# Analyst1 iteration run 43 (11 Oct 2026) - scripts work/analyst1/m1/ (v1,v2,s1-s15)

Focus: v4 no-hook pools (WETH/native), the biggest unwatched runner cell. Data: r6weth/r6nat swap tapes (2607+1366 nohook v4 pools with >=5 swaps, 21 Sep-9 Oct), canon, creators.db.

1. Token address vanity (zeros, repeated chars): no lift (ledger #410). DISCARD.
2. Token mint -> first swap gap: no monotone power, v3 not covered (#411). DISCARD.
3. **Gradual cap-cross entry on v4 nohook, pull-aware (#412).** Price-only sim looks good (cap crossing 150k from >=120k prior print, age<15 min, entry +15..60 s, stop .5 tp3, 4% cost: mean 1.5-1.66 n=677, stable on 19 days, ex top2 same, 630 different creators). But 55% of entries see a >=50% LP removal before tp or stop; with those booked at 0.1 (0.4) the mean is .96 (1.19); tp2 1.06-1.16, tp1.3-1.5 ~1.0-1.08. All 677 are the fee-100 kit-template pools (first cap ~27k, same size, same reserve) = BUSY60 / FAST5 family. 1290 of 1291 template pools with >=100 swaps get pulled: median 15 min after birth, at a median 3.9x first cap (p90 12x). No feature separates pulled from tp pools (age, swaps, ramp speed, first cap, reserve). DISCARD as new test; it confirms the family is a scripted ramp-and-pull and the open number is what a pull leaves to sell.
4. Aged slow climbers (#413): of 171 tape pools that reach 1M, 76 smooth <1h, 70 jump <1h, 21 jump >=1h, only 4 smooth >=1h. Lane not tradable. DISCARD.
5. Kit concurrency regime (#414): with a +-30 min window mean falls 1.36 -> .84 (pull% 41 -> 63), but the window includes future births (look-ahead); trailing-only is flat. DISCARD (trap logged).
6. Request ppdata (below).

Request to ppdata: table liq_txfrom (/opt/sw/ppdata/liq_txfrom.db): for every ModifyLiquidity event (kind 10 add, kind 12/negative remove) in r6weth/r6nat for v4 pools with hooks=0, fee 100, from 13 Sep to now, refreshed daily: pool, block, li, kind, liq_delta, tx hash (full), tx.from, tx.to. Why: r6 'sender' on these events is the position manager (0x4e34, 0x58da, 0x9982), tx15 is a truncated hash, so the real LP owner who pulls is unknown. Test: group the 1290 pulls by tx.from; if a handful of addresses make most pulls, build per-adder hazard (pull age distribution by adder from earlier pools) and test an exit-before-typical-pull-age rule and a skip rule on pools whose adder has pulled >N earlier pools, on the 677 entries above (needs forward OOS: split by date).