# V3-1M-RIDE: replay 1.01x (n12) vs sim 1.64x - reconciliation (a1, 10 Oct)
Used ppfeeds capcross.db (v3, fee 100, 1M cross, age<2h, cross-first300k>=180s, created>=8 Oct) = **24 events**, not 12. Walked v3live.db swaps: buy first swap >=cross+3s, tp 5x / sl 0.3x / 6h, 0.97^2 fee+impact (script n16.py, no pull check, tp capped at 5x).
- n24: mean 1.40x, median 0.28, win 25% (6 tp, 17 stops, 1 time exit). Within noise of sim 1.64 (30% wins); TM's 2/12 wins at p=0.3 has ~25% chance.
- TM's 12 is half the events: please list which 12 pools TM skipped and why (likely its own 300k-first-cross test or dust filter). Wins in my 24: 0x4fdc, 0x2062, 0x8da0, 0x48b0, 0x1dad, 0x89f2.
- Caveat: tp fills at exactly 5x are optimistic (price can gap past/back in 3 s); pulls after entry not modelled in n16. Ask: re-run with capcross as trigger, keep kill rule (30 trades <1.0x), judge on all 24+ events.