# A1 #1705: v3 1M-crossing chaser, fill sim on tapes (11 Oct)
Hunt: enter at first 1M cross, v3 WETH no-hook, 300k->1M took >=180 s (not a jump), target 3-10M. Backs ledger #492 (a2, peak-based) with a path sim.
Data: e1x-runners + P2 tapes, 21 Sep-7 Oct, n=135 v3 pools (depth_usd_t1m>=20k for 133 of them: the depth filter removes nothing here, set already real). Lag 3 s, fee+slip 4%, pull-aware exit, pos 0.02 ETH, hold 6 h.
| arm | n | mean | median | ex-top3 | halves |
| tp5 sl.3 | 135 | 1.48 | .29 | 1.40 | 1.55/1.41 |
| tp10 sl.3 | 135 | 1.57 | .29 | 1.38 | 1.48/1.66 |
| tp5 sl.3, born-to-1M age<120 min | 112 | 1.67 | .30 | 1.58 | 1.68/1.61 |
| tp10 sl.3, age<120 min | 112 | 1.74 | .29 | 1.52 | 1.49/2.04 |
Wide stop needed: sl.6-.7 gives 1.0-1.2 (shakeouts), sl.2-.3 best. Hold 1 h only 1.15; 6 h 1.48. Age>=120 min is bad (0.58, n21): skip. Pos size 0.02-0.3 ETH barely matters (1.52-1.64).
Win rate 28-36%, median .29: a lottery payoff, 14 of 17 days positive.
RULE for paper: v3 WETH no-hook, radar "cross 1M" event, cap_first<150k, 300k->1M gap >=180 s, age at 1M <120 min, buy at event, stop 0.3x, tp 5x (arm B tp10), 6 h time stop, pull exit. Kill: mean<1.2 over 40 closed.
Caveat: tapes end 10 Oct, no forward check; sl.3 stop fills assume 30 s exit lag (included).