Gold & Silver Research Library

Annotated one-pagers for the 20 foundational papers underlying the gold/silver trading system. Each file contains the paper's thesis, core math, replication verdict, and mapping to our signal/strategy layer.

Index

Valuation & Fundamentals

Paper	Signal Layer	Verdict
Erb & Harvey (2013) -- The Golden Dilemma	Strategic allocation (gold/CPI ratio)	CONFIRMED
Barsky & Summers (1988) -- Gibson's Paradox and the Gold Standard	Primary macro factor (real rates)	CONFIRMED
Gorton & Rouwenhorst (2006) -- Facts and Fantasies about Commodity Futures	Portfolio construction, roll yield	PARTIALLY CONFIRMED

Risk Premia & Factor Strategies

Paper	Signal Layer	Verdict
Moskowitz, Ooi & Pedersen (2012) -- Time Series Momentum	Core trend signal (SS5.1)	CONFIRMED w/ caveats
Koijen et al (2018) -- Carry	Carry cost adjustment (SS5.2)	PARTIALLY CONFIRMED
Bessembinder (1992) -- Hedging Pressure	COT positioning signal (SS5.3)	PARTIALLY CONFIRMED
Baur & Lucey (2010) -- Is Gold a Hedge or a Safe Haven?	Tail-risk hedging (SS5.6)	CONFIRMED
Szakmary, Shen & Sharma (2010) -- Trend-Following in Commodity Futures	Trend signal calibration (SS5.1)	CONFIRMED w/ caveats
Fuertes, Miffre & Rallis (2010) -- Tactical Allocation in Commodity Futures	Momentum-carry interaction (SS5.1+SS5.2)	PARTIALLY CONFIRMED

Safe Haven & Portfolio Construction

Paper	Signal Layer	Verdict
Baur & McDermott (2010) -- Is Gold a Safe Haven? International Evidence	DM/EM safe-haven regime (SS5.6)	CONFIRMED
Hillier, Draper & Faff (2006) -- Do Precious Metals Shine?	Strategic allocation (5-15% PM weight)	CONFIRMED
Lucey & Li (2015) -- What Precious Metals Act as Safe Havens?	Safe-haven hierarchy, metal selection	CONFIRMED

Volatility & Macro Drivers

Paper	Signal Layer	Verdict
Tully & Lucey (2007) -- A Power GARCH Examination of the Gold Market	Vol model specification (APARCH)	CONFIRMED
Batten, Ciner & Lucey (2010) -- Macro Determinants of PM Volatility	Macro-augmented vol model	CONFIRMED

Silver-Specific

Paper	Signal Layer	Verdict
Batten, Ciner & Lucey (2015) -- Silver Stylized Facts	GSR mean-reversion (SS5.5), vol regime	CONFIRMED

Execution & ML Infrastructure

Paper	Signal Layer	Verdict
Almgren & Chriss (2000) -- Optimal Execution	Execution engine (SS5.7)	CONFIRMED
Lopez de Prado (2018) -- AFML	ML pipeline (SS5.4)	CONFIRMED

Overfitting Defense & Statistical Rigor

Paper	Signal Layer	Verdict
Bailey et al (2014) -- Probability of Backtest Overfitting	Strategy validation gate (SS5.8)	CONFIRMED
Bailey & Lopez de Prado (2014) -- Deflated Sharpe Ratio	Strategy validation gate (SS5.8)	CONFIRMED
Harvey, Liu & Zhu (2016) -- ...and the Cross-Section of Expected Returns	Multiple testing correction (SS5.8)	CONFIRMED

Verdict Summary

CONFIRMED: 13/20 papers
CONFIRMED w/ caveats: 2/20 (Moskowitz TSMOM, Szakmary trend-following)
PARTIALLY CONFIRMED: 4/20 (Gorton-Rouwenhorst, Koijen carry, Bessembinder hedging pressure, Fuertes-Miffre tactical)
REJECTED: 0/20

The four "partially confirmed" papers share a common theme: their conclusions hold for diversified commodity portfolios but are weaker when applied narrowly to gold and silver. Gold's persistent contango (negative carry), lack of traditional producer hedging pressure, and post-financialization correlation shifts all reduce the applicability of broad commodity research to precious metals specifically. The two "confirmed with caveats" papers (trend-following) work for precious metals but require metal-specific parameter calibration -- gold favors longer lookbacks, silver favors shorter breakout windows.

Key Findings Across the Library

Gold vs. Silver: Structural Divergence

The most consistent finding across the 20 papers is that gold and silver are fundamentally different assets despite moving together ~65% of the time: - Gold: monetary asset, safe haven, symmetric vol, driven by real rates and the dollar - Silver: hybrid monetary-industrial asset, NOT a safe haven, asymmetric vol, driven by both macro and business-cycle factors

Signal Hierarchy (Multiple-Testing Adjusted)

From Harvey-Liu-Zhu, our signals ranked by adjusted \(t\)-statistic: 1. Real rate sensitivity (\(t \approx 4.1\)) -- highest conviction 2. Trend/TSMOM (\(t \approx 3.2\)) -- high conviction 3. GSR mean-reversion (\(t \approx 2.7\)) -- supporting signal 4. COT hedging pressure (\(t \approx 2.3\)) -- supporting signal 5. Carry (\(t \approx 1.8\)) -- filter only, not standalone

Safe-Haven Architecture

From the Baur-Lucey, Baur-McDermott, and Lucey-Li papers: - Gold is a safe haven for developed-market equities only - Silver is NOT a safe haven (it is a risk asset) - Safe-haven effect activates with a 3-5 day lag in liquidity crises - Duration: ~15 trading days before dissipating

Key Dependencies

Real Rates (Barsky-Summers) ──> Macro Regime ──> Trend Signal (Moskowitz TSMOM)
         │                           │                    │
Erb-Harvey (Gold/CPI) ──> Strategic Allocation      Carry Filter (Koijen + Fuertes)
         │                           │                    │
Baur-Lucey + Baur-McDermott ──> Tail Hedge      COT Signal (Bessembinder)
  + Lucey-Li (Safe Haven)       Overlay                  │
         │                           │              Trend Calibration (Szakmary)
Hillier-Draper-Faff ──> PM Weight (5-15%)                │
         │                           │                    │
Batten (Silver Facts) ──> GSR Signal ──> Position Sizing ──> Execution (Almgren-Chriss)
         │                                                     │
Tully-Lucey (APARCH) ──> Vol Model ──> Risk Limits       ML Pipeline (AFML)
         │                                                     │
Batten-Ciner-Lucey ──> Macro Vol Drivers                 Meta-Labeling
  (Rates, DXY, IP)                                            │
                                         PBO + DSR (Bailey) + HLZ (Harvey) ──> Validation Gate