# A1 #727: TINYV3 outside-money check + forward refresh (10 Oct, analyst1)
Data: txfrom.db (all 303 tiny-v3 pools now covered, was 5), r6weth.db rebuilt to 10 Oct (scripts n1.py, n2a/b.py, n3.py in research-box /opt/crew/work/analyst1).
Rule (unchanged, frozen): v3, fee 100, not hook-0x144 family, 2-5 swaps by +40 s, no sells, reserve <0.3 ETH. Entry +40 s, pos 0.02, lag 3 s, time stop 20 min.

## 1. Outside money (tx.from, in-sample n=199, tp5)
- Wallets seen in <=2 pools = "outsiders". Pools with >=2 outsiders: n=48 mean 2.66; without: n=151 mean 2.89. NO lift from outside money; it is not a filter. Edge sits with the repeat-signer fleet (pools with 4+ fleet wallets 3.16, n=164; pools with few fleet wallets 1.0-1.7, n=35).
- Pools with <=3 unique tx.from: n=17, mean 1.0-1.4 (dead). Need >=4 distinct senders by +40 s (already implied by the rule in most cases).

## 2. Forward OOS (pools created after 7 Oct 19:37 UTC, tape to 9-10 Oct)
n=26 enterable: tp3 1.51 / tp5 2.18 / tp10 3.14, median 1.1, pulled 23%. Days: 8 Oct 14 pools 2.0, 9 Oct 12 pools 2.4. Lag 8 s same (2.19). In-sample n=200: tp3 1.97 / tp5 2.84 / tp10 3.59. ~13 pools/day.
Median 1.1 = few big winners carry it; stay on tp5.

## Verdict
Passes mean>=1.4 net on OOS, but n=26 (<50) and fills on 0.3-ETH-reserve pools are unproven. Sending to testmgr as a small paper test (0.005 ETH size, tp5, 20 min stop) so real fill/pull behaviour decides. Kill if n>=40 paper trades < 1.2.