Skip to content

Chapter 3: Our Precious Metals Universe

Reading time: 14 minutes | Difficulty: Beginner-Intermediate | Prerequisites: Chapters 1-2

Universe = the exact set of instruments the system is allowed to touch. Nothing outside this list gets traded. Nothing inside this list gets ignored. This chapter is the canonical reference, derived directly from qgtm_core/pm_universe.py.

The 20 PM Instruments We Trade

The QGTM system trades 20 precious metals instruments across 7 sectors. Each instrument has a sector classification, liquidity tier, and defined role in the portfolio.

 +======================================================================+
 |  GOLD ETFs (5)                                                       |
 |  Physical gold held in vaults -- core directional exposure            |
 +----------------------------------------------------------------------+
 |  GLD   SPDR Gold Shares              0.40% ER    8M ADV    Tier 1   |
 |  IAU   iShares Gold Trust            0.25% ER    5M ADV    Tier 1   |
 |  SGOL  abrdn Physical Gold           0.17% ER    2M ADV    Tier 2   |
 |  AAAU  Goldman Sachs Physical Gold   0.18% ER    400K ADV  Tier 3   |
 |  OUNZ  VanEck Merk Gold Trust        0.25% ER    300K ADV  Tier 3   |
 +======================================================================+
 |  SILVER ETFs (2)                                                     |
 |  Physical silver held in vaults                                      |
 +----------------------------------------------------------------------+
 |  SLV   iShares Silver Trust          0.50% ER    12M ADV   Tier 1   |
 |  SIVR  abrdn Physical Silver         0.30% ER    600K ADV  Tier 2   |
 +======================================================================+
 |  CROSS-PRECIOUS (3)                                                  |
 |  Platinum, palladium, and baskets                                    |
 +----------------------------------------------------------------------+
 |  GLTR  abrdn Physical PM Basket      0.60% ER    150K ADV  Tier 3   |
 |  PPLT  abrdn Physical Platinum       0.60% ER    150K ADV  Tier 3   |
 |  PALL  abrdn Physical Palladium      0.60% ER    30K ADV   Tier 3   |
 +======================================================================+
 |  GOLD MINERS (5)                                                     |
 |  Equity exposure to gold mining companies                            |
 +----------------------------------------------------------------------+
 |  GDX   VanEck Gold Miners            0.51% ER    20M ADV   Tier 1   |
 |  GDXJ  VanEck Junior Gold Miners     0.52% ER    6M ADV    Tier 1   |
 |  NUGT  Direxion Gold Miners Bull 2X  1.17% ER    3M ADV    Tier 2   |
 |  RING  iShares Global Gold Miners    0.39% ER    100K ADV  Tier 3   |
 |  GOAU  US Global GO GOLD & PM Miners 0.60% ER    200K ADV  Tier 3   |
 +======================================================================+
 |  SILVER MINERS (3)                                                   |
 |  Equity exposure to silver mining companies                          |
 +----------------------------------------------------------------------+
 |  SIL   Global X Silver Miners        0.65% ER    500K ADV  Tier 2   |
 |  SILJ  ETFMG Prime Junior Silver     0.69% ER    400K ADV  Tier 3   |
 |  SLVP  iShares Global Silver Miners  0.39% ER    50K ADV   Tier 3   |
 +======================================================================+
 |  LEVERAGED (2)                                                       |
 |  2x daily rebalanced -- tactical use only                            |
 +----------------------------------------------------------------------+
 |  UGL   ProShares Ultra Gold          0.95% ER    200K ADV  Tier 3   |
 |  AGQ   ProShares Ultra Silver        0.95% ER    300K ADV  Tier 3   |
 +======================================================================+

Total tradeable ETFs: 20 | Gold: 10 | Silver: 5 | Cross: 3 | Leveraged: 2

Sectors: What Each Group Does

The sector classification from PMSector determines how the compliance gateway applies concentration limits and how the regime allocator distributes exposure.

Gold ETFs (5 instruments)

The core of the portfolio. In a typical allocation, gold ETFs represent 30-50% of gross exposure. The five instruments provide optionality on vehicle selection (Chapter 2) without changing the underlying view.

Key relationship: All five track the same thing -- the London gold spot price. Correlation among them is 0.99+. The system never holds long GLD and short IAU (that would be an expense-ratio arbitrage too small to matter after transaction costs).

Silver ETFs (2 instruments)

Silver carries higher vol than gold (25-35% vs 15-20% annualized), so it receives smaller position sizes for the same risk budget. SLV's 12M share ADV makes it the most liquid instrument in the entire universe by raw volume.

Cross-Precious (3 instruments)

GLTR, PPLT, and PALL are niche. Their role is primarily as feature inputs: platinum and palladium prices inform the Gold/Platinum pairs strategy (S1.C.02) and the Cross Carry strategy (S1.B.03). Direct trading is limited by Tier 3 liquidity.

Gold Miners (5 instruments)

GDX is the workhorse miner ETF. It provides leveraged exposure to gold (1.5-2.5x beta) without the path-dependence drag of leveraged ETFs. GDXJ (junior miners) has higher beta (2.0-3.5x) but also higher idiosyncratic risk.

NUGT (2x gold miners) combines miner beta with daily leverage. At 3-5x effective beta to gold, it is the highest-beta instrument in the universe. Used only for short-duration, high-conviction trades.

Silver Miners (3 instruments)

SIL and SILJ are the silver equivalents of GDX and GDXJ. Lower liquidity than gold miners. SLVP at 50K ADV is essentially signal-only.

Leveraged (2 instruments)

UGL and AGQ. Tactical only, never held longer than 5 days due to volatility drag (see Chapter 2). The Leveraged ETF Decay strategy (S1.C.04) is designed to exploit the structural drag in these products.

Liquidity Tiers

The system classifies every instrument into three liquidity tiers based on Average Daily Volume (ADV). The tier determines position sizing limits and execution algorithm selection.

Tier 1: Full Sizing (ADV > 5M shares)

  GLD    8,000,000    Gold ETF
  IAU    5,000,000    Gold ETF
  SLV   12,000,000    Silver ETF
  GDX   20,000,000    Gold Miners
  GDXJ   6,000,000    Junior Gold Miners
  IEF    5,000,000    Cross-asset: Treasury signal
  IBIT  30,000,000    Cross-asset: Bitcoin signal

Tier 1 instruments can be traded at full strategy weight. VWAP execution with up to 10% participation rate means we can move \(500K-\)1M in a single rebalance without market impact.

Tier 2: Reduced Sizing (ADV 500K-5M)

  SGOL   2,000,000    Gold ETF
  SIVR     600,000    Silver ETF
  SIL      500,000    Silver Miners
  NUGT   3,000,000    2x Gold Miners
  AGQ      300,000    2x Silver
  UUP    3,000,000    Cross-asset: Dollar signal
  TIP    4,000,000    Cross-asset: TIPS signal
  CPER     200,000    Cross-asset: Copper signal (borderline)

Tier 2 instruments are sized at 50-75% of full weight. Execution switches to limit orders with longer fill windows.

Tier 3: Signal-Only or Minimal Sizing (ADV < 500K)

  AAAU     400,000    Gold ETF
  OUNZ     300,000    Gold ETF
  GLTR     150,000    PM Basket
  PPLT     150,000    Platinum
  PALL      30,000    Palladium
  RING     100,000    Gold Miners
  GOAU     200,000    Gold Miners
  SILJ     400,000    Junior Silver Miners
  SLVP      50,000    Silver Miners
  UGL      200,000    2x Gold

Tier 3 instruments contribute price data and features to the system but receive minimal or zero allocation. Trading them at meaningful size would exceed 10% of daily volume.

Cross-Asset Inputs: Not Traded, But Critical

Five instruments provide signal inputs without being directly traded:

Symbol Name ADV What it tells us
UUP Invesco DB US Dollar Index Bullish 3,000,000 Real-time dollar strength (DXY proxy for ETF accounts)
TIP iShares TIPS Bond ETF 4,000,000 Real interest rate expectations
IEF iShares 7-10 Year Treasury Bond 5,000,000 Nominal rate environment
IBIT iShares Bitcoin Trust 30,000,000 Digital gold narrative / risk appetite
CPER United States Copper Index Fund 200,000 Industrial demand proxy (copper-gold ratio)

These inputs feed the feature engine. Examples: - The TIP/IEF ratio approximates breakeven inflation, used by the Breakeven Inflation Gold strategy. - The CPER/GLD ratio (copper-gold) is a leading indicator of risk appetite. Rising copper-gold = risk on, falling = risk off. The Regime Classifier uses this as an input feature. - IBIT correlation with gold has been unstable (ranging from -0.3 to +0.5 over rolling 60-day windows). The Meta Labeller ingests this as a feature to detect regime shifts in the "digital gold" narrative.

FRED Macro Series: The Invisible Inputs

Eight FRED time series are fetched daily at 6:00 AM ET. They are never "traded" but drive half the strategies:

Label FRED ID Frequency Used by
Real yield 10Y DFII10 Daily Real Rate Gold, Meta Labeller
Nominal 10Y DGS10 Daily Breakeven Inflation Gold
Trade-weighted dollar DTWEXBGS Daily DXY Gold
Breakeven 10Y T10YIE Daily Breakeven Inflation Gold
Fed funds rate EFFR Daily Regime Classifier
VIX VIXCLS Daily VIX Haven, Meta Labeller
M2 money supply M2SL Weekly Central Bank Gold
Fed balance sheet WALCL Weekly Regime Classifier

Staleness protection: If any FRED series is not updated for 7 days (FEATURE_STALENESS_MAX_DAYS in constants.py), the features derived from it are marked stale and excluded from signal generation. The affected strategies revert to their last valid signal rather than trading on stale data.

Why We Do Not Trade Energy or Agriculture Anymore

The original QGTM system traded 42 ETFs across 6 sectors including energy (USO, UNG, BNO) and agriculture (DBA, CORN, WEAT). We removed them because:

1. Diluted edge. Our strongest signals come from the real-rate/gold relationship (R-squared > 0.70). No comparable structural relationship exists for crude oil or wheat. The strategies we ran on energy and agriculture had lower Sharpe ratios and higher decay rates.

2. Data pollution. Mixing gold signals with oil signals in the same aggregator reduced the meta-labeller's accuracy. Gold strategies produce better signals when they are not competing for allocation with strategies driven by completely different fundamentals.

3. Operational focus. Twenty instruments are enough to run 29 strategies with full capacity. Adding 22 more instruments would require proportionally more data feeds, more feature engineering, more backtest maintenance, and more monitoring -- without proportional return improvement.

4. Clear positioning. A gold/silver specialist system has a clearer value proposition for subscribers than a generalist commodity platform. Edge comes from depth, not breadth.

The energy and agriculture code still exists in qgtm_core/universe.py (the legacy broad universe module) for historical backtesting purposes, but the live system uses only qgtm_core/pm_universe.py.

Universe Summary (from pm_universe.universe_summary())

  Total ETF symbols:          20
  Total symbols (all):        30 (includes futures, FX, cross-asset)
  ──────────────────────────────
  Gold ETFs:                   5
  Silver ETFs:                 2
  Cross-precious:              3
  Gold miners:                 5
  Silver miners:               3
  Leveraged:                   2
  Futures:                     3  (GC, SI, MGC)
  FX pairs:                    2  (XAUUSD, XAGUSD)
  Cross-asset inputs:          5  (UUP, TIP, IEF, IBIT, CPER)
  FRED macro series:           8
  ──────────────────────────────
  Tier 1 (full sizing):       ~8 instruments
  Tier 2 (reduced sizing):    ~8 instruments
  Tier 3 (signal-only):       ~14 instruments

Correlation Structure

Within the PM universe, correlations are high but not uniform. This matters for risk management:

Pair Typical Correlation Why
GLD-IAU 0.99+ Same underlying (gold spot)
GLD-SLV 0.75-0.85 Both precious metals, but silver has industrial component
GLD-GDX 0.70-0.85 Miners leverage gold, but have equity market risk
GDX-GDXJ 0.90-0.95 Both gold miners, juniors slightly more volatile
SLV-SIL 0.65-0.80 Silver metal vs silver miners (equity risk dilutes)
GLD-UUP -0.40 to -0.60 Gold inverse to dollar (key trading relationship)
GLD-TIP 0.20-0.50 Gold correlates with inflation expectations
GLD-IBIT -0.30 to +0.50 Unstable: varies by regime and narrative

The compliance gateway uses these correlations to ensure the portfolio does not accidentally take a 5x concentrated gold bet by being long GLD + IAU + GDX + NUGT + UGL simultaneously (which, despite being 5 different tickers, is effectively one massive gold position).

The 40% sector concentration limit applies at the PMSector level: all gold ETFs (GLD, IAU, SGOL, AAAU, OUNZ) share the GOLD_ETF sector and are summed together for the limit check.

For OpB: This is the complete list of everything the system can trade. Twenty instruments, 7 sectors, 3 liquidity tiers. If it is not on this list, the system cannot touch it.

For OpA: The canonical source of truth for the universe is qgtm_core/pm_universe.py. Any change to instruments, sectors, or liquidity thresholds must be made there. The constants _TIER1_THRESHOLD (5M) and _TIER2_THRESHOLD (500K) control the tier boundaries.

Next: Chapter 4: How Algo Trading Works -- the full pipeline from data to execution, with a worked example using the Real Rate Gold strategy.