Asym v2 — Consolidated Research (4 streams complete)

A: Microstructure timing · B: Wider filters · C: No-gap original · D: No-gap + price≥$3 · 4 streams parallel research, 30+ кандидатов validated

4 DEPLOY-ready 8 PAPER candidates 15 NEEDS_MORE 10 REJECTED Apr 27, 2026

TL;DR — Combined Phase 1 upside ($15K BP)

$10-12K

Baseline v2 (honest)

$20-25K

v2.5 (improvements only)

$30-40K

v3 (multi-strategy)

×3

vs current honest baseline

Combined: organic_pump v2 baseline + Stream A timing improvements + Stream B universe expansion + Stream C/D new strategies parallel.

4 Stream summaries

Stream A — Time-of-day microstructure

Найдено: 3 huge findings для existing organic_pump. A6 ORB-no-break filter (skip 33% bad trades, +0.49 Sh swing). A1 exit at 10:00 (79% TPs до 10:00, 11:00 отдаёт winners). A7 sweep & reverse (WR 100% N=36).

Stream B — Wider filters

Найдено: ADV 500K+ universe лучший (WR 53.7% vs 52% baseline, ×2 capacity, $/yr +103%). Gap loosening 10%→5% даёт +33% capacity цена 8% edge. Stricter (gap≥15%) хуже — concentration не окупает потерю N.

Stream C — No-gap candidates (no price filter)

Найдено: first5_strong_up/dn на flat-open (WR 82.3% N=700, est daily Sh 1.9, +/-2% mean). Орthogonally от gap-based work. CAVEAT: execution 09:35 после confirmation, не MOO. PM volume patterns DEAD без gap.

Stream D — No-gap + price≥$3

Найдено: g3a_open_2atr_above_short_exADR (Sh real 1.53, $13K/yr, all 4 years pos). ATR-stretch вместо raw gap = volatility-normalized + ADR exclusion = clean US-domiciled edge. Categories A/B/F entirely DEAD без gap.

4 DEPLOY-ready кандидата (можно строить уже сейчас)

organic_pump_v2.5_improved (Stream A integration)

SHORT · improvements к existing baseline

$20-25K Phase 1

×2 baseline

A1+A6+A7

Apply 3 Stream A findings к текущему organic_pump: (1) exit 10:00 not 11:00 (+18%), (2) ORB-no-break filter (+50% Sh), (3) sweep & reverse bonus signal. Без новых данных — только code change.

gap≥+10% AND prev_close<$10 AND pm_dvol_z≥3 AND price≥$3 →
SHORT @ 09:30 → if 09:45-10:15 break above 09:30-09:45 high → COVER (skip)
→ exit at 10:00 ET (not 11:00) OR SL +1% OR TP −7%

g3a_open_2atr_above_short_exADR (Stream D)

SHORT · NEW strategy, no gap filter

N=940 4y

WR 31.1%

Sh 1.53 real

$13K/yr

Volatility-normalized stretch (ATR units, не gap %). All 4 years positive (1.15-2.30 Sh). ADR exclusion критична — без неё edge inflated foreign catch-up moves.

open_09:30 − prev_close > 2 × ATR_20 AND price ≥ $3
AND ticker NOT in [ING, UBS, UL, EQNR, BCS, TCOM, ...]
→ SHORT @ 09:30, exit 11:00, SL +1%, TP −7%

organic_pump_universe_expanded (Stream B C3)

SHORT · universe expansion к existing

WR 53.7% ↑

N ×2

$/yr ×2

Расширить universe с ADV 2M+ до ADV 500K+ для existing organic_pump. WR растёт (53.7% vs 52%) — мелкие/менее ликвидные пампы фейдят надёжнее. ×2 capacity. Caveat: spread/slip может быть хуже на иliquide tickers.

Existing organic_pump rule, but universe = ADV 500K+ (was 2M+)
Adds ~4,200 tickers to scope
Estimate: +$10-15K/yr Phase 1 incremental

first5_strong_directional (Stream C)

DYNAMIC · NEW strategy, completely orthogonal

N=700 per side

WR 82.3%

Sh 1.9 est daily

FLAT-open stocks (|gap|<2%) + первые 5 min directional move >2% with no opposite wick = institutional positioning revealed. Continuation play. Орthogonally от gap-based work — completely separate signal stream.

|gap_pct| < 2% AND first_5min_move > +2% AND no_opp_wick → LONG @ 09:35
|gap_pct| < 2% AND first_5min_move < −2% AND no_opp_wick → SHORT @ 09:35
Exit 11:00, SL 1%, TP 7%

⚠️ Execution 09:35 (after confirmation), не MOO. Different routing model.

v3 multi-strategy portfolio projection

Combined Phase 1 ($15K BP shared across 4 strategies)

Strategy	Side	Daily $ baseline	Phase 1 $/yr	Notes
organic_pump v2.5 (with A1+A6+A7)	SHORT	$80	$20K	Baseline + improvements
g3a (Stream D) — ATR-stretch SHORT	SHORT	$50	$13K	NEW orthogonal SHORT
universe expansion (Stream B C3)	SHORT	$40	$10K	Adds incremental signals
first5_directional (Stream C)	DYNAMIC	$30	$8K	Orthogonal signal stream
TOTAL combined Phase 1	DIVERSIFIED	$200	$50K	If all signals independent
− Overlap discount (BP shared, signals overlap)	−	−40%	$30-35K	Realistic with BP sharing

Combined estimate: $30-35K Year 1 Phase 1 with 4 strategies parallel deployed (vs $10-12K baseline single strategy = ×3). Year 2-3 plateau (Phase 6): combined может достичь $200-300K/yr (×2-3 от single strategy).

Stream A — детально (microstructure findings)

A6 ORB-no-break filter

FILTER — applies к existing organic_pump

+0.49 Sh

−33% bad trades

Watch 09:30-09:45 high. Если в 09:45-10:15 цена пробивает выше → SHORT DIES (mean −5.32%, WR 15%). Если НЕ пробивает → SHORT работает (mean +2.51%, WR 73%). Простое skip 33% trades.

A1 exit @ 10:00 not 11:00

EXIT RULE — applies к existing organic_pump

79% TPs до 10:00

+18% performance

Per-trade PnL peaks at 10:00 (+0.10%), потом decay −0.06% к 11:00. 79.5% TP-7% hits до 10:00, 91% до 10:30. 11:00 force-close отдаёт winners.

A7 sweep & reverse

BONUS signal — high quality, low N

WR 100% N=36

Sh 1.18

Open spike >+3% from open в первые 5 min, потом возвращается ниже open к 09:55. Perfect quality — 36 instances 4y all winners. Используем как scoring boost.

CLEAN stack (A6+A2 combined)

FILTER stack — 60% pass rate

WR 77.4%

+2.61% mean

Combined A6 (no break) + A2 (specific selection) → 60% events pass. Per-trade Sh 0.61 — solid.

⚠ Caveat for Stream A

Findings basis на 384 pump events из cache (470 tickers). Только 1/26 April 2026 trades был в cache (sub-2M ADV pumps часто не в cache). Нужна cache extension на full universe.

Stream B — детально (filter sweep)

Config	N (3y)	WR	Mean PnL	Edge / trade vs baseline	$/yr @ $1K	Verdict
Baseline v2 (gap≥10/px<10/vol≥3x/ADV2M)	3,232	52.0%	+3.11%	1.00×	$33K	baseline
C1 gap≥5%	4,597	50.6%	—	0.92×	$44K	BETTER +33%
C2 prev<$20	3,490	51.8%	—	0.97×	$35K	EQUIVALENT
C3 ADV 500K+ ⭐	6,367	53.7%	—	1.02×	$67K	BEST quality
C4 combined wide	10,536	50.7%	—	0.88×	$96K	Capacity, low edge
C5 narrow strict (gap≥15)	1,681	—	—	1.03×	$18K	WORSE

Сюрприз: illiquid universe (ADV 500K-2M) даёт выше WR — мелкие пампы фейдят более reliable. Stricter filters не помогают — pump fade не масштабируется выше 10% gap.

Stream C — детально (no gap filter)

first5_strong_up/dn_no_gap_follow

DYNAMIC · 09:35 entry

N=700/710

WR 82.3%

+/−2.0% mean

Sh 1.9 est daily

|gap_pct| < 2% (FLAT open) AND first_5min_move > +2% AND no_opposite_wick → LONG @ 09:35
|gap_pct| < 2% AND first_5min_move < −2% AND no_opp_wick → SHORT @ 09:35

ret5d_strong_up/dn_flat_open

DYNAMIC · momentum

N=1,711/1,232

WR 67.4%

+/−0.93%

Sh 0.96 est daily

5-day momentum ≥+/−10% AND flat open today → continuation в направлении momentum.

tested_prev_close_from_above_short

SHORT · massive capacity

N=11,079

WR 63.9%

+0.46%

Sh 0.67

Цена тестирует prev_close сверху → SHORT. Audit needed для execution model (09:45 entry).

Stream D — детально (no gap + price≥$3)

Candidate	Side	N (4y)	WR	Mean PnL	Real Sh	$/yr @ $15K	Verdict
g3a (open >2 ATR + ex-ADR) ⭐	SHORT	940	31.1%	+0.30%	1.53	$13K	DEPLOY
c1_monday_universe_long	LONG	44,911	43.9%	+0.06%	1.39	$128K basket	PAPER
d1_open_100bps_below_pmclose	LONG	6,012	39.4%	+0.06%	0.64	$17K	RESEARCH
g_stretched_up_combo_short	SHORT	2,716	26.7%	+0.09%	0.61	$11K	RESEARCH

g3a year-by-year stability

Все 4 года positive

Year	Sh real	Verdict
2023	1.81	Strong
2024	1.15	OK
2025	1.55	Strong
2026 YTD	2.30	Best year

Stability across 4 different regimes (2023 hawkish Fed, 2024 election, 2025 AI mania, 2026 YTD) = robust edge, not regime artifact.

Action plan

Immediate (1-2 weeks)

Apply A1 exit @ 10:00 к existing organic_pump (1 line yaml change). +18% performance free.
Apply A6 ORB filter к existing organic_pump (skip if break above 09:30-09:45 high). +0.49 Sh.
Build g3a plugin — Stream D top finding. Code в moo_1100_asym/plugins/. Paper trade 2 weeks.

Short-term (1 month)

Расширить universe organic_pump до ADV 500K+ (Stream B C3). +103% capacity, +2% WR.
Build first5_directional plugin (Stream C). Orthogonal LONG+SHORT signal stream.
Cache extension to full universe → re-validate Stream A на real April 2026 trades.

Mid-term (2-3 months)

Build Stream D's g_stretched_up_combo as Tier 2.
Build c1_monday_basket (Stream D, 44K events 4y).
Build d1_auction_undershoot (Stream D, 6K events).
Aggregate Sh measurement live across all 4-5 plugins.

Risk management

Halt threshold per strategy: −5% from start of Phase 1.
Aggregate halt: −10% combined → kill engine.
Max concurrent across all strategies = 5 (was 2).
Side balance ≤70% per side даже combined.