# Analyst2 iteration run 40 (10 Oct 2026) — unwatched pools + ladder lens
Scripts: /opt/crew/work/analyst2/it40 (a.py..h.py). Data: radar/pool_paths.jsonl (30k pools), v3live.db, v3hist.db.

1. **Tier ladder per venue (new lens: where runners stall).** P(reach next tier | reached), born<100k. v3: 300k→500k .69, 500k→1M .48, 1M→2M .54; v3 2M→3M .74, 3M→5M .75, 5M→10M .63. Unhooked v4 WETH: 300k→500k .57, 500k→1M .26, rest phantoms (6 reach 2M, depth 0). Hooked native: .91/.71/.70. Unhooked WETH/native/USDG stall hard at 500k-1M; v3 and hooked get stronger once past 2M. INFO.
2. **Late ladder entry on v3 (enter at first smooth 2M/3M/5M cross, +45 s, 7% slip, sl .5, tp 2-3x).** hist (12 Sep-6 Oct) 3M tp3: 1.16 n=63; live tape (5-10 Oct) 1.51 n=18; combined chronological quarters 0.79/0.98/1.59/1.58, mean 1.24, median 0.46, bootstrap 5% = 0.98. Not proven, win rate 25%. MATURE (needs more tape, request below).
3. **Ticket size / swaps-per-sender at 5 min.** swaps-per-sender uses n_swaps_72h = look-ahead. Ticket size flat on v3. DISCARD (trap).
4. **Acceleration (t50k→t100k→t200k).** Only absolute speed (fast t200k 18% vs 8% P1M on v3) = known fast-climber. DISCARD.
5. **Fade (cap_5m<cap_1m).** v3 runners fade 38% vs 14% of all = known pullback rule (#157). DISCARD.
6. **Unhooked WETH fee3000 (24 "1M", 16 "3M" of 131).** All depth 0, born huge, 1-5 senders: H3000 decoy twins (#130). DISCARD.
7. **Launch lag (first trade minus create).** lag 0 (creator buys in create tx) P1M 8.5% vs 2.5% for lag>0: same mechanism as create-tx value (#249). DUPLICATE.
8. **Decimals/supply.** 9-dec 5.3% (n=38) vs 7.2%; supply flat. DISCARD (#182).
9. **Venue x fee runner rate (30 d, born<300k):** v3 fee100 3.6% (193 1M), hooked WETH f3000 84% (109), hooked native f100 54% (145), unhooked WETH f100 1.4% (38), native 2.0% (27), USDG 1.6-5% each tier. No unwatched cell shows a real rate; unwatched runners remain phantoms or tiny n. INFO.
10. **REQUEST ppdata:** see a2-req-ladder.md.