# A1-V2DEC9: v2 seed-1 pairs whose token reserve is 1e18 raw (9-decimals, 1B supply "classic" meme kit)
analyst1, 9 Oct 2026. Scripts r9.py (+ i8.py loader, v2.pkl). Data: r6tape.db v2 syncs 21 Sep-7 Oct (v2.pkl), v2feed live (pairs.jsonl, v2tape.jsonl).

## Finding
Inside the V2RAMP seed-1 family (first reserve0 exactly 1.000 ETH, factory 0x8bceaa40...) a small sub-family has first token reserve = 1.0e18 raw
(token has 9 decimals, supply 1e9: e.g. Hachiko Inu / HACHI). 64 pairs in 17 days (~4/day, every day but two), all from the same v2 factory.
- Entry +5 s (+3 s lag, first sync after), exit at pair age 350 s, constant-product fill both ways incl. 0.3 % fee both sides, pull (reserve < 10 % of max) = 0.
- Mean return per $1 (x): pos 0.02 ETH 35.1, 0.1 ETH 22.3 (median 16.5, p10 1.45), 0.25 ETH 13.4, 0.5 ETH 8.1, 1 ETH 4.6. Zero (pulled) 3 %. Top 3 trades = 15 % of the sum (not a jump: ETH reserve grows 1 -> median 5.4 -> p90 11 ETH by 350 s with ~160 buys and ~57 sells; sell/buy ETH 0.59, only 8 % of pools have zero sells => sellable).
- Halves (time order) 14.8 / 29.8 at 0.1 ETH. Rest of seed-1 family (normal token seed): cap10 2.43 (the known V2RAMP arm). So this is a conviction arm inside V2RAMP, not a new rule family.
- Entry delay matters hard: +5 s 22.3, +30 s 19.1, +60 s 10.4, +120 s 4.9 (0.1 ETH). Need the trigger at PairCreated, same as V2RAMP.
- Exit age: 120 s 7.5, 200 s 12.6, 350 s 22.3 (zero 3 %), 500 s 29.3 but 8 % pulled and p10 0.02, 800 s 38 with 17 % pulled. Pulls start ~400-500 s. Keep 350 s (V2RAMP exit age 300-350).
- Forward check on live v2feed (1 day): 2 pairs with token reserve < 2e18 and seed 1 ETH; one has tape: reserve 1 -> 7.8 ETH at 113 s, 11.7 at 193 s, 11.2 at 274 s, 13.2 at 382 s (same pattern). n=1, not evidence, only consistency.

## Caveats
- n=64 in-sample, one factory. Token-seed split was one 2-bin cut in a 5-bin scan, not a deep search; the 64 are 3 % of seed-1 pairs.
- Sim is on Sync reserves (no tx.from / recipients): cannot test outside money; whoever buys 10 ETH into a 1 ETH pool is real net ETH, but the buyers could be one fleet. Real position sizes above 0.5 ETH lose most of the edge (impact), so this is a $-small, high-multiple lane.
- Sellability: sells exist in 92 % of pools, but a honeypot-style token that allows only some sellers is not excluded; paper test with real eth_call sells is the check.

## Rule for paper (testmgr)
v2 pair created by factory 0x8bceaa40b9acdfaedf85adf4ff01f5ad6517937f, WETH side first reserve exactly 1.000 ETH, token reserve in [0.5e18, 1.5e18] raw, sibling=false. Buy at first sync after PairCreated (+<=5 s), 0.02-0.1 ETH. Sell everything at pair age 350 s (or earlier if WETH reserve < 50 % of its max). Log fill price, WETH reserve at entry and exit, and whether the sell succeeded.