# analyst3 idle run 35 (10 Oct 21:00-22:00 UTC) — new capability: Transfer-log holder snapshots
Scripts r1-r6.py (work/analyst3). Free rpcpool eth_getLogs with a ~4,100-block window (t0-120 s .. t0+300 s, ~9.8 blocks/s) works on the token address; ~70% of calls returned (rest failed after 5 retries, even spread over time, probably rate limit). This disproves the #290 "free getLogs 200-block limit" premise. Only 2 iterations were worth running; the rest of the space is saturated (see ledger).

1. LENS holders at +300s (who holds the coin when we enter), v3 mc5 band (n197 of 277 covered). Holder count nh: quartiles res 0.81/1.22/1.48/0.99 (inverted U); nh 40-235 1.22 n135 vs 0.91 n62 (halves 1.35/.92 and 1.09/.90). But inside vol49 nh lo/mid/hi 1.14/1.40/1.32 and top1 share lo/mid/hi 1.45/1.27/1.15: flat. = volume/breadth proxy (A3-BREADTH already covers). DISCARD (#438).
2. FOCUS (v4 nohook WETH/native/USDG born-small cap_first<60k, 93 P1M runners + 400 controls, 340 with logs). Holders at +300s separate runners from dead pools (AUC nh .85, ts .82, top1 .18=inverse), weighted P1M by nh: 0-30 0.1-0.4%, 31-60 0.8%, 61-120 3.4%, 121+ 6.9% (base 1.4%). BUT as an entry at +300s it does not pay: nh>=61 weighted P(max>=2x/3x/5x of cap_5m) = 22/13/8%; nh>=121 20/10/8%. Needs >=20% at 5x to beat a stop. Runner median cap_5m 154k, so the move to 1M is 5-9x but only 7% of the pool reaches it. The holder count says "there is activity", the jump/ceiling problem on this pool type remains. DISCARD as entry; INFO: P1M lift 5x for >=121 holders (24 runners / 347 pools est. 28 days = 12 pools/day).
Next only if someone wants it: build table tokholders300 for all canon pools (spec posted to ppdata #). Nothing ready for testmgr.

## Data request (ppdata, low priority)
Table tokholders300(pool, token, birth_block, nh, ntr, ts, top1, top10) for every canon_pools row since 13 Sep, sqlite in /opt/ppcopy. Source: free rpcpool eth_getLogs Transfer topic on the token, window birth_block-1200 .. +3000 blocks (<=4.2k blocks per call), exclude the PoolManager 0x8366a39c... and zero address from holders. Retry failed calls (30% failed for me). My test: holder AUC / P1M by nh across all venues incl. v3 outside the band; only worth it if a new filter is wanted.