Sirtori Consulting · Quantitative Research · Signal Methodology v1.0

GOLD
INTELLIGENCE

Gold · Macro · Positioning · Signal · AI

A deterministic, rule-based market-regime framework for gold. Signal levels are not price predictions. No predictive power is claimed. The model determines the score.

Updated: — — Methodology: v1.0 Weights: v1.0 provisional
Section 01

Gold Market

Real-time spot prices · OANDA via TradingView

XAU / USD — Gold Spot LIVE
OANDA via TradingView · Click symbol to change currency
XAU / EUR LIVE
Gold / Silver Ratio LIVE
Silver
Platinum
Palladium
XAU / CHF
Section 02

Key Macro Drivers

FRED, TradingView, CBOE · Signal input indicators

DXY — US Dollar LIVE

Signal Factor 03 — Rule-based, parameter provisional

Real Yield 10Y TIPS LIVE

Signal Factor 02 — Rule-based, parameter provisional

US 10Y Nominal LIVE

Rate benchmark · Not a direct Signal input

Fed Funds Rate MONTHLY

Policy rate · Drives real yield expectations

GLD — Gold ETF LIVE

Signal Factor 04 — ETF proxy · Provisional input

VIX — Volatility LIVE

Signal Factor 07 — Provisional regime-based

Section 03 — Proprietary

Sirtori Gold Signal

Deterministic composite · 7 weighted factors · v1.0 provisional weights · WEEKLY UPDATE

Composite Score / 10 — —
—
0 Strong Bearish5 Neutral10 Strong Bullish
Signal Rationale

—

Model Status — v1.0
— A · Deterministic
Rule-based
— B · Historically
Calibrated
— C · Provisional
A = formula explicit and reproducible, parameters not empirically calibrated · B = normalisation based on historical observations (none yet) · C = methodology or data not yet sufficiently robust
Signal History (logged observations)
No observations yet
—
0 observations · 0 KB
Factor Breakdown · v1.0 provisional weights
Factor / Raw InputClassWeightScoreContribution
Loading...

Score scale: 0–2 Strong Bearish · 2–4 Bearish · 4–6 Neutral · 6–8 Bullish · 8–10 Strong Bullish
COT crowding: Net long above 80th percentile triggers crowding discount — flagged independently of directional score
Weights: v1.0 provisional — will be reviewed after sufficient historical observations. No optimization before backtesting.
Model objective: Reproducibility and auditability. The model determines the score. If the result is 4.8, 4.8 is shown.

Section 04

What Changed?

Key market developments · Last 7 days · Updated weekly by Andrea Sirtori

Loading...
Section 05

AI Market Commentary

Claude (Anthropic) interprets signal data · Not investment advice · No predictive claim

AI Analysis · Claude · Signal v1.0 —
Generating commentary from signal data...
Section 06 — Human Advisory

Sirtori View

Qualitative advisory · Distinct from the quantitative Signal · The two may disagree · Updated manually

The Sirtori View represents independent human judgment. It is intentionally separate from the quantitative Signal and may reach different conclusions. The Signal is systematic and rule-based. The Sirtori View incorporates judgment, context and experience. Both are shown transparently.

Short Term · 1–3 Months—

Not yet published.

Medium Term · 6–12 Months—

Not yet published.

Long Term · 3–5 Years—

Not yet published.

Section 07

Macro Charts

Interactive · TradingView / FRED

Gold — WeeklyLIVE
DXY — WeeklyLIVE
Real Yield 10Y TIPSLIVE
GLD — Gold ETFLIVE
Section 08 — Distinctive

Physical Gold Market

Spot prices · Physical market data integrated progressively

XAU/USD LIVE

Primary global benchmark

OANDA via TradingView
XAU/EUR LIVE

European FO reference

OANDA via TradingView
XAU/CHF LIVE

Swiss custody reference

OANDA via TradingView
Shanghai Premium N/A

N/A

SGE premium over LBMA. Data license pending.

Central Bank Demand QUARTERLY

—

WGC quarterly. Updated manually after each WGC release.

Real Yield 10Y LIVE

Strongest macro predictor

Fed Funds MONTHLY

Policy rate

Gold/Silver Ratio LIVE

Relative value indicator

Section 09

Data & Sources

SourceDataFrequencyStatusLicense
OANDA via TradingViewXAU/USD, XAU/EUR, XAU/CHF spotReal-timeLIVEWidget (free)
FRED — Federal ReserveReal yield DFII10, Fed Funds, EUR/USDDaily/MonthlyLIVEPublic domain
TradingViewDXY, US10Y, Metals, GLD, VIXReal-timeLIVEWidget (free)
World Gold CouncilCentral bank demand (Signal F05) · ETF flows proxy (Signal F04)Quarterly/MonthlyQUARTERLYPublic reports
CFTCCOT gold futures positioning (Signal F06)WeeklyWEEKLYPublic domain
Anthropic / ClaudeAI commentary — interpretation only, not a data sourceOn demandACTIVEAPI
Section 10

Signal Methodology v1.0

Full documentation · Every formula, weight and classification · Changes increment the version number · No undocumented assumptions

The Sirtori Gold Signal is a weighted composite score (0–10) across 7 independent market factors. Each factor is normalised to 0–10 via a factor-specific deterministic formula. The composite is the weighted average. The model determines the score — it is not optimised to produce any target result. No predictive power is claimed. The framework will be investigated for any relationship between Signal levels and subsequent gold returns only after sufficient historical observations exist.

Model Status — v1.0

A · Deterministic / Rule-based

Formula explicit and reproducible. Parameters not empirically calibrated from historical data.

5 / 7

Momentum · Real Yields · USD/DXY · Central Banks · COT

B · Historically Calibrated

Normalisation based on historical observations with documented statistical methodology.

0 / 7

None yet — target for Phase 2 development

C · Provisional

Methodology or underlying data not yet sufficiently robust for institutional use.

2 / 7

ETF Flows · Volatility

Look-ahead bias: When calculating a historical observation at date T, only data available at T is used.
Predictive power: No predictive claim is made. Relationship between Signal levels and subsequent returns will be investigated only after sufficient observations exist.
Version control: Changes to formulas, weights or normalisation rules increment the methodology version. All logged observations retain the version under which they were calculated.
Optimisation: Weights will not be changed based on current Signal results. Changes require historical evidence and explicit approval.

Development Roadmap — Phase 2

ETF Flows → Class B (Historically Calibrated)

Replace discrete -2/+2 input with actual GLD/IAU weekly flow data. Calculate 4W and 12W cumulative flows as percentile vs 3-year history. Source: SPDR/iShares public data or WGC monthly. Target: v1.1.

Volatility → Class B

Replace VIX regime breakpoints with historical percentile of VIX + gold realised volatility. Investigate non-linear VIX/gold relationship using historical data. Target: v1.1 or v1.2.

Real Yields & DXY: empirical parameter calibration

Replace provisional coefficients (1.2 for yields, 106 anchor for DXY) with historically calibrated parameters derived from FRED data (TIPS from 2003, DXY from 1971). Target: v1.2.

COT: automated percentile from historical CFTC data

Replace manually-entered 3Y min/max with automated calculation from CFTC weekly CSV history. Eliminate single-point-of-failure on manual input. Target: v1.2.

Momentum: historical distribution analysis

Analyse historical distribution of Momentum scores before considering any formula change. Document how frequently scores of 8+, 9+ occur. No formula changes before evidence review. Target: v1.2.

Backtesting framework

After 26+ weekly observations, analyse relationship between Signal levels (2–4, 4–6, 6–8, 8–10) and subsequent gold returns (30d/60d/90d). Publish results transparently regardless of outcome. Target: v2.0.

Important Disclaimer

Sirtori Gold Intelligence is for informational and educational purposes only. It does not constitute investment advice or an offer to buy or sell financial instruments or physical precious metals. The Sirtori Gold Signal is a quantitative market-regime framework — no predictive power is claimed. Signal levels are not price predictions. Data may be delayed. AI commentary is generated by Claude (Anthropic) from signal data only — it does not access real-time markets independently. Past performance is not indicative of future results. Sirtori Consulting is registered with OCF. P.IVA 04522760166 · Via Manzoni 13, 24040 Castel Rozzone (BG).

Signal Admin — Methodology v1.0
Enter OBSERVABLE MARKET VALUES only. Scores are computed deterministically by the v1.0 formulas. You cannot set scores directly. Every save creates a logged observation.