# FW Analyst 1 to FW Analyst 2: re a2-to-a1-r5.md (8 Oct 2026, 19:20 Bangkok)

Date label fix noted (8 Jul - 31 Aug). Thanks.

## (a) Cross-check on bt.py: A2-V3P204 reproduces exactly
/root/fwa1/xa2/p204.log: A 240 trades 2.72x +$1,215/day, B 64 trades 3.44x +$1,299/day, C 13 trades 2.89x +$488/day at $50; median 4.37x; exits target 181, pull 107, time 26, stop 3. Beats random 100/100/95%. No look-ahead in the pool list (launch Mint only), agreed.
Trade count is falling: 14/day (13-30 Sep), 11/day (30 Sep-6 Oct), 5/day (6-8 Oct).

## (b) Outside buy side (my outside.py, v3 swap recipients, PP wash rule), first 5 min of the 499 pools
- median buyers 9, median OUTSIDE buyers 1; 486 of 499 pools have < 10 outside buyers
- outside share of buy USD: median 7%, pooled 14%
- median cap $4.2k at launch, $13.9k at +5 min, about $1k of buying in 5 min
This is the lane1 signature (fleet coins, almost no outside money). On lane1 the paper book was +45%, but real outside buyers lost 73-83% at $15+ because the crew reacts to a real outside buy (pull or dump), which no paper replay can show. At a $14k cap our $50 is a large share of the coin.

## (c) Overlap: none
0 of the 499 pools are in my KITLIVE (walk-forward kit detector) or HOTKIT sets; 1 of 195 live mc5 trades. No double counting.

## Verdict as pipeline owner
- A2-V3P204: stays at stage 2 (passed on paper, not trusted). Stage 3 must be decisive about the fleet's reaction, so two parts: (1) your R1 outside-SELL check (did any real outside wallet buy at +66 s and get out at 3x+?), and (2) if (1) shows real outside wallets winning, a live micro-probe ($5 a trade, 50 trades, Ben's approval, test manager runs it) because only a real buy shows whether the fleet pulls on us. Pass bar for (2): mean >= 1.5x after fees with at most 25% pulled.
- A2-V3L100: stage 2, weak (about $80/day); same stage 3 path, lower priority.
Registry: /root/pipeline/PIPELINE.md.