Chapter 2: Precious Metals Vehicles
Reading time: 14 minutes | Difficulty: Beginner-Intermediate | Prerequisites: Chapter 1
Vehicle choice matters. The same gold view can be expressed through a physical ETF, a miner ETF, a leveraged product, a futures contract, or an FX pair. Each has different cost, capacity, tax treatment, and risk profile. This chapter covers all of them.
Gold ETFs: GLD vs IAU vs SGOL vs AAAU vs OUNZ
All five hold physical gold in vaults. The differences are cost, structure, and liquidity.
| Symbol | Name | Expense Ratio | ADV (shares) | Liquidity Tier | Notes |
|---|---|---|---|---|---|
| GLD | SPDR Gold Shares | 0.40% | 8,000,000 | Tier 1 | Largest gold ETF globally. Deepest options chain. Default vehicle for the system. |
| IAU | iShares Gold Trust | 0.25% | 5,000,000 | Tier 1 | Lower fee than GLD. Better for buy-and-hold legs. |
| SGOL | abrdn Physical Gold | 0.17% | 2,000,000 | Tier 2 | Lowest fee. Swiss vaults (Zurich). Good for longer-duration positions. |
| AAAU | Goldman Sachs Physical Gold | 0.18% | 400,000 | Tier 3 | Comparable to SGOL on cost. Lower liquidity limits sizing. |
| OUNZ | VanEck Merk Gold Trust | 0.25% | 300,000 | Tier 3 | Unique: allows physical delivery. Signal-only in the system. |
How the system uses them:
The Vehicle Router in the signal aggregator selects among gold ETFs based on the trade's expected holding period and size:
- Short-duration tactical trades (< 5 days): GLD. The tight bid-ask spread and deep options chain minimize execution cost.
- Medium-duration positions (5-60 days): IAU. The 15 bps expense savings over GLD compounds over weeks.
- Long-duration core positions (> 60 days): SGOL. The 23 bps savings over GLD matters when holding for months.
- Signal-only / thin liquidity: AAAU and OUNZ generate features and signals but are rarely traded directly due to Tier 3 liquidity.
For OpB: Think of GLD as the express lane (fast, costs a bit more), IAU as the regular lane (slightly slower fills, lower toll), and SGOL as the scenic route (cheapest, but you need patience).
For OpA: The routing decision is made post-aggregation in the VehicleRouter class. It considers ADV participation rate (capped at 10% of ADV per slice via VWAP), estimated hold time from the strategy's decay halflife, and the expense drag over that hold time.
Silver ETFs: SLV vs SIVR
| Symbol | Name | Expense Ratio | ADV | Liquidity Tier | Notes |
|---|---|---|---|---|---|
| SLV | iShares Silver Trust | 0.50% | 12,000,000 | Tier 1 | Dominant silver ETF. Deep options chain. |
| SIVR | abrdn Physical Silver | 0.30% | 600,000 | Tier 2 | Lower fee, lower liquidity. |
Silver ETFs are more expensive than gold ETFs because silver is heavier per unit of value (silver is ~\(29/oz vs gold at ~\)2,300/oz). Storing the same dollar value of silver requires 80x more vault space.
SLV's 0.50% expense ratio is the system's biggest drag on silver positions. For positions held longer than 30 days, the system routes to SIVR when order size is below 1% of SIVR's ADV (6,000 shares). Above that threshold, SLV's superior liquidity wins despite the higher fee.
The SLV options chain is the primary vehicle for our Vol Risk Premium (S1.E.01) and Skew Trades (S1.E.02) strategies. SLV implied volatility typically runs 5-8 points above realized vol, creating a persistent risk premium to harvest.
Gold Miners: GDX, GDXJ, NUGT, RING, GOAU
Miner ETFs do not hold physical gold. They hold stocks of gold mining companies. This creates a fundamentally different return profile:
| Symbol | Name | Expense Ratio | ADV | Tier | Beta to Gold |
|---|---|---|---|---|---|
| GDX | VanEck Gold Miners | 0.51% | 20,000,000 | Tier 1 | ~1.5-2.5x |
| GDXJ | VanEck Junior Gold Miners | 0.52% | 6,000,000 | Tier 1 | ~2.0-3.5x |
| NUGT | Direxion Gold Miners Bull 2X | 1.17% | 3,000,000 | Tier 2 | ~3.0-5.0x |
| RING | iShares Global Gold Miners | 0.39% | 100,000 | Tier 3 | ~1.5-2.5x |
| GOAU | US Global GO GOLD & PM Miners | 0.60% | 200,000 | Tier 3 | ~1.5-2.5x |
Why Miners Amplify Gold Moves
A gold miner's cost structure is mostly fixed (labor, energy, equipment). If gold is at $2,300 and all-in sustaining cost (AISC) is $1,200, the miner's margin is $1,100/oz. If gold rises 10% to $2,530, the margin jumps to $1,330/oz -- a 21% increase. This operating leverage is why GDX has 1.5-2.5x beta to gold.
The catch: beta decay. Miners underperform physical gold over long holding periods because of equity market risk (management, dilution, geopolitics, energy costs). GDX has lagged GLD over most 3-year rolling windows since 2006.
How the System Uses Miners
Our Miners vs Metal strategy (S1.C.03) exploits the mean-reverting relationship between GDX and GLD: - GDX/GLD ratio drops below historical mean: long miners, short metal (miners are cheap) - GDX/GLD ratio rises above historical mean: long metal, short miners (miners are expensive) - Decay halflife: 35 days
The regime classifier modulates miner allocation: in crisis regimes, the system shifts away from miners toward physical ETFs because miner equity risk compounds with gold downside.
Silver Miners: SIL, SILJ, SLVP
| Symbol | Name | Expense Ratio | ADV | Tier | Beta to Silver |
|---|---|---|---|---|---|
| SIL | Global X Silver Miners | 0.65% | 500,000 | Tier 2 | ~2.0-3.0x |
| SILJ | ETFMG Prime Junior Silver Miners | 0.69% | 400,000 | Tier 3 | ~2.5-4.0x |
| SLVP | iShares Global Silver & Metals Miners | 0.39% | 50,000 | Tier 3 | ~2.0-3.0x |
Silver miners carry the same operating leverage as gold miners, but with silver's higher volatility (25-35% annual). This makes SILJ one of the most volatile instruments in the universe -- annualized vol often exceeds 50%.
The system uses silver miners primarily as a high-conviction vehicle when the gold/silver ratio strategy signals silver outperformance. The sizing is scaled down by the Tier 2/3 liquidity constraints.
Leveraged ETFs: UGL and AGQ
| Symbol | Name | Leverage | Expense Ratio | ADV | Tier |
|---|---|---|---|---|---|
| UGL | ProShares Ultra Gold | 2x daily | 0.95% | 200,000 | Tier 3 |
| AGQ | ProShares Ultra Silver | 2x daily | 0.95% | 300,000 | Tier 3 |
Path Dependence: The Critical Risk
Leveraged ETFs reset daily. They deliver 2x the daily return, not 2x the cumulative return. Over time, volatility erodes returns through a mathematical effect called volatility drag:
Example with UGL (2x gold): - Day 1: Gold +2%. UGL: +4%. UGL NAV: $104. - Day 2: Gold -2%. UGL: -4%. UGL NAV: $104 * 0.96 = $99.84. - Gold net: +0.0%. UGL net: -0.16%.
Over a year, if gold is flat but volatile (15% annualized vol), UGL loses approximately:
Add the 0.95% expense ratio, and UGL costs 5.4% per year when gold is flat.
When the System Uses Leveraged Products
The Leveraged ETF Decay strategy (S1.C.04) exploits this drag by shorting leveraged ETFs against long physical positions. It is currently in shadow mode (not trading live) due to the borrow cost and availability constraints on shorting UGL and AGQ.
The system also uses UGL and AGQ tactically for short-duration, high-conviction directional trades where the 2x leverage is desired and holding period is < 5 days (minimizing path dependence). These trades are sized at 50% of normal to account for the 2x leverage.
Cross-Precious: GLTR, PPLT, PALL
| Symbol | Name | Underlying | Expense Ratio | ADV | Tier |
|---|---|---|---|---|---|
| GLTR | abrdn Physical PM Basket | Au, Ag, Pt, Pd mix | 0.60% | 150,000 | Tier 3 |
| PPLT | abrdn Physical Platinum | Platinum | 0.60% | 150,000 | Tier 3 |
| PALL | abrdn Physical Palladium | Palladium | 0.60% | 30,000 | Tier 3 |
These are used in the Gold/Platinum strategy (S1.C.02), which pairs gold against platinum. Platinum's industrial demand profile (automotive catalytic converters) creates divergence from gold during manufacturing cycles.
PALL (palladium) has extremely thin liquidity (30K shares/day). It is signal-only in the system -- the model generates features from palladium price data, but never trades PALL directly.
Futures vs ETFs vs FX: When to Use Each
| Dimension | ETFs (GLD, SLV) | Futures (GC, SI) | FX (XAUUSD) |
|---|---|---|---|
| Capital efficiency | 100% margin (cash account) | ~5-10% margin | ~2-5% margin |
| Expense drag | 0.17-0.50% annually | Zero (but roll cost) | Zero |
| Tax treatment | LTCG if held > 1 year (but collectibles rate 28%) | 60/40 rule (60% LTCG, 40% STCG) | 60/40 rule |
| Trading hours | 9:30-4 ET | Nearly 24/5 | 24/5 |
| Options available | Deep chain (GLD, SLV) | Deep chain (GC, SI) | Limited |
| Capacity | Excellent (GLD 8M shares/day) | Excellent (GC 200K contracts/day) | Effectively unlimited |
| Operational complexity | Low (equity account) | Medium (futures account, rolls) | Medium (FX account, swaps) |
| System default | Yes (Alpaca equity account) | Optional (requires CME connection) | Optional (requires FX feed) |
The system's default is ETFs through Alpaca because: 1. Single account handles all 20 ETF symbols 2. No roll management required 3. Options strategies use GLD/SLV options natively 4. Commission-free at Alpaca
Futures and FX are configured as alternatives when capital efficiency or tax treatment warrants the operational overhead.
Options on GLD/SLV: The Vol Risk Premium
GLD and SLV have among the most liquid options chains of any ETF. This is critical for four of our strategies:
| Strategy | What it does | Vehicle |
|---|---|---|
| Vol Risk Premium (S1.E.01) | Sells straddles to harvest implied > realized spread | GLD/SLV options |
| Skew Trades (S1.E.02) | Trades put/call skew dislocations | GLD options |
| Gamma Scalp (S1.E.03) | Long gamma, delta-hedge around events | GLD options |
| Vol Term Structure (S1.E.04) | Calendar spreads on vol term structure shape | GLD/SLV options |
The vol risk premium in gold options is persistent because: - Institutional hedgers (central banks, miners) are natural put buyers, pushing implied vol above realized - Retail traders are natural call buyers during gold rallies, adding to the premium - The spread between implied and realized vol on GLD averages 3-5 vol points over rolling 30-day windows
Our Tail Hedge strategy (S1.H.01) takes the other side: it is a permanent buyer of OTM put spreads on GLD, funded by short vol in calm regimes. This provides portfolio insurance against extreme drawdowns.
Vehicle Selection Decision Tree
Strategy generates gold LONG signal
|
+-- Holding period < 5 days?
| |
| +-- YES: Trade GLD (tight spreads, fast execution)
| +-- YES + high conviction: Consider UGL (2x, accept drag)
|
+-- Holding period 5-60 days?
| |
| +-- Order < 1% IAU ADV: Trade IAU (lower expense)
| +-- Order > 1% IAU ADV: Trade GLD (better liquidity)
|
+-- Holding period > 60 days?
| |
| +-- Order < 1% SGOL ADV: Trade SGOL (lowest expense: 0.17%)
| +-- Order > 1% SGOL ADV: Trade IAU (0.25%)
|
+-- Want leveraged gold exposure?
| |
| +-- Short duration (< 5 days): UGL
| +-- Long duration: Use GDX miners (natural leverage, no drag)
|
+-- Want to express via options?
|
+-- Vol sell: GLD straddle/strangle
+-- Directional: GLD call spread
+-- Protection: GLD put spread (tail hedge)
For OpB: The key takeaway is that GLD is the system's workhorse for gold, SLV for silver, and GDX when you want amplified exposure. The system automatically picks the best vehicle based on trade duration and size.
For OpA: The VehicleRouter integrates expense ratio, ADV participation rate, estimated holding period, and available margin to select the optimal instrument. Override is available via the strategy config for specific vehicle preferences.
Next: Chapter 3: Our PM Universe -- the complete 20-instrument universe, sector map, and liquidity tiers from pm_universe.py.