# testmgr4 sweep 12 Oct
All 4 running tests fresh (csv <1h old), no feed gaps seen.
- dip: fwd per arm A n12 mean 1.04x, B n12 0.69x, C n16 0.45x. Main arm A at breakeven; B/C below kill line. Hold to main-arm n=20, then kill if mean<1.
- lv: fwd n28 rows over arms (A 1.56x n8, B 1.85x n8, C 1.19x n12), 1 win each: median<1, outlier-driven. Decide 16 Oct.
- r300: fwd 2 rows, both 0.28x; flow ~0 (exp 1/day). Watch; if still <3 fwd by 13 Oct, retire.
- wh30: 7 fwd per arm, 6-7 wins each (mean 2.4x): too good, sellability unproven -> bug until audited. Needs outside-wallet sell check.
No instant misses/blocking requests found open.