# analyst2 idle run, 11 Oct 2026 (scripts work/analyst2/it59/a-m.py; data pool_paths.jsonl 23k born-small pools 13 Sep-9 Oct, nm2 names, prices.db)
Organic runner = max_cap>=1M, depth_usd_5m>=5k, swaps72h>=30. Entry sims on v3hist (stop .5, tp 5x, 4 h).
1. ETH/BTC regime (ETH 1h/6h, BTC 6h/24h return before birth) vs P(1M) per cell: flat in every cell and both halves (v3 2.8/2.9/3.1% terciles). DISCARD. Ben's "15 losses while BTC dumped" is not visible as a BTC effect on hit rate.
2. Sparse launches ("lone launch": <=3 same-cell pools born in prior 60 min) vs P(organic 1M): v3 5.5% vs 2.1% (halves 3.8/7.9 vs 1.6/3.2), hooked WETH 11.3% vs 1.5% (8.6/12.0 vs 1.4/4.4), matched within day 16-28 days, 10.7 vs 3.0. REAL but entirely a pre-entry effect: survivors at +5 min (cap5m 100-500k) show no lift (v3 29.6% vs 38.5% hit3x), and entry sims at +66 s / +300 s on v3 are flat (1.88 vs 2.02 / 1.22 vs 1.52). Attention-per-pool is priced in by the 5-min cap. PARK (info).
3. Sympathy names (new, lens narrative spillover): pool shares a rare word (doc freq <=15) with a runner (peak >= 2M) born in prior 48 h and already peaked: P1M v3 13.9% vs 2.7% (n173, halves 7/21), hooked WETH 11.9% vs 4.2% (n59), v4 WETH nohook 4.7% vs 1.9%. Entry +66 s v3 sim: mean 1.78 vs 1.97 (n109), hit5 18% vs 20%. >=20x 11% vs 6.4% (12 hits). Known at birth, but no return lift. PARK (A3-COPY #256/#333 is the same signal).
4. Supply/decimals fingerprint (new): non-1e27 supply is a kit marker. v3 0.9% vs 3.1% P1M, native nohook 0.5% vs 2.5% (both halves): skip filter at best (base rates 1-3%). v4 WETH nohook fee 3000 supply 1e29 hooks 0: 16 of 21 "runners" with max caps up to 700M are full-range dust pools (L~1e18, price swings 1e33-1e48, swaps of 1e-5 ETH by two bots). Same as #61. 93 of 250 v4 WETH nohook 1M "runners" in pool_paths are dust (full-range mint L<3e18): DISCARD, but the runner map should drop them.
5. Hot-hand spillover (organic 3M runner peaked in the prior 60 min) vs new pools P(organic 1M): v3 3.2 vs 2.1 (H2 flat 3.5 vs 3.6), WETH nohook 1.3 vs 1.2, hooked WETH worse. DISCARD.
Nothing for testmgr. Universe saturated: stopped at 5 tests (no honest 6-10).
Data request (ppdata): pool_paths with true crossing times (t300k_s, t1m_s, t3m_s, depth at cross) and a refresh (stale 38 h). Reason: iterations 3 and 5 approximate runner peak time as t_first+max_age_s.