Chapter 1: Precious Metals Primer
Reading time: 15 minutes | Difficulty: Beginner-Intermediate | Prerequisites: Chapter 0
This chapter answers one question: why does this system trade gold and silver instead of a diversified commodity basket?
Why Gold and Silver, Not Broad Commodities
QGTM started as a 42-ETF commodity platform covering energy, agriculture, industrial metals, and precious metals. We narrowed to gold and silver for three reasons:
1. Edge concentration. In quantitative trading, edge comes from knowing something the market has not yet priced. Gold and silver have a unique set of structural drivers -- real interest rates, central bank reserves, physical demand from jewelry and industry -- that produce persistent, measurable signals. Spreading the same research effort across 42 instruments diluted the signal quality in each.
2. Cross-strategy synergy. All 29 of our strategies interact with the same macro variables (DFII10 real yields, DXY dollar index, COMEX open interest, COT positioning). When one strategy detects a regime shift, the information propagates naturally to the others because they share the same driver set. In a broad commodity system, a corn weather model and a gold real-rate model share almost no signal.
3. Capacity. GLD alone trades 8 million shares per day. SLV trades 12 million. GDX trades 20 million. This liquidity means we can run 29 strategies simultaneously without moving the market. Try running 29 strategies on cocoa futures (NIB trades 30K shares/day) and you become the market.
Gold: The Monetary Metal
Gold is not like other commodities. You do not consume it. The gold mined 3,000 years ago still exists somewhere -- in a central bank vault, in jewelry, in an ETF. Total above-ground gold stocks are approximately 210,000 tonnes. Annual mine production adds roughly 3,500 tonnes (1.7% of stock). This stock-to-flow ratio is the highest of any commodity.
What this means for trading: Gold's price is not determined by supply-demand balance the way oil or wheat prices are. Instead, gold prices are driven by the opportunity cost of holding a non-yielding asset. That opportunity cost has a name: the real interest rate.
The Real-Rate Relationship
The 10-year TIPS real yield (FRED series: DFII10) is the single most important variable in the QGTM system. The relationship:
- Real yields fall (rates decline or inflation rises) -- gold becomes more attractive vs. bonds. Price rises.
- Real yields rise (rates increase or inflation falls) -- gold becomes less attractive. Price falls.
This inverse correlation has held with R-squared above 0.70 for most of the past two decades. When it breaks (2022-2023, when gold rallied despite rising real yields), the break itself is a signal: central bank buying was overwhelming the rate channel.
Our Real Rate Gold strategy (S1.A.01) directly trades this relationship with a z-score threshold model: - DFII10 falls below -0.5%: scale into gold - DFII10 rises above +1.5%: scale out - Rate of change matters more than level in regime transitions
The Dollar Relationship
Gold is priced globally in USD. When the dollar weakens, gold becomes cheaper for non-US buyers, increasing demand. The trade-weighted dollar index (FRED: DTWEXBGS) is the second most important input.
Our DXY Gold strategy (S1.A.02) trades this: - DXY trending down (20-day SMA below 60-day): long gold - DXY trending up: reduce gold exposure - DXY extreme (z-score > 2.0 on 252-day lookback): mean-revert via gold
Central Bank Buying
Central banks hold approximately 36,000 tonnes of gold -- 17% of all above-ground stock. When central banks are net buyers (as they have been since 2010, with acceleration since 2022), it creates persistent upward pressure that does not show up in traditional rate models.
Our Central Bank Gold strategy (S1.A.05) uses IMF IFS quarterly data to trend-follow central bank net purchases. It has a 180-day decay halflife because central bank behavior changes slowly.
Safe-Haven Flows
Gold's third macro driver is risk aversion. When equity volatility spikes (VIX > 25), capital flows into gold as a safe haven. This effect is asymmetric: gold responds more to fear than to greed.
Our VIX Haven strategy (S1.A.04) enters gold positions when VIX crosses above its 90th percentile of trailing 252 days, then scales out as vol normalizes.
Silver: The Industrial-Monetary Hybrid
Silver is fundamentally different from gold. Approximately 50% of silver demand is industrial (solar panels, electronics, medical devices), while the other 50% is investment and jewelry. This dual nature makes silver:
- More volatile than gold -- silver's annualized vol is typically 25-35% vs. gold's 15-20%
- Higher beta to gold -- when gold moves +1%, silver often moves +1.5-2%
- Correlated with industrial activity -- silver underperforms in recessions, outperforms in industrial booms
Why This Matters for the System
The gold-silver duality creates two distinct strategy opportunities:
-
Macro strategies work on silver's gold-like component. When real rates fall, silver rises (but with more noise than gold). When the dollar weakens, silver rallies harder.
-
Industrial demand creates divergence. When gold rallies on fear but industrial activity contracts, the gold/silver ratio spikes. This divergence is tradeable.
The Gold/Silver Ratio
The ratio of gold price to silver price is one of the oldest and most studied relationships in finance. Key levels:
| Ratio | Interpretation | Historical context |
|---|---|---|
| 40-50x | Silver relatively expensive | Typically during industrial booms or silver short squeezes |
| 60-70x | Fair value range | Long-run average oscillates here |
| 80-90x | Silver relatively cheap | Usually during recessions or flight-to-quality episodes |
| >100x | Extreme fear | Seen briefly in March 2020 (COVID panic: ratio hit ~125x) |
Our Gold/Silver Ratio strategy (S1.C.01) mean-reverts this ratio using an Ornstein-Uhlenbeck model: - Ratio > 85: long silver / short gold (ratio expected to compress) - Ratio < 55: long gold / short silver (ratio expected to expand) - Entry threshold: 1.5 standard deviations from OU mean - Exit: mean reversion to 0 z-score
The strategy has a 40-day decay halflife because the ratio can trend for weeks before reverting.
For OpB: Think of the ratio like a rubber band. It stretches (gold gets expensive relative to silver), and eventually snaps back. We trade the snap.
For OpA: The OU calibration uses a 252-day rolling window for the mean-reversion rate (kappa), long-run mean (theta), and diffusion (sigma). Entry/exit thresholds are in z-score space. The model is re-estimated weekly as part of the self-learning loop.
Key Drivers Summary
The QGTM system ingests these drivers through 8 FRED series and multiple alternative data feeds:
| Driver | FRED Series | Primary effect on gold | Used by strategies |
|---|---|---|---|
| Real yields (10Y) | DFII10 | Inverse: yields down = gold up | Real Rate Gold, Meta Labeller |
| Nominal yields | DGS10 | Indirect via breakeven | Breakeven Inflation Gold |
| Dollar index | DTWEXBGS | Inverse: dollar down = gold up | DXY Gold |
| Breakeven inflation | T10YIE | Positive: inflation up = gold up | Breakeven Inflation Gold |
| Fed funds rate | EFFR | Policy signal, affects real yields | Regime Classifier |
| VIX | VIXCLS | Positive in spikes: fear = gold up | VIX Haven, Meta Labeller |
| Money supply (M2) | M2SL | Positive: expansion = gold up | Central Bank Gold |
| Fed balance sheet | WALCL | Positive: expansion = gold up | Regime Classifier |
Beyond FRED, the system ingests: - CFTC Commitments of Traders (COT): Weekly speculative and commercial positioning in gold and silver futures. Released Fridays with a 3-day lag. Used by COT Precious and Hedging Pressure strategies. - COMEX warehouse stocks: Daily eligible and registered gold/silver inventory. Used by COMEX Warehouse strategy. - Shanghai Gold Exchange (SGE) withdrawals: Monthly data as a proxy for Chinese physical demand. Used by SGE Withdrawals strategy. - ETF fund flows: Daily creation/redemption data for GLD, IAU, SLV. Used by ETF Flow PM strategy. - LBMA AM/PM fix: The London gold fix prices, used by Fix Dislocation strategy to trade spot-vs-fix deviations.
Physical Markets: COMEX, LBMA, SGE
Gold and silver are unique among financial assets because large physical markets operate alongside the paper markets.
COMEX (CME Group, New York)
The primary futures exchange for gold (GC) and silver (SI). Key concepts:
- Front-month contract: The most liquid, actively traded contract. GC front-month trades 200,000+ contracts per day.
- Contango: When the futures price is above spot. Normal for gold because holding physical gold costs money (storage, insurance). The contango equals roughly the risk-free rate minus the convenience yield.
- Backwardation: When futures trade below spot. Rare in gold, signals extreme physical demand. Our Backwardation Stress strategy trades this.
- Roll yield: When you hold a futures position, you must roll from the expiring contract to the next one. In contango, rolling is a cost (sell cheap, buy expensive). In backwardation, rolling is a profit.
- Mini-gold (MGC): 1/10th the size of a standard GC contract. Allows more precise position sizing.
LBMA (London)
The London Bullion Market Association sets the benchmark gold and silver prices twice daily: - Gold AM Fix: 10:30 AM London time (5:30 AM ET) - Gold PM Fix: 3:00 PM London time (10:00 AM ET) - Silver Fix: 12:00 PM London time (7:00 AM ET)
These fixings are reference prices for physical transactions worldwide. Our Fix Dislocation strategy (S1.F.02) trades the gap between the fix price and the live spot price when it exceeds historical norms.
Shanghai Gold Exchange (SGE)
The world's largest physical gold exchange. Monthly withdrawal data is a leading indicator of Chinese consumer and central bank demand. When SGE withdrawals spike, it often precedes a gold rally as physical demand tightens the market.
Futures: GC, SI, MGC
Our system trades both ETFs and futures (when configured). Here is what matters:
| Contract | Underlying | Size | Tick | Daily ADV |
|---|---|---|---|---|
| GC | Gold | 100 troy oz (~$230K at $2,300/oz) | \(0.10 (\)10/tick) | 200K+ contracts |
| SI | Silver | 5,000 troy oz (~$145K at $29/oz) | \(0.005 (\)25/tick) | 50K+ contracts |
| MGC | Mini-gold | 10 troy oz (~$23K) | \(0.10 (\)1/tick) | 30K+ contracts |
Contango and roll yield example: If gold spot is $2,300 and the 2-month futures price is $2,310, the annualized contango is approximately:
This is the cost of carrying a long futures position. Our GC Term Structure strategy (S1.B.01) harvests roll yield by positioning on the part of the curve where carry is most favorable.Backwardation example: If spot is $2,300 and 2-month futures are $2,290, physical demand is strong enough that holders demand a premium for lending their gold. This is rare and bullish for spot gold.
Connecting This to the System
Every concept in this chapter maps directly to a strategy or data feed in QGTM:
| Concept | System component | Strategy/feed |
|---|---|---|
| Real yields drive gold | FRED DFII10 | Real Rate Gold (S1.A.01) |
| Dollar inverse to gold | FRED DTWEXBGS | DXY Gold (S1.A.02) |
| Gold/silver ratio mean-reverts | Feature engine z-score | Gold/Silver Ratio (S1.C.01) |
| COT positioning extremes | CFTC weekly data | COT Precious (S1.D.01) |
| Futures term structure | COMEX GC/SI curves | GC/SI Term Structure (S1.B.01/02) |
| London fix dislocation | LBMA AM/PM data | Fix Dislocation (S1.F.02) |
| Central bank buying | IMF IFS quarterly | Central Bank Gold (S1.A.05) |
| VIX fear spikes | FRED VIXCLS | VIX Haven (S1.A.04) |
| ETF fund flows | GLD/SLV creation/redemption | ETF Flow PM (S1.D.03) |
| Physical demand (China) | SGE monthly withdrawals | SGE Withdrawals (S1.D.05) |
For OpB: You now understand why we trade gold and silver and what drives their prices. The next chapter covers the specific instruments (GLD, SLV, miners) and when to use each.
For OpA: You now have the macro framework. Chapter 2 covers vehicle selection, expense structures, and the cost/capacity/tax tradeoffs between ETFs, futures, and FX.
Next: Chapter 2: PM Vehicles -- GLD vs IAU, miners vs metal, leveraged ETFs, and why vehicle choice matters.