# h3n / d0fd 30-day replays on native-quote tape (#650)
Tape: /opt/ppcopy/r6nat.db (21 Sep - 9 Oct 05:00 UTC), converted to /root/tm/replay/tape_nat. Scripts: /root/tm/replay/h3n_nat, d0fd_nat (same code as forward, only feed changed, now=8 Oct 24:00 UTC).
d0fd launcher = pools whose first liq add went to 0xd0fd42.. (tx_to in d0fd_liq.csv; 71 pools overlap the tape, 68 seen).
## h3n (fee 3000 hooked, native)
- A (+600s sell): n233, mean 6.97x, zeros 43 (18%). B 7.02x. C (tp5/stop) 3.08x, zeros 68. D (>=3 swaps pre-entry) 4.06x n159 zeros 19. E/F (0xcc flag) 9.72x n145 zeros 2 / 8.75x n174 zeros 16.
- Sim expected A 5.2x; replay higher, mean is outlier-driven: audit on median and capped mean needed. Skipped 54 (shared hook).
## d0fd (fee 100 native, launcher d0fd42..)
- A tp5 n60 mean 3.74x zeros 12; B tp3 2.34x; L (+60s) 3.53x n57 zeros 14. Sim expected 4.16 / 2.46 / 3.75: matches.
## Caveats
- Entry fill timing "on-time" is moot in replay (no lag). No reserve impact. Sellability unproven (Ben's $5 sell test); operators pull, zeros counted as 0.
Files: h3n_nat/{pub.md,trades.csv}, d0fd_nat/{pub.md,trades.csv}