# Shadow variant grid: live mc5 rule (paper replay, Test Manager)

Made 23 Oct 02:58 Bangkok. Same coins and fill model as the live rule's research (bt.py tick-walk on real pool depth, 1% each way + gas), every night. Paper only; nothing touches a wallet. Only Ben changes the live rules (D-030). Numbers are tickets: +1.0 = one ticket of profit ($50 at $50). 'w/o top 3' removes the 3 best trades. Random = share of random draws from the same pools (same exits) that this rule beats, per period (dumb baseline, D-032). Kill rule: behind the live rule on both all days and the last 7 days in 3 runs in a row = dropped.

| variant | last 7 d: trades, net, w/o top 3 | since 13 Sep: trades, net, w/o top 3, per day | wins | beats random | behind live (runs) |
|---|---|---|---|---|---|
| 0.3x, 4h, 150k-1M | 41, +5.2, -19.5 | 346, +326.4, +227.2, +9.87 | 19% | 97/79/96 | 0 |
| stop 0.3x | 35, +4.5, -20.2 | 275, +335.3, +236.1, +10.85 | 21% | 98/85/98 | 0 |
| no stop | 35, +3.8, -20.9 | 275, +344.1, +244.9, +11.13 | 21% | 98/85/99 | 0 |
| no stop, 4h, 150k-2M | 51, -3.1, -27.8 | 396, +364.5, +265.2, +11.02 | 19% | 96/77/97 | 0 |
| **LIVE (0.5x, 4h, 250k-1M, 5m)** | 35, -3.5, -23.7 | 275, +292.9, +193.7, +9.48 | 17% | 98/85/96 | - |
| band 250k-2M | 45, -9.4, -29.5 | 325, +309.7, +210.5, +10.02 | 16% | 97/85/95 | 0 |
| entry 3m | 41, -14.9, -30.4 | 303, +341.8, +246.4, +10.66 | 15% | 95/76/97 | 0 |

**Leader (last 7 days): 0.3x, 4h, 150k-1M** vs live: +5.2 vs -3.5 tickets this week, +326.4 vs +292.9 since 13 Sep.
Dropped so far: time 1h (11 Oct), time 12h (11 Oct), no stop, 12h (11 Oct), 0.3x, 12h (11 Oct), no stop, 12h, 250k-2M (11 Oct), 0.3x, 12h, 3m (11 Oct), band 150k-1M (15 Oct), entry 8m (18 Oct)

A leader is a candidate for Analyst 1, not a switch: a rule change goes through the pipeline and only Ben switches (D-030).