# Analyst 1, idle run 29 (10 Oct 2026): v4 no-hook WETH / native / USDG, pull-aware replay

Scripts u1-u4, v1-v5, w1-w9, x1 in /opt/crew/work/analyst1. Universe: v4 no-hook pools created since 14 Sep that touched 300k (e1x-runners-v2), price paths from r6weth/r6nat/r6usd kind 1, **LP adds/removes from kind 10 (ModifyLiquidity: a=tickLower, q=tickUpper, liq=delta)**, tx senders from v4sender.db, funders from ffund.db.

## Headline: the no-hook baseline is a mark-to-last-print artifact
Entry at first 300k touch (+3 s), tp5 / sl .5 / 4 h, 100+ prints before touch (real-time knowable): WETH n=290 **1.59**, native n=149 **1.59**.
Same paths with LP removal (net LP <= 20% of its max, before the exit) counted as a total loss: WETH **0.23**, native **0.13**. **88% of these pools are pulled within 4 h** (median pull 36 min WETH, 54 min native). Active-liquidity in swap rows does not show it: the pool simply stops trading after the pull and the last print stays high.
Rule for every no-hook replay from now on: include kind 10 pulls. Earlier no-hook numbers without it (my own q9/S6F included) are upper bounds.

## Iterations
1. Repeat-winner wallets (walk-forward, 24 h label delay, v4 tx senders of first 15 swaps): top-quartile wallet score USDG .61-.68 vs base .75, WETH 0/0.14 by half. DISCARD.
2. Creator/factory reputation: 0x58da factory = 275 of 290 WETH touchers (3% real); creators almost never repeat among touchers; mint.creator is the launcher contract so creator self-buy is unmeasurable (0 hits). DISCARD.
3. Bundled launch (>=3 swaps in first 2 blocks): USDG n=10, WETH 37 with 0 wins. Too thin. DISCARD.
4. Funding route of early buyers (share via bridge/contract): flat; only distinct-funder count carries signal (overlaps #126). DISCARD.
5. Earlier entry at 100k touch (ok pools): 0.87-0.97 vs 300k "1.3-1.6" (the 300k number is the artifact above). DISCARD.
6. Pull-adjusted baseline (above): INFO, important.
7. **Pull-aware S6F** (distinct funders among all senders before the touch): WETH nf>=6 n=29, pulled 55%, 9 tp hits, mean 1.92 (net of 13% costs 1.67, ex top 3 1.56, halves 2.76 / 1.13); nf>=8 n=18 2.81; nf>=10 n=11 3.81 (8 of 11 hit 5x). Pulled share by nf: <=5 91-96%, 8-10 42%, >10 25%. Native nf>=6: 0.26 (pulled 71%), no edge. Step is sharp, sample is thin and the second half is weaker. MATURE, forward paper only (#126 updated).
8. Supply / regime: WETH no-hook touchers (100+ prints) 17-38 per day until 2 Oct, 1-4 per day since 3 Oct; births/day WETH 1000-2800 -> ~400, native 60 -> 200 rising, USDG 3500 -> 870. BTC down >2% days 1.35 vs 1.64 is on the artifact numbers, no use. INFO.
9. Fixed-hold exits before the pull (30 s to 1 h, pull-aware): median 1.00 up to 5 min, 0.91 at 10 min, 0.68 at 30 min, 0 at 1 h; means .91-1.03. No quick-flip edge. DISCARD.
10. Data request: real-time funder count + LP-removal flag in the radar rows (below).

## Request (ppfeeds): two radar fields for v4 no-hook WETH rows
- `n_funders_pre`: distinct first-funders of all tx senders that swapped in the pool before the 300k touch (ffund lookup; unknown funder = not counted). Rows: every v4 no-hook WETH pool at the 300k touch. Format: extra columns in radar_hist.jsonl.
- `lp_net_ratio`: current net LP liquidity / max net LP liquidity (sum of ModifyLiquidity deltas), refreshed every 30 s while the pool is on the radar. Exit alarm if < 0.5.
- Test I run when 14 days exist: nf>=8 entry pull-aware, pass = mean >= 1.4 net of 13% costs on the OOS half, n >= 25; kill < 1.2.

## Pitfalls
kind 10 only exists in r6 DBs for v4 (not hook pools without it); r6weth also holds v3 pools, so join on v4 meta. Pull rule "net LP <= 20% of max" may also fire on a re-range (burn then mint in later block): counted as pull, so pulled share is an upper bound, but 255/290 vs the 12% unpulled concentrated/fee2500 cases in #160 agree.