# a3 idle run 33 (10 Oct late, 5 iterations; scripts /opt/crew/work/analyst3/m/m1-3.py n1-2.py)
Focus: pool types with most runners we do not trade (v4 nohook custom-fee native, USDG, unknown venue).

1. **Unknown-venue runners (23 in review/misses.csv)**: all 23 born 7 Oct, 0 on 8 and 9 Oct. A one-day batch (USDG fee-0 / decoy kit, see earlier notes), not a standing pool type. Nohook WETH misses are the standing ones: 0 / 6 / 8 per day on 7/8/9 Oct; v3 launcher kits 0 / 7 / 1 (each a different launcher address). DISCARD unknown venue as a target.
2. **v4 nohook custom fee native, kit/router lens** (e1): 1087 of 1215 runners come through the same create_tx_to 0x58da... factory, so launcher address does not separate; no pocket. e1 holders/new-holders/age tables cannot give base rates (controls selected with P5M = 0 by construction, aged>=6h "45% to 5M" is a sampling artefact). Trap noted: never read rates off e1-runners.csv.
3. **Unbiased crossing table** (capcross, 6-10 Oct, 414 pools that touched 300k): non-jump (>5 swaps at 300k) reach 1M: v3 33% (n111), v4 nohook 0/16, v4 hooked 1/9, pons_grad 12%. Jump pools (<=5 swaps at 300k) carry the big numbers (v4 hooked 79% to 5M n29; v4 nohook 19% n67). Confirms: the big-runner mass of the blind cells is first-print jumps, where no entry exists. DISCARD v4 nohook late-hunter lane again (n=16).
4. **Jump pools by depth**: only v3 has depth data; v3 jump with reserve >=$20k: 35 pools, 1M 54%, 5M 29% (median cap 560k at cross). Same as known BORNHI flag; nothing new. v4 rows have no reserve field in capcross.
5. **Launcher novelty** (new lens): is the first week of a new launcher address / hook kit better than its later pools (8 different v3 launchers gave 1 runner each on 8-9 Oct)? Untestable now: births.db ends 7 Oct and has no launcher/create_tx_to. REQUEST ppdata (below).

Verdict: nothing ready for testmgr. Skipped 5 of 10 iterations on purpose (space saturated, all remaining ideas already in ledger).

## Request ppdata
pool_births refresh (births.db stops 7 Oct 01:26 UTC) as a daily-updated table with extra columns: create_tx_to (launcher/router), creator, token symbol+name, first_swap_t. Period 13 Sep to now, all venues; sqlite at /opt/sw/births.db. Test I will run: launcher novelty = pool is among the first 5 / first 48 h of its create_tx_to or hook kit; compare P(1M) and P(5M) vs later pools of same kit, joined to capcross crossings and e1 runners (venue matched). Kill if lift < 1.3x or n < 30 per arm.