14. Commodity Investing
Teaching rewrite of commodities as an investment: who invests and why, commodity indices and the roll-yield debate, total return swaps, exchange traded products (ETCs, ETFs, ETNs), and structured notes. Original wording; worked numbers recomputed. Not investment advice.
Introduction
Earlier lessons looked at commodities through the eyes of producers and consumers hedging real exposures. Here the user is an investor who has no physical exposure but wants commodity returns — for diversification, inflation protection, or a view.
Because few investors want to store crude or cattle, almost all commodity investing runs through futures, swaps, and securities wrapped around them. That creates a subtle difference between the spot price you see quoted and the return you actually earn.
This lesson covers:
- Who invests and over what horizon; the main instruments
- Why invest: diversification, inflation, USD, risk premia
- Commodity indices (S&P GSCI): construction, spot vs excess vs total return
- The roll-yield myth, worked through in numbers
- Total return swaps, ETCs / ETFs / ETNs
- Structured products: capital-protected, basket, income, reverse convertible, autocallable, outperformance, worst-of
14.1 Investors and instruments
| Investor | Traits | Typical horizon |
|---|---|---|
| Hedge funds, bank desks | “Leveraged” accounts; tactical, relative value | Under 1 year |
| Commodity trading advisors (CTAs) | Futures-based, often trend-following, multi-asset | Short–medium |
| Private bank / retail | Mutual funds, ETPs, structured notes | 1–5 years |
| Asset managers | Funds, index tracking | 1–5 years |
| Real money (pensions, insurers) | Can’t borrow; buy and hold | 5+ years |
| Sovereign wealth funds | Strategic diversification | 5+ years |
By horizon:
- Under 1 year: tactical trades (e.g. mean reversion), relative value (platinum vs palladium, gold vs silver, or looser pairs like crude vs gold), index tracking. Blow-ups happen — e.g. Amaranth (lesson 3).
- 1–5 years: diversification, growth, harvesting volatility with options, ETPs, structured notes.
- 5+ years: strategic allocation to a broad index.
| Instrument | What it is |
|---|---|
| Total return swap (TRS) | OTC swap: receive an index’s total return, pay a money-market rate plus a fee |
| Exchange traded product (ETP) | Listed security on one commodity or an index: ETC (physically backed), ETF (derivatives-based, bankruptcy-remote), ETN (unsecured bank debt) |
| Medium term note (MTN) / structured product | Bank-issued note with embedded derivatives, mainly for retail and private banks |
Market size (Barclays): commodity assets under management grew from ~USD 15 bn (2001) to a peak of ~USD 418 bn (2012), then ~USD 311 bn (2017). TRS and ETPs dominate; structured notes are small. Precious metals are the most popular sector, then energy, agriculture, base metals.
14.2 Why invest in commodities?
Return and diversification
The IMF all-commodity index (Jan 1992 = 100) peaked above 400 in July 2008 — versus 309 for the S&P 500 on the same base — then halved by January 2009. Since then: 388 (2011), 174 (2016), 265 (2018), 168 (April 2020). High return, high volatility.
Two-asset portfolio volatility:
σ_p = √( w₁²σ₁² + w₂²σ₂² + 2 w₁ w₂ ρ σ₁ σ₂ )
Unless ρ = +1, portfolio volatility is below the weighted average of the parts.
| Period | Correlation with equities | Bonds | T-bills |
|---|---|---|---|
| 1970–2005 | −0.28 | −0.10 | −0.01 |
| 2006–2016 | +0.40 | +0.28 | +0.10 |
Worked example: diversification benefit
80% equities (vol 16%) + 20% commodities (vol 25%):
- ρ = −0.28 → σ_p = √(0.8²·0.16² + 0.2²·0.25² + 2·0.8·0.2·(−0.28)·0.16·0.25) ≈ 12.4%
- ρ = +0.40 → σ_p ≈ 15.5%
- All equities → 16.0%
The benefit shrank as correlation turned positive.
Why correlations shift (Morgan Stanley, 2009): near full capacity (late cycle), scarce commodities rise while squeezed margins hurt equities → negative correlation. With slack (early cycle / recession), demand moves both together → positive correlation.
| Low inflation | High inflation | |
|---|---|---|
| High growth | Both can do well | Commodities up, equities squeezed |
| Low growth | Both weak | Commodities hold up better |
Inflation hedge
Commodities correlate with realised inflation (partly because they are in inflation indices). Real return ≈ how much more you can consume later: a 2% coupon with 3% inflation → 1.02 ÷ 1.03 − 1 ≈ −0.97% in real terms.
US dollar hedge
A weaker USD makes dollar-priced commodities cheaper abroad → more demand → higher USD prices. For a non-US investor holding USD assets, commodities can offset currency losses.
Risk premia strategies
Systematic strategies that earn an excess return for bearing a known risk (Barclays, 2011):
| Strategy | Trade | Risk |
|---|---|---|
| Curve (term premium) | Long deferred, short nearby futures | Curve moves against you |
| Value | Long cheap, short dear (e.g. in steep backwardation: sell nearby, buy deferred) | Convergence takes too long |
| Momentum | Long recent winners, short losers | Trend reversal |
| Arbitrage | Positions that should converge | Model error, funding, liquidity |
| Volatility | Sell implied vol (usually above realised) | Vol spikes |
Checkpoint 1: investors and rationale
Checkpoint — sections 14.1 and 14.2
1 / 5Which product exposes the investor to the issuing bank’s credit risk?
Pick an answer to see if you are right.
14.3 Commodity indices
The S&P GSCI (launched 1991) was the first major investable commodity index — a beta index giving broad market exposure.
| Feature | S&P GSCI |
|---|---|
| Constituents | 24: 6 energy, 5 industrial metals, 8 agriculture, 3 livestock, 2 precious metals |
| Weights | World production value (5-year average), reviewed annually → energy > 60% |
| Instruments | Front-month futures, not physical |
| Roll | 5th–9th business day of the month, 20% a day |
| Sub-indices | Energy, non-energy, industrial metals, precious metals, agriculture, livestock… |
(The Bloomberg Commodity Index, BCOM, is the main alternative: it uses production and liquidity, caps sectors at about a third, and rolls on the 6th–10th business days.)
Three versions of the index
| Version | Includes |
|---|---|
| Spot | Price changes of the futures held |
| Excess return (ER) | Spot + gains/losses from rolling (“roll yield”) — what an unfunded futures position earns |
| Total return (TR) | ER + interest on fully collateralised positions (e.g. USD 100,000 exposure: 5,000 margin, 95,000 in T-bills) |
TR ≠ ER + T-bill rate exactly, because interest is reinvested and collateral changes with the futures value.
Index evolution
- Beta — passive, long-only (S&P GSCI)
- Enhanced beta — different constituents or weights, long/short, deferred contracts, different roll dates (pre-/post-roll), different use of cash
- Active alpha — proprietary strategies: momentum, liquidity choices, volatility selling
14.4 The roll-yield myth
Conventional view: “roll yield” is the gain or loss on the day you roll — sell expiring at 100, buy next at 90 in backwardation = +10; the reverse in contango = −10.
Revised view (Campbell & Co., 2014): rolling itself makes no P&L — you sell and buy at market. Roll yield is the net benefit or cost of owning the asset (its carry):
- Futures return = spot return + roll yield
- Roll yield = futures return − spot return
Worked example: gold in contango
Spot rises 5 a day from 1,000. Financing 3%, lease 1% → net carry 2% a year, so futures sit above spot (contango). F = S × (1 + 0.02 × days/360).
| Day | Spot | Future #1 (expires day 5) | Future #2 (34 days left on day 1) |
|---|---|---|---|
| 1 | 1,000 | 1,000.22 (buy) | 1,001.89 |
| 2 | 1,005 | 1,005.17 | 1,006.84 |
| 3 | 1,010 | 1,010.11 | 1,011.79 |
| 4 | 1,015 | 1,015.05 (sell) | 1,016.75 (buy) |
| 5 | 1,020 | 1,020 (= spot) | 1,021.70 |
| … | … | … | |
| 10 | 1,045 | 1,046.45 (value) |
- Future #1 profit (days 1→4): 1,015.05 − 1,000.22 = 14.83
- Future #2 profit (days 4→10): 1,046.45 − 1,016.75 = 29.70
- Futures return = 44.53; spot return = 45.00
- Roll yield = 44.53 − 45.00 = −0.47
The conventional “roll loss” on day 4 is 1,015.05 − 1,016.75 = −1.70. Subtracting that from the spot gain implies 43.30 — wrong; the actual futures P&L was 44.53.
Reconciling the two: roll yield = basis return + cumulative roll adjustment.
- Basis day 1: 1,000.22 − 1,000 = 0.22; basis day 10: 1,046.45 − 1,045 = 1.45 → basis return 1.23
- Roll adjustment = −1.70
- 1.23 − 1.70 = −0.47 ✓
Because the basis is bounded by the curve’s shape, over long periods the cumulative roll adjustment dominates — so the old rule of thumb survives:
- Contango → longs underperform spot (negative carry)
- Backwardation → longs outperform spot (positive carry). Swapping the rates (financing 1%, lease 3%) gives a futures return of 45.47 and roll yield +0.47.
The source quotes the futures return as both 44.53 and 44.54, the roll yield as both −0.46 and −0.47, and the second future’s roll price once as “1,106.75”. The consistent figures are 44.53, −0.47, and 1,016.75.
Checkpoint 2: indices and roll yield
Checkpoint — sections 14.3 and 14.4
1 / 5Why is the S&P GSCI dominated by energy?
Pick an answer to see if you are right.
14.5 Total return swaps
The investor receives the index return and pays a T-bill rate plus a fee — unfunded (no cash up front). Investors that can’t take unlimited leverage use a funded version (a note).
flowchart LR
I[Investor] -- T-bill interest + fee --> B[Bank]
B -- Index total return --> I
B -- buys futures in index weights --> F[Futures markets]
Worked example: first monthly period
Notional USD 200 m, S&P GSCI TR from 2,922.42 to 2,955.33, 31 days, fee 0.25% (act/365), T-bill 0.5%.
| Leg | Calculation | USD |
|---|---|---|
| Index amount | 200 m × (2,955.33 − 2,922.42) ÷ 2,922.42 | +2,252,243 |
| Fee | 200 m × 0.25% × 31/365 | −42,465.75 |
| T-bill leg | 200 m × [(1 − 91/360 × 0.5%)^(−31/91) − 1] | −86,184 |
| Net to investor | 2,123,593.25 |
Next period the floating leg uses an adjusted notional of 200 m × 2,955.33 ÷ 2,922.42 = 202,252,243, mimicking the growth of a real holding.
The source’s term sheet states a notional of USD 100 million, but its worked example uses USD 200 million. The figures above use 200 million throughout.
The bank hedges by buying futures in index weights; the 0.25% covers hedging costs and profit. Investor risks: index falls and bank credit (usually unsecured). Variants: sub-indices, zero-coupon (one payment at maturity), excess-return swaps (investor pays only the fee).
14.6 Exchange traded products
| ETC | ETF | ETN | |
|---|---|---|---|
| Backing | Physical commodity in a vault | Derivatives (futures/swaps) with collateral | Unsecured issuer debt |
| Issuer | Bankruptcy-remote SPV | Bankruptcy-remote SPV | Bank (not remote) |
| Size | Varies with creation/redemption | Varies | Fixed issue |
| Main extra risk | Fees erode entitlement | Swap counterparty, termination, leverage path-dependence | Issuer credit |
ETC: Gold Bullion Securities (2004)
Jersey SPV (WisdomTree), physically backed by allocated LBMA good-delivery gold at HSBC London; trustee holds the gold for investors; fee 0.40%; USD, unhedged.
An SPV is a “robot firm”: one purpose, fixed rules, no staff, outsourced administration, designed so it can’t practically go bankrupt.
Creation/redemption: authorised participants (APs) pay cash → the issuer buys gold → new securities to the AP, who sells them and makes markets. Redemption runs in reverse (even for physical gold).
Fees are paid in gold: each security’s entitlement shrinks daily.
- 1 July 2005: gold 427.30; entitlement 99.550959% of 0.1 oz → 0.0995510 oz × 427.30 ≈ USD 42.54
- 4,000 securities ≈ USD 170,160 (≈ 398 oz — about one 400 oz bar)
- Next day: 99.550959% − 0.40%/365 = 99.549863%
- Mid-2020: entitlement 93.599%, gold 1,720.21 → USD 161.01 per security
ETF: 2× daily leveraged gold
References the Bloomberg Gold Subindex TR (futures-based), so the roll-yield issues apply. Fees 0.98% (issuer) + 1.30% (swap counterparty) + 0.05% (index).
Structure: investor cash → issuer → swap counterparty → segregated collateral account (third party) supporting a total return swap with 2× notional.
Daily valuation: P_t = P_(t−1) × [1 + capital adjustment + leverage × index % change]
Example: 50.00 × [1 − 0.0001 + 2 × 0.05] = 54.995 — a 5% index rise gives ~10%. A 50% one-day index fall wipes out the security (losses can’t exceed the investment).
Daily reset = path dependence: index +10% then −10% → index × 0.99; the 2× product × 1.20 × 0.80 = 0.96 — it loses more than twice the index’s 1%.
Risks: counterparty (mitigated by collateral); the counterparty may terminate if it can’t hedge — WisdomTree closed several leveraged oil ETPs in 2020 after crude briefly went negative.
ETN: Barclays gold 3-month index note
USD 25 notes, 30-year maturity, NYSE Arca; tracks a Barclays index on gold futures 3 months out, plus T-bill interest; unsecured — no principal protection; issuer call and investor put.
Value = previous value × (today’s index ÷ yesterday’s index). Index +5% then +2% → 25 × 1.05 × 1.02 = 26.775. Market price can differ (liquidity, issuer downgrade).
Checkpoint 3: swaps and ETPs
Checkpoint — sections 14.5 and 14.6
1 / 5TRS: notional 50 m, index 1,000 → 1,020. Index amount?
Pick an answer to see if you are right.
14.7 Structured products
A structured product offers a risk-return profile investors can’t easily build themselves: higher participation, limited downside, or a way to be short. Modern ones are often model-based, so not easily reverse-engineered. All carry issuer credit risk — “capital protection” is not a guarantee.
Capital-protected note
= zero-coupon bond + call option.
5-year gold note, USD 100,000; zero rate 5%:
| Step | Value |
|---|---|
| ZCB to return 100,000 | 100,000 ÷ 1.05⁵ = 78,353 |
| Left for options | 21,647 |
| 5-yr ATM call (spot 1,500; vol 25%; rate 5%; lease 1%) | 433 → 433 ÷ 1,500 = 28.87% of notional → 28,867 |
| Participation | 21,647 ÷ 28,867 ≈ 75% |
Payout = 100% + 75% × max(0, gold return). Gold +10% → 107,500; flat or down → 100,000.
Levers for a better participation rate: OTM strike, longer maturity, partial protection, lower-vol commodity, capped upside (call spread or barrier), drop protection at very low prices, Asian or correlation-dependent options, sell above par.
What’s special about commodities
- Forward curve: a European call’s moneyness depends on the forward, not spot. With rates 5% and lease 1%, the 5-year forward is 1,500 × e^(0.04×5) ≈ 1,832 — the “ATM” call is deep ITM. Reverse the rates (1% rates, 5% lease) and the forward is 1,228, the call is OTM, and it costs about 174.
- Volatility term structure often slopes down (near-term uncertainty) — cheaper long-dated options.
- Low correlation between commodities → cheaper basket options.
The source computes a 187% participation rate by pricing the cheaper call with 1% rates but still funding the zero-coupon bond at 5%. With 1% rates, the bond costs 100,000 ÷ 1.01⁵ ≈ 95,147, leaving only 4,853 for options: 4,853 ÷ (174/1,500 × 100,000) ≈ 42%. The lesson stands — the forward curve matters — but rates hit both legs.
Basket note
4-year, capital-protected, 100% participation in an equally weighted WTI + cocoa basket (initial 50 and 2,400). Basket value = Σ wₖ × finalₖ ÷ initialₖ.
| Scenario | WTI | Cocoa | Basket | Payout |
|---|---|---|---|---|
| Both rise | 70 | 2,500 | 0.5 × 1.40 + 0.5 × 1.0417 = 1.2208 | 122% |
| Both fall | 40 | 2,300 | 0.8792 | 100% |
| Diverge | 70 | 2,300 | 1.1792 | 118% |
The buyer of a basket call is long correlation: lower correlation → cheaper option → higher participation.
Income note: Baltic Dry Index
3-year capital-protected note; each year pays min(5%, max(0, BDI_t ÷ BDI_0 − 1)) — a strip of annual call spreads (100%/105%). BDI +10% → 5%; +3% → 3%; −8% → 0%.
Reverse convertible
12-month, 8% coupon, barrier 80% on crude. Investor = deposit + short down-and-in put (strike 100%).
- Barrier never touched → 108%
- Barrier touched → 100% + 8% − max(0, 1 − final ÷ initial)
Example: initial 60, barrier 48. Touches 47, ends at 45 → 108% − 25% = 83%. Never touched, ends at 55 → 108%.
The source writes the loss term as max(0, P_final ÷ P_initial − 1), which would penalise rising prices. The short put loses when prices fall: the term is max(0, 1 − P_final ÷ P_initial).
Worst-of autocallable
6-month note on Brent and silver; knock-in barrier 85%.
- At 3 months: if both ≥ 100% → redeems at par early
- Coupons at 3 and 6 months: principal × N × 3% (N = assets above their initial level)
- At maturity: if the worst asset ≥ 85% → 100%; else worst final ÷ worst initial
Building blocks: long digitals (coupons), short down-and-in put on the worst asset (funds it), an up-and-out feature for early redemption.
Example: at 3 months Brent 105%, silver 98% → no call; coupon 1 × 3% = 3%. At 6 months Brent 90%, silver 80% → coupon 0; silver < 85% → redemption 80%.
Outperformance note
5-year: 100% + 300% × max(oil return − gold return, 0); WTI 50, gold 1,500.
| Scenario | Oil return | Gold return | Payout |
|---|---|---|---|
| Both rise (70; 1,550) | 1.40 | 1.0333 | 100% + 3 × 0.3667 = 210% |
| Both fall (45; 1,300) | 0.90 | 0.8667 | 100% + 3 × 0.0333 = 110% |
| Diverge (40; 1,550) | 0.80 | 1.0333 | 100% |
- The source says the note is “based on the belief that gold will outperform oil”, but the payoff pays when oil outperforms gold.
- Its 211% and 109% come from rounding gold’s return to 1.03 and 0.87 first; unrounded they are 210% and 110%.
Worst-of call
3-year, coal and BDI: 100% + max(0, worst asset return − 1).
| Coal | BDI | Worst | Payout |
|---|---|---|---|
| +10% | +5% | BDI +5% | 105% |
| +10% | −5% | BDI −5% | 100% |
| −10% | +5% | Coal −10% | 100% |
| −10% | −5% | Coal −10% | 100% |
Pays only if both rise — cheaper than a call on either, so it suits investors bullish on all assets.
Checkpoint 4: structured products
Checkpoint — section 14.7
1 / 6ZCB rate 4% for 5 years, USD 100,000 note. Cost of the ZCB (nearest dollar)?
Pick an answer to see if you are right.
Rank: investing practice
Investing practice
1 / 2Rank the generations of commodity index investing.
OrderDrag the handle to rearrange — top should be first.
- 1Active alpha
- 2Enhanced beta
- 3Passive long-only beta
Drag items into the correct order (or use ↑↓), then check.
Check: commodity investing
Commodity investing
1 / 6Why do most commodity investments reference futures rather than physical goods?
Pick an answer to see if you are right.
Calculate: numeric practice
Type the number (no units; minus sign for losses).
Formula sheet
| Item | Formula |
|---|---|
| Portfolio vol | √(w₁²σ₁² + w₂²σ₂² + 2w₁w₂ρσ₁σ₂) |
| Real return | (1 + nominal) ÷ (1 + inflation) − 1 |
| Gold future | S × (1 + (financing − lease) × days/360) |
| Roll yield | futures return − spot return = basis return + roll adjustments |
| TRS index amount | notional × (I_m − I_(m−1)) ÷ I_start |
| TRS fee | notional × fee × days/365 |
| ETC value | entitlement × 0.1 oz × gold price |
| Leveraged ETF | P × (1 + adj + L × index change) |
| ZCB cost | 100 ÷ (1 + r)^T |
| Participation | option budget ÷ cost of 100% call |
| Basket | Σ w × final ÷ initial |
| Reverse convertible | 100% + coupon − max(0, 1 − final/initial) if barrier hit |
Calculate: commodity investing
1 / 1780% equities (vol 16%), 20% commodities (vol 25%), correlation −0.28. Portfolio vol in % (2 dp)?
Enter a number (at most 2 decimal places).
Randomized practice
Practice: investing drills
1 / 6Gold ETC: entitlement 94.38% of 0.1 oz per security; gold USD 2987/oz. Value per security? Round to 2 decimals.
Enter a number (at most 2 decimal places).
Takeaways
- Investors range from hedge funds (short horizon) to pensions and SWFs (long); instruments are TRS, ETPs, and structured notes
- Rationale: diversification (correlation regimes shift), inflation and USD hedges, risk premia
- S&P GSCI: production-weighted, energy-heavy, front-month futures, rolled 5th–9th business day; spot vs ER vs TR
- Roll yield = futures return − spot return = carry; rolling itself makes no P&L, but contango still hurts longs and backwardation helps
- TRS: receive index, pay T-bill + fee, with adjusted notional
- ETC (physical, fee paid in metal), ETF (synthetic, collateral, path-dependent leverage), ETN (issuer credit)
- Structured notes = ZCB + options; participation depends on rates, vol, the forward curve, and correlation
Commodities rallied strongly in 2021–22 (energy shock), then fell back; gold hit successive records in 2024–25, passing USD 3,000/oz in 2025 amid heavy central-bank buying. The 2020 negative WTI episode forced several oil ETPs to restructure into deferred contracts — a real-world reminder of roll and counterparty risk.
Further Reading
- 13. Commodity-linked financing
- 2. Derivative valuation — forwards, carry, options
- 4. Gold — lease rates and forwards
- 5. Base metals — baskets, barriers, collars
- Campbell & Company (2014), Deconstructing Futures Returns: The Role of Roll Yield; S&P GSCI and Bloomberg Commodity Index methodologies
- Background: N. Schofield, Commodity Derivatives: Markets and Applications (Wiley) — used as a study outline