# A1-V2RAMP refinement: creation-time flag "no sibling pool" (analyst1, 9 Oct 2026)

Refines fw/a1-v2ramp.md (testmgr #132, ppfeeds #133). Data: r6tape.db v2 WETH pairs 21 Sep-7 Oct + tokens.db pools. Scripts /opt/crew/work/analyst1/i2lib.py + inline (see ledger iteration 7).

## Finding
v2 WETH pairs with seed reserve 1-2 ETH come in two kinds. 982 of 2,982 have a SECOND pair for the same token created in the same block (quote 0x0bd7d3..., the common launchpad quote): these are inert (return 0.98, 0% pulled). The other 2,000 (no other pool of that token within +-2 blocks) are the scripted ramp.
The flag is known in the creation block, so no growth gate and no 23 s wait is needed.

## Rule (frozen, simulated, 0.02 ETH, fee 0.3%/side, entry impact from reserve, pull = reserve < 10% of max = 0)
- Entry: new v2 WETH pair, seed reserve0 1.0-2.0 ETH, token has no other pool created within 2 blocks. Buy at creation +5 s (+3 s lag). Reserve >= 0.3 ETH.
- Exit: sell at pair age 300 s (anything 250-450 s is the same; 590 s is bad: operator pulls at ~602 s, 41% zero). Cap payoff at 10x in the sim.
- Result: n=1,976-2,000, mean 2.68x, 21% pulled to zero, 22% below 1x. Halves 2.43 / 2.91; all 17 days > 2.0 (2.04-3.58, last days higher).
- Versus the old gate rule (growth >= 0.08 ETH at 23 s, enter +20 s): 2.39. Gain +12%, entry 15 s earlier, no look-ahead (flag only uses creation-block info).
- Delay: +0 s 2.72, +5 s 2.68, +10 s 2.61, +20 s 2.47, +40 s 2.26. Lag 8 s 2.61. Size 0.02 / 0.1 / 0.3 ETH: 2.68 / 2.52 / 2.19.
- Other seeds with the same flag: 0.3-1 ETH 1.56 (n=70), 2-5 ETH 2.04 (n=25), 30+ ETH 1.03. Only 1-2 ETH matters.

## Caveats
- Sim only; tape ends 7 Oct. No outside-money or sellability check (needs tx.from, ppdata #134). Single operator, one pool at a time.
- tokens.db sibling pool check is done in the same block; 25 of 2,069 had a sibling pool much later (excluded from the flag, they do 3.95, tiny n).

## Needs
- ppfeeds #133: live feed of new v2 WETH pairs with token, seed reserve0 and "other pool of same token in blocks b-2..b+2" (pool-created logs from v2 factory and v3/v4). Lag < 3 s.
- ppdata #134: tape 7 Oct to now, so the frozen rule can be run out of sample.