# A2-H3000: v4 fee-3000 per-coin-hook WETH pools (ramp pools) - 9 Oct 2026, analyst2

Data: ppdata r6weth.db (board #80), pools created 21 Sep-8 Oct. Scripts: /opt/crew/work/analyst2/w1-w12.py (research-box).

## Method (fixes the mirage from the stable tape)
Price from sqrtPriceX96, quote reserve = cumulative net WETH into the pool, LP-pull time from v4 liq events (cum liq <= 10% of peak = dead; v3: first burn). Entry first swap +60 s at >= 0.4 WETH reserve; position 0.04 WETH; exit fill = next swap after trigger +10 s, needs reserve >= 2x position*mult, else half value; pulled before exit = 0. Pool fee + 3% slippage each side. Hold 4 h, stop 0.5.

## Dumb baseline (whole WETH universe, 3,240 pools passing entry gate, no picking)
tp5/sl.5: mean 1.31 (CI 1.24-1.38), median 0, 27% win; 59% of entries are pulled to zero. Hold-only 0.86. Weekly stable. (Universe is mildly positive under TP; consistent with earlier bt.py baselines 1.2-1.8.)

## Segment found: venue v4, fee 3000, non-zero hook (one hook contract per coin)
- n=110 pools (136 exist; 119 traded), 21 Sep-8 Oct, every day 3-10 pools. tp5/sl.5: mean 4.47x (bootstrap CI 4.31-4.63), 96% reach 5x, 2 of 110 zero. Halves 4.2/4.6. Trim top 5: 4.35.
- tp3 2.77, tp10 8.87 (95% reach 10x), tp10 stays at 8.4+ CI. Entry delay 120/180/300 s: 4.49/4.46/4.36. Exit latency 60 s: 4.65. Slippage 10%/side: 3.89. Position 0.2 WETH: 4.45 (n=53, reserve-gated), 0.5 WETH slip 6%: 4.09 (n=15).
- Time from entry to 5x: median 634 s (q1 514, q3 790). Reserve at the 5x moment median 20 WETH. LP pulled median ~10 h after creation (q1 6 h).
- Who is on the other side: 3 contracts (0xc955efd6..., 0xe0395226..., 0x7680ecc6...) do 97% of 1.19M swaps (10k per pool): the ramp is their buying. Outsiders: 196 senders, 32k swaps; 96% of 924 outsider wallet-pool pairs end net negative (-122 WETH total, mostly 0x092c2f7f... -71), because they hold bags when the LP is pulled. Outsiders that sold early did fill at multiples (0x05007022..., 0x88767899... net +3.0 / +8.5 WETH; sells in the first 4 h priced 13-47x the +60 s price). So selling into the ramp works for non-operator senders.
- Kill if: sells from a fresh wallet revert (hook allowlist / honeypot); forward mean under 2.0x over 30 trades; ramp stops appearing for 3 days.

## Open risks (must be checked before any real money)
1. SELLABILITY: tape shows outsider sells, but a hook can block specific wallets. Needs eth_call simulation of a sell from a fresh address on a live pool of this kind, or a $5 live test (Ben only).
2. Tape ends 8 Oct 11:24 UTC. Is the ramp still running? Asked ppdata for refresh.
3. Detectability at +60 s: rule is fee==3000, hooks != 0, venue v4, WETH quote, hook is a fresh per-coin contract. Needs live feed to confirm the pool init is visible at launch (ppfeeds).
4. Paper-test fills must use the reserve gate, not tick prices.

## Rule for testmgr (paper)
Pool init v4, quote WETH, fee 3000, hooks != 0. Buy 0.04 WETH at first swap +60 s if reserve >= 0.4 WETH. Arm A: sell all at 5x. Arm B: half at 5x, half at 10x. Stop 0.5. Time stop 4 h. Skip if reserve < 2x position*mult at exit (record as stuck).