Centaur_Quant_Architecture/00_CONCEPT/EXP003_RESULTS/RAW_MANIFEST.md

EXP-003 — RAW EVIDENCE MANIFEST (Phase 3)

Tanggal: 2026-08-13 · Baseline produksi (git HEAD): f81d85defc1768a4ab4adaf316cbe79dcf2dec29 Protokol: EXP-003_SCANNER_ACTIVATION.md v1.0.0 (sha256 493cd7b35628e684...) Window: XAUUSD M15, 2026-02-01 → 2026-07-31 · Model=1 (OHLC 1m) · InpEnableTrading=false semua arm

1. Hasil per Arm (raw)

Arm	Konfigurasi (.set)	Setup count	FUSION calls	exp003_setups.csv
C0	baseline v2.8.0 (prox 0.25, depth 200, exp 1.2, no-mitigated)	0	0	(tidak dibuat)
B1	prox 0.50	0	0	(tidak dibuat)
B2	depth 400	0	0	(tidak dibuat)
B3	prox 0.50 + depth 400	0	0	(tidak dibuat)
B4	exp 1.0 + allow_mitigated=true	0	0	(tidak dibuat)

Semua 5 run: Test passed (teknis OK). exp003_setups.csv hanya dibuat bila Scan() menemukan kandidat → tidak ada file = 0 kandidat (evidence itu sendiri).

2. Verifikasi Input (dari log agent "started with inputs")

Run	Arm	InpEnableTrading	InpScanDepth	InpProximityMult	InpExpansionMin	InpAllowMitigatedOB
1	C0	false	200	0.25	1.2	false
2	B1	false	200	0.50	1.2	false
3	B2	false	400	0.25	1.2	false
4	B3	false	400	0.50	1.2	false
5	B4	false	200	0.25	1.0	true

3. Deteksi Scanner (log agent, seluruh run eksperimen)

Bullish OB+FVG @ : 0
Bearish OB+FVG @ : 0
FUSION lines     : 0
no-zone logs     : 191   (throttle 1/10 dtk wall-clock; ~26 scan/6 bulan per arm)

4. Diagnostic Density Pola Mentah (data terminal, 2.848 bar M15, 2026-06-18 → 07-31)

Pola dasar triple-candle displacement+FVG dihitung tanpa gate expansion/mitigation/proximity:

expansion >= 1.0 : bull=0  bear=0  total=0
expansion >= 1.2 : bull=0  bear=0  total=0

→ Data mentah menunjukkan pola dasar (OB bearish + impulse bullish close>high[i] + FVG low[i-2]>high[i], dan cerminnya) tidak muncul sama sekali di sampel tersebut.

5. Keputusan Gate A (per protokol §6)

setup_frequency < 2/bulan pada SEMUA arm (0 setup / 6 bulan) → Gate A TIDAK LULUS. Per protokol: masalah signal generation lebih dalam (definisi pola OB/FVG, bukan threshold proximity/depth/expansion/mitigation) → evaluasi hipotesis baru. Tanpa tuning lanjutan.

6. Hash Artifact (SHA-256)

Artifact	Hash
EXP003_C0.set	732d2230e1fb76ed38b8263e055f2a1d26055752b79ece0ab600b731b678dbfb
EXP003_B1.set	7da55578375bbc96d34d95d4f365637ee0229474fca634dd576beffe77cdfb25
EXP003_B2.set	39ceb93973026c5e298ee894edb4cae0845252bea1218aea5d4ac8ac3f598fa9
EXP003_B3.set	2d505027d8ae13c1ae9369bd018f415800b78987591b6438d177f2815b5d8689
EXP003_B4.set	7bca325e29da661df7e2c772b0e036d30cf3431829ae5654f5f514ce80648212
EXP-003_SCANNER_ACTIVATION.md	493cd7b35628e6842693ba196e8dbb0141172615d6c55d09f9818dd946bc725a
diagnostic_pattern_density.py	6d9b5f0d0c2c8f341cf865a9c92b35de408d318beb21277cb43a6a73b918fd58
raw_history_sample_2026-06-18_2026-07-31.txt	7bde8c9c339634589c61e57402d9cbb4d71ec293a5a1f21975299c9d43625e85
Baseline produksi (HEAD)	f81d85defc1768a4ab4adaf316cbe79dcf2dec29