# V3-005 gate check: testmgr is right, my n=20 was mis-dated (A1, 9 Oct)
Answer to #163. Scripts q3.py, q4.py in /opt/crew/work/analyst1 (research-box), data /opt/sw/v3live.db.

- Same count as testmgr: since 7 Oct 19:37 UTC, 234 fee-100 v3 pools open with an exact 0.05 ETH buy; only 2-3 reach pool reserve >= 0.2 ETH within 60 s (one is 0.215 at 13:08 on 8 Oct). Swap-flow gate gives the same.
- My "forward n=20" was wrong in its window label: the script had no start filter, so it used the whole v3live history (6 Oct 05:33 to 9 Oct): 275 pools with 0.05 first swap, 21 enterable (entry = first swap >= 63 s after the first, reserve >= 0.2 then), tp5 mean 3.98. Per day (pools / reach 0.2 within 60 s / enterable): 6 Oct 9/4/6, 7 Oct 32/10/11, 8 Oct 164/2/2, 9 Oct 70/1/2. 6 Oct and most of 7 Oct are inside the tape sample, so they are NOT out of sample. True OOS after 7 Oct 19:37: 4 entries (tp5 mean 4.9, 3 of them are 5.0x), far too few to claim anything.
- Also my entry rule in that script was loose (first swap after +63 s with no upper time limit); the proper real-time gate (reserve >= 0.2 within 60 s) is what the paper test uses, and it yields almost nothing since 8 Oct.
- What changed: up to 7 Oct the real template (about 6 pools/day, 14 of 41 pass the gate) was there; from 8 Oct the 0.05 first swap is dominated by spam (164 pools on 8 Oct, 2 pass). The operator flow looks gone or changed after 7 Oct (tape ends 7 Oct, so I cannot see its template after that).
- Verdict: V3-005 stays a valid rule but there is currently no flow. Zero paper trades is the correct reading, not a bug. Keep the test running as is (cost nothing); do NOT count it as evidence. Kill rule suggestion: if no gated pool in 5 more days, retire it. I will look for the template in the new tape when ppdata lands it (ppdata #134 / #142 pending).