# A1-QUIET60-V4S: refinement of QUIET60 (analyst1, 9 Oct, iterate run)
Data: r6weth.db WETH tape 21 Sep-8 Oct, txfrom_598.db (ppdata #91). Sim as in a1-quiet60.md (entry +60 s +3 s lag, impact from depth, pool fee, pull=0, exit needs reserve >= 2x pos, TP capped). Scripts: /opt/crew/work/analyst1/c1-c9.py, d1-d2.py.

## Rule (v4 only)
Pool: v4 WETH-quoted, fee 100, reserve >= 0.5 ETH at entry. In the first 60 s after the first swap: **<= 17 swaps AND >= 2 SELL swaps**. Entry first swap +60 s (or right after the 2nd sell, ~+52 s: same 2.23). TP 3x (2.5-6 all 1.95-2.05), stop irrelevant (0.5), time stop 20 min, size 0.01-0.1 ETH (0.3 ETH 1.80, 1 ETH 1.28).
## Numbers (mean multiple per trade, net of fees/impact, pulls = 0)
- v4 fee100 n<=17 & sells>=2: **n=302, mean 2.21**, halves 2.25/2.13 (2.21/2.20 by count), 17 days all 0.7-2.9 (only day with 3 trades < 1). Pulled-to-zero 19%.
- Same cohort with 0-1 sells: n=182, 1.16. The same filter on v3 (n=50): 1.08 -> v4 only. >17 swaps: 1.15 regardless of sells.
- Delay: +20 s 2.13, +60 s 2.05 (all-venue version), +120 s 1.85, +5 min 1.31 (all-venue). Exit lag 3/6/10 s: 2.05/2.03/2.01.
- Chase filter: price already > 1.475x at +60 s: 1.19 vs 1.8-1.9 below (quiet cohort, mild).
## What it is
Most hits are one scripted launch shape: swap sequence BBBBBBSBBSB(B) (200 of 302), 11-12 swaps, first sell ~37 s, 2nd sell ~52 s, self-matched buy/sell sizes. A bot that fakes two-sided activity, then pumps. Honeypot-free proof (sells go through) is a plausible reason pools with sells beat all-buy pools.
## Outside-money (txfrom, 595 of the 598 pools)
Outside buyer = tx_from seen in <= 3 of the cohort's pools before entry. 309 of 595 pools have ZERO outside buyers pre-entry (all buyers are reused fleet wallets). Their mean: 1.98 (vs 1.36 with >=1 outside). Inside v4 sells>=2: ob0 n=232 2.24, ob>=1 n=70 2.08. So the edge does NOT depend on outsiders being there before us, but it is a fleet pump: we are paid by the pool/LP depth and by outsiders who arrive later, and the rug comes ~10 min in. Real risk = sellability and copy-fleet behaviour, not seen in tick data. Needs a paper test and (Ben's call) a micro-probe.
## Kill conditions
Forward 40 trades: mean < 1.2 net, or > 35% zero-outcomes, or the BBBBBBSBBSB share of pools drops (operator gone).