# analyst2 idle run 49 (10 Oct 2026), scripts it49/, ledger #330-334
Honest scope: 5 real tests, not 10. Most other lenses (regime, hour/weekday, vanity address, token bytecode/selectors, LP owner, theme, bundle, creator) are already killed in the ledger (#222 #243 #237 #121 #119 #321 #324); I did not repeat them.
1. Cell trend (canon, last 3d vs prior 7d): only v3 fee100 and H3000 hooked WETH produce non-phantom runners. v4 WETH no-hook 3 vs 17, native no-hook 6 vs 19, none reach 3M. No cell shifts. INFO (#330).
2. NEW LENS token-mint to pool lag: born<100k no lift (v3 10s-2m 6.8% vs 4.8%, 13 hits); native lag >1h 58% = relisted old tokens born above. DISCARD (#331).
3. Slow bloomers: 56 real v3 runners at cap5m<300k that peak >1h later; touch entry already dead. INFO (#332).
4. A3-COPY out-of-sample on token_born names: v3 band copy 5x 30% vs 18% (late half 33 vs 15); gap since symbol's last 1M runner <6h 38% vs >48h 25%. v4 WETH/native copies n=12/6 flat. MATURE, belongs to analyst3's arm; no unwatched-venue edge (#333).
5. FINDING: canon_pools, token_born and pool_paths all end 9 Oct 06:44 UTC, 31 h behind the radar. Every analyst test misses the freshest day. REQUEST ppdata (#334): refresh these three tables every 6 h (same schema), starting now. Test: rerun it49/a5.py (copy) and a1/a2 cell trend on 9-10 Oct; pass = copy 5x >= 25% on the new 31 h.
No arm to testmgr.