XAU-COMEX Intelligence — Owner Observatory

Data source lock

GOLD FUTURES REFERENCE
Symbol: GC=F
Source: Yahoo Finance via yfinance
Canonical library: github.com/ranaroussi/yfinance
Target market: Exness XAUUSDm (only — no substitution)

NOT CME DIRECT FEED

No fallback provider. No interpolation. No invented bars. Fail-closed on degraded data.

XAU now

InstrumentExness XAUUSDm

Last observed bar (UTC)2026-10-09T20:57:00Z

Last close4194.988

Spread240 pts

Data age1d 7h 31m

FreshnessSTALE (market closed — weekend)

COMEX GC=F now

Reference symbolGC=F

Last bar (UTC)2026-10-09T20:59:00Z

Last close4216.30

SourceYahoo Finance via yfinance

Interval1m

FreshnessSTALE (market closed — weekend)

GC → XAU relationship

Frozen study window 2026-10-05 .. 2026-10-09 — 5,386 verified 1-minute intersection observations.

GC minutes6,533

XAU minutes5,632

Intersection5,386

Contemporaneous r (range)0.971 – 0.987

Verdict: GC and XAU move contemporaneously — the strongest correlation sits at lag 0 in every session. Incremental predictive power of GC(t) for XAU(t+k) beyond XAU(t) is ≤ 0.23% (partial r ≤ 0.011). No evidence that GC=F leads XAUUSDm at 1-minute resolution.

Live 1-minute

Continuous per-minute capture is LIVE (Phase 2 research lane deployed 2026-10-11). This panel reads the collector's own health feed and always shows the true age — never a synthetic "live" figure.

Source last bar (UTC)2026-10-09T20:57:00Z

Received (UTC)—

Age1d 7h 31m

Freshness stateSTALE (market closed — weekend)

Symbol lockXAUUSDm

Broker time ↔ UTCUTC_CONFIRMED

Reading live collector feed…

ASIA

Intersection minutes1,917

Contemporaneous r (0m)0.9714

Best |r| lag0m

Incremental R² (best)0.00226

Same-minute agreement0.915

Shock follow-through +1m0.527

0/±1..5-minute correlation table

Lead/lag	n pairs	Pearson r
XAU LEAD +5m	1879	0.0104
XAU LEAD +4m	1886	-0.0123
XAU LEAD +3m	1893	0.0025
XAU LEAD +2m	1900	-0.0025
XAU LEAD +1m	1907	0.0482
CONTEMPORANEOUS (0m)	1914	0.9714
GC LEAD +1m	1907	0.0434
GC LEAD +2m	1900	0.0103
GC LEAD +3m	1893	-0.0130
GC LEAD +4m	1886	-0.0068
GC LEAD +5m	1879	0.0024

EUROPE

Intersection minutes2,362

Contemporaneous r (0m)0.9819

Best |r| lag0m

Incremental R² (best)0.00112

Same-minute agreement0.944

Shock follow-through +1m0.506

0/±1..5-minute correlation table

Lead/lag	n pairs	Pearson r
XAU LEAD +5m	2337	-0.0292
XAU LEAD +4m	2342	0.0067
XAU LEAD +3m	2347	-0.0338
XAU LEAD +2m	2352	-0.0206
XAU LEAD +1m	2357	0.0394
CONTEMPORANEOUS (0m)	2362	0.9819
GC LEAD +1m	2357	0.0431
GC LEAD +2m	2352	-0.0162
GC LEAD +3m	2347	-0.0372
GC LEAD +4m	2342	0.0051
GC LEAD +5m	2337	-0.0253

AMERICA / COMEX

Intersection minutes2,392

Contemporaneous r (0m)0.9833

Best |r| lag0m

Incremental R² (best)0.00088

Same-minute agreement0.932

Shock follow-through +1m0.500

0/±1..5-minute correlation table

Lead/lag	n pairs	Pearson r
XAU LEAD +5m	2362	-0.0279
XAU LEAD +4m	2367	0.0105
XAU LEAD +3m	2372	-0.0223
XAU LEAD +2m	2377	-0.0168
XAU LEAD +1m	2382	0.0105
CONTEMPORANEOUS (0m)	2387	0.9833
GC LEAD +1m	2382	0.0145
GC LEAD +2m	2377	-0.0158
GC LEAD +3m	2372	-0.0212
GC LEAD +4m	2367	0.0125
GC LEAD +5m	2362	-0.0246

EUROPE-AMERICA OVERLAP

Intersection minutes1,162

Contemporaneous r (0m)0.9874

Best |r| lag0m

Incremental R² (best)0.00075

Same-minute agreement0.950

Shock follow-through +1m0.500

0/±1..5-minute correlation table

Lead/lag	n pairs	Pearson r
XAU LEAD +5m	1137	-0.0264
XAU LEAD +4m	1142	0.0107
XAU LEAD +3m	1147	-0.0347
XAU LEAD +2m	1152	-0.0224
XAU LEAD +1m	1157	0.0547
CONTEMPORANEOUS (0m)	1162	0.9874
GC LEAD +1m	1157	0.0563
GC LEAD +2m	1152	-0.0179
GC LEAD +3m	1147	-0.0346
GC LEAD +4m	1142	0.0112
GC LEAD +5m	1137	-0.0230

Daily

UTC day	GC	XAU	Intersection
2026-10-05	1,140	1,378	1,137
2026-10-06	1,377	1,378	1,375
2026-10-07	1,379	1,378	1,377
2026-10-08	1,379	998	997
2026-10-09	1,258	500	500

Automated daily reporting is not yet deployed (Phase 6).

Weekly

Persistence research across weeks requires an extended window. Automated weekly reporting is not yet deployed (Phase 7). Longitudinally, the XAU side is characterised over 69 months (2,009,997 bars) — see RESEARCH.

Research

Session-conditioned 1-minute return volatility, XAU side, 2021-02-02 → 2026-10-08 (2,009,997 verified bars, 1,767 days):

ASIA0.9764

EUROPE1.196

AMERICA1.205

OVERLAP1.473

The EU-US overlap is the most volatile window — the place to look for any GC→XAU relationship.

Hypothesis status

ID	Status
HYP-001	NO_EVIDENCE (contemporaneous co-movement dominates; no incremental lead detected)
HYP-002	WEAK
HYP-003	WEAK
HYP-004	WEAK
HYP-005	NO_EVIDENCE
HYP-006	NO_EVIDENCE
HYP-007	NOT_TESTED (GC volume/impulse model deferred)
HYP-008	NO_EVIDENCE

Spread regimes

Regime (pts)	n	1m vol	mean \|ret\|
tight	509,666	1.3550	0.7229
sub_normal	61,437	0.8175	0.4957
normal_200	1,052,394	0.4518	0.2874
wide_201_280	214,462	1.9234	1.2406
very_wide_281_350	131,655	0.6092	0.2866
extreme_gt350	38,774	2.4186	1.2899

Spread bands are confounded with calendar era (r(spread,|return|) = 0.070), so HYP-006 is not cleanly testable on this corpus.

Workers

Worker	Status
YFINANCE-GC-COLLECTOR	NOT DEPLOYED (Phase 2)
EXNESS-XAUUSDM-COLLECTOR	DEPLOYED (Phase 2 research lane)
SESSION-CLASSIFIER	SPEC FROZEN (15/15 DST vectors)
GC-XAU-LEADLAG-WORKER	RUN (offline, frozen window)
XAU-MINUTE-ANALYST	NOT DEPLOYED (Phase 5)
XAU-DAILY-ANALYST	NOT DEPLOYED (Phase 6)
XAU-WEEKLY-ANALYST	NOT DEPLOYED (Phase 7)
DATA-QA-WORKER	RUN (gap classification, fail-closed)

The Phase 2 research lane runs as its own read-only service (xau-comex-research-lane), fully separate from the MT5 strategy/execution lane: no MT5 session, no order path, no public port, read-only data mount. It never places orders.

Data health

Provider calls in analysis0

InterpolationNONE

Fallback providerNONE

Broker-time ↔ UTCMATCH

Gaps — reported, never filled

Kind	Count
daily_break	1,134
weekend	296
other	40

Known XAU gap: 2026-10-08T16:38Z → 2026-10-09T12:37Z (1,200 min) — preserved and reported, not filled.

Session-clock validation (DST)

Verdict US_DST_FINGERPRINT_CONFIRMED — 58/58 strict month checks match, 0 mismatches. The daily gold break is pinned to 17:00 America/New_York and its transition months are March/November (US DST), confirming a US/CME-anchored session clock.

Provenance

Artifact	SHA256 (16)
`analysis/intersection/GC_XAU_INTERSECTION_1m_2026-10-05_2026-10-09.csv`	`900610e2f45851fb`
`analysis/leadlag/LEADLAG_BY_SESSION.json`	`81f51c95f1a83c2f`
`analysis/leadlag/LEADLAG_CORR_MATRIX.csv`	`7affac2330192e62`
`analysis/leadlag/LEADLAG_INCREMENTAL.csv`	`2bb0f37996b8716f`
`analysis/leadlag/SESSION_DESCRIPTIVES.csv`	`5f3c4e99acc9d926`
`analysis/leadlag/SHOCK_RESPONSE.csv`	`2bb68e08c84e7991`
`analysis/longitudinal/XAU_DST_VALIDATION.json`	`5b5cbaa8be41ca61`
`analysis/longitudinal/XAU_LONGITUDINAL_SUMMARY.json`	`426b6333a6f9e063`
`analysis/longitudinal/XAU_MONTHLY_SESSION.csv`	`e88162f856c57c73`
`analysis/longitudinal/XAU_SPREAD_REGIMES.csv`	`d42bfbe3a6849763`
`docs/PROJECT_BASELINE.md`	`a851f2dd1e4ab51b`
`evidence/intersection/INTERSECTION_DATASET_MANIFEST.json`	`7fe9bec6dbc0b24b`
`evidence/leadlag/LEADLAG_ANALYSIS_MANIFEST.json`	`98a8d9826396cd88`
`evidence/longitudinal/LONGITUDINAL_MANIFEST.json`	`4042315e3fb8fe1c`
`intake/OWNER_DIRECTIVE_BASELINE_V1.0.md`	`271eb988b886a6ca`
`normalized/gc_1m/GC_F_1m_2026-10-05_2026-10-09.csv`	`5a5266106ef638af`
`normalized/xau_1m/XAUUSDm_1m_2026-10-05_2026-10-09.csv`	`79d1119cfd2655f5`
`raw/gc_yfinance/GC_F_1m_2026-10-05_2026-10-09.csv`	`ddf59d20cd3396a7`
`raw/xau_exness/XAUUSDm_1m_2026-10-05_2026-10-09.raw.jsonl`	`4fbe5fbb057732d9`
`scripts/build_intersection.py`	`5340e030f59dfac6`
`scripts/build_leadlag.py`	`e8a9e02854f0a99e`
`scripts/build_longitudinal.py`	`a24243d777a5f2c6`
`scripts/build_manifest.py`	`90325e5897b9b53b`
`scripts/build_observatory.py`	`39f6d6e5a29a01c8`
`scripts/session_vectors.py`	`324b065ee84170da`
`scripts/verify_baseline.py`	`9f8017ac1843e9b1`

Full manifest: BASELINE_MANIFEST.json (26 hashed artifacts). Baseline v1.0 · intake 271eb988…e219.