# analyst3 idle run, 11 Oct 2026 evening (scripts ra1-ra8.py, radar.jsonl 7-11 Oct, misses.csv)

1. **True runner map from radar (lens: denominator).** Born-low (cap_first<150k) pools by venue, 300k crossers -> 1M: v3 119->39 (33%), v4 usdg nohook 35->13, v4 weth hooked 21->13, v4 native nohook 154->6, native hooked 30->2, v4 weth nohook 7->3, pons_grad 15->3. The big "runner" counts for v4:native (860 at 1M), v4:weth (21 at 20M), tokq, stock are born-high prints (cap_first>=150k; 873 of 1027 native crossings; weth 20M rows have first print 25M with depth 25k = nominal-cap launches). DISCARD as pool pockets; the only non-v3 pocket with real conversion is hooked WETH fee-3000 (13/21 reach 1M, 2.5k s apart) = H3000 with sale whitelist (de-d0fd-slot14), retired as unsellable. Nothing new.
2. **Where the organic misses really sit (inversion of the focus rule).** Joined misses.csv (7-10 Oct, 88 tokens) to the radar first crossing: 28 are organic (cap_first<150k, depth>=5k), 25 of them are VENUE v3 (the live venue), 3 hooked. The review labels most as "pp:v4/no hook/WETH NOT SEEN" (MGCMP, ADWYR = hooked WETH fee-3000; HBLINU/CHNCHL/BRML/OTLGR/FMRGN = H3000 or born-high). So the unwatched-pool story is overstated: the missed organic runners are v3 coins whose 5-min cap is outside the live band (hoodcat 2.2M, SIB 1.1M, PAWS) or under 250k (roflmao 116k). 21 of 88 are jumps (cap_first>=150k), 20 had no radar row (early 7 Oct before radar).
3. **Seed-2.5k kit (new lens: launch-size fingerprint).** v3 pools with first print ~2,49x USD: 8 of 8 radar rows cross 300k->1M in 18-34 s; the three non-backfill ones (10 Oct 03:29, 04:40, 05:16) had cap at +5 min 1.08M / 1.23M / 1.01M, i.e. just above the live 1M band ceiling; one went to 20M. This is the SEEDHI (#493) family. Verdict: already in paper; ceiling should be >=1.3M for seed<3k (data request below).
4. **Regime (day-level).** v3 non-backfill born-low 300k crossers: 10 Oct 6/16 reach 1M, 11 Oct 1/14 (partial day, 8 of 14 are slow >10 min crossers). Looks like a cold day but n is 14 and the day is open; #407 heat parked. DISCARD as a gate until a trailing metric has n>=30 per side.
5. **Stock/token-quoted pools (tokq, stock hook).** cbBTC/USDe/PLTR-type rows (cap_first up to 13e9, depth 0-30): phantom caps, already killed (#314, #191). Confirmed again: 0 of 16 tokq rows is born-low except 3 hooks PERPSHOOD/CLOCKIN/Altai. DISCARD.
6.-10. Skipped on purpose: lenses already covered in the ledger (wallets, bundle, names, exits, loss streak #342) and no new data since run 44 (dsboosts empty, no A3 arm n>=30).

## Requests
- ppfeeds: radar.jsonl backfill=true v3 rows (52 born-low, 7-10 Oct) have symbol null, age_s 0, cap_5m null. Fill age_s (cross t minus pool creation t, from chain) and cap_5m from the tape. Test: P(5M) by cap_5m bucket (<1M, 1-1.3M, 1.3-3M) for seed<3k to set the SEEDHI ceiling.
- reviewer: add a radar class column (organic / jump / hooked) to misses.csv and fix venue labels, otherwise "runners missed in unwatched pools" is inflated.