Most commodity investors focus on spot prices — whether oil is going up or down. But a futures investor's return is not determined by the spot price alone. It is shaped critically by the shape of the forward curve: whether deferred contracts are priced above or below the current spot. This guide builds the complete framework — from the mechanics of a futures contract to the cost-of-carry pricing formula, the dynamics of contango and backwardation, the arithmetic of roll yield, and the documented historical evidence for roll yield as a source of return. It concludes with two case studies demonstrating how ignoring the curve structure has destroyed retail capital.
How Futures Contracts Work
A futures contract is a legally binding, standardized agreement to buy or sell a specific quantity of an asset at a predetermined price on a specific future date. Contracts are traded on centralized exchanges — the CME Group, ICE, Eurex — which act as counterparty guarantor for both sides, eliminating default risk.
The key operational distinction: futures investors rarely take or make physical delivery. They maintain exposure by rolling positions forward before expiration — selling the expiring contract and buying the next one. This rolling process is where roll yield is generated. But before that, the basic mechanics require understanding.
Margin, Leverage, and Mark-to-Market
Futures do not require payment of the full contract value upfront. Instead, traders post an initial margin — a good-faith performance bond — typically 2–12% of the contract's notional value depending on the asset's volatility. This creates substantial leverage. A crude oil contract worth $70,000 in notional value might require only $7,000 in margin, yielding 10:1 leverage.
At the close of each trading day, the exchange calculates each account's gain or loss based on the daily settlement price and immediately credits or debits cash. There is no accumulation of unrealized P&L — every day the position is repriced and settled in cash. If losses reduce the account below the maintenance margin threshold, the broker issues a margin call: deposit additional funds immediately or the position is forcibly liquidated. This daily settlement process eliminates counterparty credit risk at the exchange level.
| Day | Settlement Price | Daily P&L | Margin Balance | Action Required |
|---|---|---|---|---|
| Entry | $70.00/bbl | — | $7,000 | None |
| Day 1 | $68.00/bbl | −$2,000 | $5,000 | Margin call if maintenance = $6,000 |
| Day 2 | $69.50/bbl | +$1,500 | $8,500 (post deposit) | None |
| Day 3 | $72.00/bbl | +$2,500 | $11,000 | None — excess available |
Illustrative example. One WTI crude oil contract = 1,000 barrels. A $2/bbl move = $2,000 P&L. Initial margin assumed $7,000 on $70,000 notional (10:1 leverage).
Physical vs. Cash Settlement
At expiration, futures contracts settle in one of two ways. Physical delivery requires the short side to deliver the actual commodity — 1,000 barrels of WTI crude oil at Cushing, Oklahoma; 5,000 bushels of corn at approved Illinois elevators. Financial institutions rarely hold positions to physical delivery; instead they roll before the first notice day. Cash settlement contracts (equity indices, VIX, some currencies) simply mark the final settlement price to an agreed reference index and net the cash difference. No physical commodity changes hands.
The April 2020 WTI crude oil price collapse to −$37.63/barrel was directly caused by the physical delivery constraint. WTI settles at Cushing, Oklahoma specifically — a landlocked hub with fixed storage capacity. When that storage filled completely and traders holding May contracts couldn't take delivery, they were forced to pay counterparties to take the contracts. A Brent crude contract, which allows alternative delivery options, did not go negative.
The Cost of Carry Pricing Formula
A futures price is not a market prediction. It is determined by an arbitrage relationship with the current spot price — specifically, the cost of buying the physical commodity today, storing it, and delivering it at the contract expiration. Any deviation from this relationship creates a riskless profit opportunity that arbitrageurs immediately eliminate.
The intuition: to buy a commodity today and deliver it into a futures contract at expiration, you must pay interest on the capital deployed (r) and pay to store and insure it (u). These costs push the futures price above the spot price. Against this, the convenience yield — the value of having the physical commodity in your possession right now rather than a futures claim on it — pulls the futures price below spot.
The convenience yield behaves like an option on physical supply. When inventories are ample, convenience yield is near zero — there is no particular benefit to holding physical crude oil rather than a futures contract. When inventories are severely depleted, convenience yield spikes: a refinery facing a potential shutdown will pay almost anything to obtain physical crude oil today rather than wait for a futures delivery. This convenience yield spike is what drives markets into backwardation during supply shocks.
The formula also reveals why gold futures always trade in contango. Gold has minimal storage costs, essentially zero industrial convenience yield (there is no factory that will shut down without gold), and positive interest rates. With y ≈ 0, the formula simplifies to F = S × e^(r × T) — a pure financing cost. Gold's contango is not a signal of supply glut; it is the interest rate expressed in commodity form.
Contango & Backwardation
The futures curve plots the prices of sequential contracts across their expiration dates. Its shape — whether deferred contracts are more or less expensive than the spot — is the single most important determinant of a buy-and-hold futures investor's return.
Contango
Futures prices are higher than spot. The curve slopes upward. Deferred contracts are more expensive than near-term ones.
Signals abundant near-term supply, high inventory levels, or low immediate demand. Common in oil during supply gluts, gold in positive rate environments, natural gas between winter peaks.
Impact on rolling investor: You sell the cheaper expiring contract and buy the more expensive deferred contract each month. Negative roll yield bleeds your capital mechanically even if spot prices are flat.
Backwardation
Futures prices are lower than spot. The curve slopes downward. Near-term contracts are more expensive than deferred ones.
Signals physical scarcity, supply disruption, or extreme near-term demand. Occurs during geopolitical shocks, production outages, and harvest failures.
Impact on rolling investor: You sell the more expensive expiring contract and buy the cheaper deferred contract. Positive roll yield generates returns even if spot prices are flat. This is Keynes' hedging pressure premium in action.
The Keynes / Normal Backwardation Theory
John Maynard Keynes proposed in 1923 that commodity futures are structurally biased toward backwardation — not just during supply shocks, but as a persistent feature driven by the risk transfer needs of commercial producers. Producers (oil companies, farmers, miners) are naturally long the physical commodity through their operations. To lock in revenue certainty, they sell futures forward — accepting a price below the expected future spot price to guarantee income. Financial speculators who take the other side of this trade earn a risk premium as the futures price drifts upward toward spot at maturity. Keynes called this systematic underpricing of futures relative to expected spot the "normal backwardation" premium.
The financialization of commodity markets over the past two decades has complicated Keynes' theory. As institutional capital flooded into commodity long-only index strategies, the natural hedging pressure imbalance was partially offset by large systematic long positions that pushed some markets into persistent contango. The theory remains a valid explanation for backwardation in specific markets with concentrated producer hedging — crude oil, copper, natural gas — but is not a universal law across the entire commodity complex.
Convergence: The Mechanical Engine of Roll Yield
Regardless of whether a curve is in contango or backwardation, one thing is certain: as a futures contract approaches its expiration date, its price must converge to the spot price. If it did not, an immediate, riskless arbitrage would exist on delivery day. This mechanical convergence is what creates roll yield. The question is only the direction — whether the deferred contract you bought drifts up toward spot (backwardation, positive yield) or down toward spot (contango, negative yield).
Roll Yield: The Hidden Return Driver
Roll yield is not the cash profit or loss on the day you switch contracts. It is the return generated over the life of the position as the futures price converges toward spot. It is entirely mechanical — determined by the curve shape when you initiate the position, not by any subsequent price movement.
Total Return Decomposition
The total return of a fully collateralized commodity futures investment has three components:
This decomposition is critical for understanding why commodity ETFs routinely fail to track spot prices. An ETF holding front-month futures can dramatically underperform the spot price if the curve is in steep contango — the spot return is positive, but the roll yield is sufficiently negative to offset or exceed it.
Crude oil spot: $80. The three-month futures contract trades at $75 — backwardation of 5/80 = 6.25%. You buy the $75 contract. Over three months, assuming spot stays at $80, convergence drives your contract from $75 → $80. You earn a 6.7% return purely from curve convergence, independent of any spot price movement. This is the carry premium Keynes theorized and Gorton & Rouwenhorst documented.
Natural gas spot: $3.00. The three-month futures contract trades at $3.45 — contango of 15%. You buy the $3.45 contract. Over three months, even if spot stays at $3.00, convergence drives your contract from $3.45 → $3.00. You lose 13% purely from curve convergence, before any spot price movement is considered. If spot also falls, the losses compound. This is the structural drag that has made passive long natural gas strategies catastrophically unprofitable.
Roll Timing: When to Roll Matters
The standard practice is to roll within a window around the first notice day — the first date on which the exchange can assign physical delivery obligations. Rolling too late means risking physical delivery; rolling too early means paying for time value before it is necessary.
More sophisticated implementations use volume-based rolling: roll when open interest in the deferred contract exceeds the front-month contract, signaling that the market has already transitioned its liquidity center. This typically reduces transaction costs by avoiding the wide bid-ask spreads that emerge in thinly traded expiring contracts.
In deep contango markets, moving the exposure further down the curve — holding the six-month contract rather than the front-month — bypasses the steepest portion of the curve where roll costs are highest. This is the core of "enhanced roll" index strategies and why they have dramatically outperformed naive front-month benchmarks like the original S&P GSCI construction.
Continuous Contract Back-Adjustment
Quantitative researchers studying roll yield across history face a data problem: individual contracts expire. Building a continuous long-term price series requires stitching successive contracts together. The method of stitching matters enormously — getting it wrong means backtests will record profits or losses that are mathematically impossible in live trading.
The problem: when you roll from the November contract trading at $51.50 to the December contract trading at $53.50, an unadjusted splice shows an artificial $2.00 jump in the price series. A backtest treating this as a real gain will dramatically overstate strategy performance.
| Date | Active Contract | Raw Price | Adjusted Price |
|---|
Best for: P&L Backtesting
Subtracts the absolute dollar roll gap from all prior data. Preserves dollar-per-contract P&L accuracy. Limitation: In sustained contango markets, historical prices eventually go negative — corrupting percentage calculations.
Best for: Volatility Analysis
Multiplies all prior data by the price ratio at the roll. Preserves percentage returns; prices never go negative. Limitation: Destroys absolute dollar levels — a $1 move in adjusted data no longer equals $1,000 P&L on a standard contract.
Some researchers use a blended series that linearly weights front and back-month contracts over a multi-day rolling window, creating a synthetic smooth price series. This is useful for macroeconomic research but is not tradeable — it understates actual transaction costs and roll friction. Never backtest a live trading strategy on perpetual data.
The Empirical Evidence Base
Two landmark studies established roll yield as a documented, persistent source of commodity return rather than a theoretical construct.
The foundational paper. Gorton and Rouwenhorst constructed an equally weighted index of fully collateralized commodity futures from 1959 to 2004 — 45 years of data, controlling for survivorship bias. Their headline finding was that commodity futures generated equity-like risk premiums with lower volatility and negative correlation to stocks and bonds. Crucially, they demonstrated that this premium did not come from rising spot prices — spot prices historically lagged inflation. Instead, the majority of the total return was generated by roll yield in markets that spent meaningful time in backwardation. Their decomposition established that most investors who thought they were profiting from commodity price appreciation were actually earning a carry premium from the curve structure.
The long-horizon validation. Extending the dataset to 138 years (1877–2015) across 29 commodities, AQR researchers confirmed the G&R findings with far greater statistical power. They separated total return into spot return and carry (roll yield) components and analyzed both across different macroeconomic regimes — inflation rising vs. falling, recession vs. expansion. Their key finding: spot returns are volatile and regime-dependent, surging during inflationary expansions and collapsing during recessions. Carry (roll yield) is far more persistent, providing a stable structural return across regimes. A long-short portfolio based purely on each commodity's carry signal — long backwardated, short contangoed — produced consistently positive Sharpe ratios independent of the directional commodity market environment.
Much of the G&R historical premium was earned in a period when U.S. energy markets were structurally backwardated — conventional oil supply was tight relative to demand. The U.S. shale revolution after 2008 dramatically altered the supply picture, pushing crude oil and particularly natural gas into prolonged contango regimes. The AQR paper's longer dataset shows the premium is more persistent when measured dynamically (long backwardated, short contangoed) than statically (long-only). The lesson: the premium exists in the curve signal, not in passive long exposure.
Real-World Pitfalls & Case Studies
The mechanics of roll yield have destroyed substantial retail capital — primarily through commodity ETFs that mechanically roll front-month contracts without accounting for curve structure. Two case studies illustrate the problem in its most visceral form.
Sector-Specific Roll Cost Drag
The magnitude of roll drag varies sharply across commodity sectors, driven by storability and structural supply dynamics.
Case Study: The United States Oil Fund (USO)
USO was designed to track daily percentage changes in WTI crude oil by holding the nearest-month futures contract and rolling it forward approximately two weeks before expiration. This mechanical design made the fund highly transparent — and structurally exposed to roll drag whenever the curve entered contango.
Following the 2008 financial crisis, the collapse in global demand left massive crude oil inventories in storage, pushing the WTI curve into steep contango. Retail investors flooded into USO believing "oil is cheap" and a recovery was imminent. They were correct about the spot price recovery — crude oil rallied substantially through 2009. But USO investors earned almost nothing. Every month, USO sold the cheaper expiring contract and bought the more expensive deferred contract. The monthly roll cost consumed the spot price appreciation. Investors who correctly predicted the direction of crude oil prices still lost money, because they owned the curve, not the spot.
The April 2020 event was the ultimate illustration of what happens when futures mechanics are misunderstood at scale. COVID-19 caused a 30+ million barrel-per-day demand collapse. Simultaneously, Saudi Arabia and Russia flooded markets with supply. Cushing, Oklahoma storage reached approximately 85% capacity — with all remaining space pre-leased. Financial traders who had bought the May WTI contract without any ability to take physical delivery found themselves in an impossible position in the final hours of trading: they could not take delivery, they could not find willing counterparties, and the clock was running. The May WTI contract settled at −$37.63 per barrel on April 20 — traders literally paid counterparties to take contracts off their hands. USO, which had already rolled into June contracts, avoided the worst of the event — but still lost approximately 75% of its value for the year and was subsequently forced by the SEC to restructure its mandate, distributing exposure much further along the curve. The SEC fined USO $2.5 million for disclosure failures related to its broker restricting new contract purchases during the crisis without timely investor disclosure.
Agricultural Seasonal Contamination
Agricultural markets present a specific pitfall for carry signal construction: the futures curve is broken into "old crop" and "new crop" segments that represent entirely different physical supply realities.
For U.S. corn, the March, May, and July contracts represent the current inventory from last autumn's harvest — a known, fixed supply that is being drawn down at a known rate. The December contract represents the next harvest — unquantified and subject to the full uncertainty of growing season weather. The spread between July corn and December corn does not measure standard inventory-driven carry. It measures the weather risk premium on an unborn crop.
To extract a genuine carry signal from agricultural futures, quantitative strategies compare contracts with the same calendar month but different crop years — for example, the July 2025 corn contract vs. the July 2026 corn contract. This comparison eliminates seasonal weather premium contamination entirely, because both contracts represent the same point in the same seasonal cycle, just one year apart. The spread between them reflects genuine structural supply and demand imbalance rather than planting uncertainty. Standard carry signals using adjacent calendar month spreads in agricultural markets are effectively noise.
Key sources: Gorton, G. & Rouwenhorst, K.G. (2006). Facts and Fantasies about Commodity Futures. Financial Analysts Journal. · Levine, A., Ooi, Y.H., Richardson, M. & Sasseville, C. (2018). Commodities for the Long Run. Journal of Investment Management. · Kaldor, N. (1939). Speculation and Economic Stability. · Keynes, J.M. (1923). A Tract on Monetary Reform. · SEC v. United States Commodity Funds LLC (2021). · U.S. Energy Information Administration — Cushing storage data, April 2020.