.5x 619, 1.5-5x 579, **>=5x: 0**.
- 1,168 (pool, wallet) pairs bought at >=5x. **None of them ever sold** (0 later sell swaps). 574 wallets bought >=5x (19 ETH).
- Reading: confirms and tightens fw/a2-h3000-outsiders.md. No outside wallet exits above 1.5x at all. Either sells are blocked for non-operators or the pool is pulled before anyone can sell. Cannot tell which from the tape.
- Verdict: A2-H3000 stays gated by Ben's sell test (eth_call / $5 sell). No live money before that. Paper numbers for H3000 exits above 5x should be treated as unproven.
- Not done here: H100 skim check (needs ERC-20 Transfer logs per tx, dataeng says not started); w120 wallet lens is covered by a2-wlink-oos-10oct.md.

## #762 lpowner feed
- Live feed starts now, history only in /opt/ppcopy/lpowner.db (4,377 pools, 1,768 owners; top owner 338 pools).
- Owner-repeat test: INVALID so far. Only 388 pools join with e1-runners.csv (that file holds only pools with peak >= $300k), so hit rate by prior history is survivorship-biased (prior>=10 shows 0.97, meaningless). Need peak for ALL lpowner pools (dataeng: peak cap per pool for lpowner.db, incl. non-runners) to test "owner with a prior winner -> this pool". Requested below.