MOC v8 Hybrid Apr 27

Why v8? User pushback on v7 hard-skips ("лучше пусть будут режимы и фазы"). Found bug in v15: size used A∪B mean even when C edge selected → BEAR_CALM × PANIC C-edge +0.191 was zeroed. v8 fixes it AND adds 0.25× floor (no hard zeros).

Headline metrics @ $10K/trade (size 1.0)

60v8 / 40v7 BEST

$+13,036/yr

Sharpe 1.76 · MaxDD $-2,498

$/DD 10.32× · active 458/496d (92%)

v8 hybrid 0.25 floor

$+13,194/yr

Sharpe 1.62 · MaxDD $-4,080

$/DD 6.39× · active 391/496d (79%)

v7 hard-skip

$+12,799/yr

Sharpe 1.59 · MaxDD $-3,502

$/DD 7.22× · active 303/496d (61%)

v15 (zeros bug)

$+9,820/yr

Sharpe 1.48 · MaxDD $-1,843

$/DD 10.53× · active 296/496d (60%)

Raw (no regime)

$+5,639/yr

Sharpe 0.54 · MaxDD $-8,305

$/DD 1.34× · active 472/496d (95%)

2026 caveat: All phase-aware variants struggle in 2026 YTD (regime shift?). v8 −$1.2K, v7 +$89, raw −$4.8K. Only v15 zeros-out works +$1.8K. Ensemble 60v8/40v7 cuts 2026 loss to −$667 while keeping 2025 upside.

Equity curves (cumulative $ on $10K/trade)

Per-year breakdown

Year	60v8 / 40v7	v8 hybrid 0.25 floor	v7 hard-skip	v15 (zeros bug)	Raw (no regime)
2024	$+3,698 Sh 1.17 · 169d	$+2,372 Sh 1.02 · 138d	$+5,686 Sh 0.98 · 126d	$+1,451 Sh 0.87 · 115d	$+2,259 Sh 0.35 · 175d
2025	$+22,738 Sh 2.47 · 228d	$+24,880 Sh 2.43 · 199d	$+19,526 Sh 2.24 · 146d	$+16,188 Sh 1.79 · 156d	$+13,735 Sh 1.30 · 232d
2026	$-667 Sh -0.45 · 61d	$-1,170 Sh -0.52 · 54d	$+89 Sh 0.08 · 31d	$+1,773 Sh 3.56 · 25d	$-4,847 Sh -1.44 · 65d

v8 phase × event recommendations

Phase	Event	A∪B N	A∪B mean%	C N	C mean%	Edge	Size×
BULL_CALM	NORMAL	1535	0.092	682	0.123	C_only	1.00×
BULL_CALM	PANIC	68	-0.049	55	-0.164	A∪B	0.25× FLOOR
BULL_CALM	GREED	18	0.090	4	-0.271	A∪B	0.75×
BULL_VOL	NORMAL	29	0.384	11	0.637	A∪B	1.50×
BULL_VOL	GREED	54	-0.196	17	0.012	C_only	0.50×
BEAR_CALM	NORMAL	566	-0.088	355	0.064	C_only	0.75×
BEAR_CALM	PANIC	140	-0.238	117	0.191	C_only	1.00×
BEAR_CALM	GREED	27	-0.094	5	-0.648	A∪B	0.25× FLOOR
BEAR_VOL	NORMAL	140	0.559	53	1.379	C_only	1.50×
BEAR_VOL	PANIC	243	-0.043	424	-0.009	C_only	0.25× FLOOR
BEAR_VOL	GREED	30	-0.362	3	-0.372	A∪B	0.25× FLOOR

Recommendations

Production deploy: 60% v8 / 40% v7 ensemble

Sharpe 1.76 — beats both standalone (v7 1.59, v8 1.62)
MaxDD −$2,498 — best of all variants
$/DD 10.32× — exceptional risk-adj
$/yr +$13,036 on $10K capital
Implementation: run BOTH v7 and v8 daily. Final position = 60% × v8_signal + 40% × v7_signal.

Pure v8 if you prefer single system aligned with "no hard skips" — Sharpe 1.62, $/yr $13.2K, but DD −$4K, weak 2026.

Pure v7 if you accept hard skips — Sharpe 1.59, robust across years.

MOC→19:00 — v8 Hybrid + 4-way + Ensemble