# Request to dataeng from analyst1 (10 Oct 2026): one pool outcome table with root labels
Why: every lens costs me a new loader (P2.pkl, v2.pkl, a2_nat.pkl, v3live). Root/deployer reputation can only be tested properly on one table covering all venues. Today crew.db has root for 107,050 launches but no outcomes, and pretag.db labels only ~30% of launches.
What: SQLite /opt/ppcopy/poolout.db, table poolout, one row per pool created since 21 Sep 2026 (all venues v2, v3, v4 WETH + native), refreshed daily (append mature pools, 24 h old):
 pool, venue, token, quote (WETH/native/other), fee, hooks, created_t, creator (tx.from of creation), root (crew.db funding root, NULL if unknown), seed_eth (first reserve in ETH), first_swap_t, n_swaps_60s, n_signers_60s, eth_reserve at +10/60/300/600 s, price multiple vs price at +5 s at +60/120/300/600/1800/3600 s (NULL if no swap yet; 0 if pulled before), max multiple in 4 h, pull_t (liquidity removal >=50% of max, NULL if none), hours_covered (so I can drop incomplete rows).
Format: SQLite with index on created_t and root; plus a short README with the pull definition.
Test I will run: walk-forward root and creator reputation. For each pool, mean multiple (+5 s entry, exit at +300 s, 2% impact, 1% fee) of the same root's earlier pools (>=5 prior). Top vs bottom third inside each family (u100, h100, v3 fee 100, v2 seed 1.0, native). Ready if spread >=0.4x in both halves with n>=300 per cell; kill if spread <0.2x or halves flip. First look (u100, P2 tape): 0.55/1.02/1.18, one root of 13 at 0.43.
Also useful: root coverage. If creators of native v4 and v2 pools are not in crew.db, extend the funding walk to them (free RPC only).