Futures Curve: Contango, Backwardation and Roll Yield

A commodity chart can mislead you without a single bad print. Pull up a multi-year continuous chart for crude oil or natural gas, watch it grind higher, and it’s tempting to assume the physical barrel or the gas itself roughly tripled over the period. Often it did nothing of the sort. You’re looking at a stitched series of separate contracts, and part of that slope is an artifact of how those contracts were joined together. The repair starts with reading the futures curve behind the chart: the set of prices for contracts on the same underlying asset that happen to expire on different dates. Once the curve is visible, contango and backwardation stop being jargon and start telling you where the cost of carry actually sits.

What a futures curve actually plots

A spot-price chart answers one question, plotted over time: what does the asset cost for immediate delivery right now. A futures curve answers a different question at a single moment. For a fixed underlying, what does each contract cost as a function of when it delivers. The horizontal axis is maturity, running from the front month out to contracts a year or more away. The vertical axis is price. Every point on the line is a live, separately traded contract, and not one of them is a projection.

Picture a hypothetical crude oil curve on one trading afternoon. Say the contracts line up like this:

• Front month: 78.20
• Two months out: 79.05
• Four months out: 80.10
• Six months out: 81.40

That upward slope is the whole object. It’s a snapshot of four prices you could transact in at the same instant, each tied to a different delivery date. The line shows contracts, not days of price history, and it makes no promise that crude will reach 81.40 in six months. When I sketch this curve for someone new to futures, I keep the spot chart open beside it so the difference stays physical. One panel is a movie of a single price through time. The other is a photograph of many prices captured at once.

Contango and backwardation are shapes, not forecasts

The curve above is in contango: later-dated contracts trade above nearer ones, so the line slopes up and to the right. Backwardation is the mirror image, where the front month sits above the deferred contracts and the line slopes down. Those two labels do all the work here, and both describe a relationship among maturities at one point in time.

One misread is worth killing early. A curve in contango doesn’t mean the asset is going up, and backwardation doesn’t mean it’s heading down. The slope measures one contract against another at the same instant. It never measures today against tomorrow. A market can sit in steep contango and fall for a year, and it can sit in backwardation and rally hard. I’ve watched newer traders treat an upward-sloping curve as a bullish tell and get the direction exactly backward, because they read a statement about carry as though it were a statement about trend. The curve reports the terms of storage and financing baked into the strip. It says next to nothing about the next move in the underlying.

Why the curve pulls away from spot

If a barrel costs 78.20 today, why would anyone pay 81.40 for delivery six months out? The gap has mechanical sources, and they differ by asset class. For a physical commodity, holding the real thing across those months costs money. You finance the purchase, you pay for storage tank space, and you insure the inventory. Add those carrying costs to spot and you get a rough floor for the deferred price. That cost-of-carry logic is the usual engine behind contango.

Backwardation normally means something pulls in the other direction, and the common culprit is convenience yield: the premium of holding the physical asset in hand right now. When a refiner can’t afford to run dry, immediate barrels are worth more than promised future ones, and the front month trades rich. Expected supply and demand ride on top of all this. A market pricing a winter shortage or a harvest glut bakes that view into specific parts of the curve, which is where seasonal patterns tend to surface in the strip. Commodity-focused investors such as Jim Rogers built entire theses on convenience yield and storage cycles that would make no sense applied to a stock-index contract. That last point matters. None of these inputs behave the same across markets. Equity-index futures carry dividends and short-term rates, currency futures reflect the interest-rate differential between two countries, and rate futures track the expected path of policy. The vocabulary of contango and backwardation travels between them. The economics underneath don’t.

Convergence: how a contract meets spot at expiry

The curve isn’t static, and the clearest way to feel that is to follow one contract to expiry. A futures contract has to converge to the spot price as delivery approaches, because on the last day the contract effectively is the spot asset. Watch a single deferred contract while spot barely moves:

• Ninety days out: spot 78.00, this contract 80.40
• Sixty days out: spot 78.10, this contract 79.60
• Thirty days out: spot 77.95, this contract 78.70
• Expiry week: spot 78.05, this contract 78.10

Spot went nowhere, drifting inside a 15-cent band. The contract lost more than two dollars, sliding from 80.40 down to meet spot. A trader holding that long contract into expiry in a contango market bleeds value even while the commodity itself stays flat. Convergence is doing exactly what it must here. There’s no fee and no slippage at work; the contract simply has to arrive at spot by the last day. It’s the single most common surprise for people coming from equities, where a share has no expiry and no built-in pull toward another price.

Roll yield: the return the price chart never shows

Because every contract expires, anyone who wants continuous exposure has to roll. You close the expiring contract and open a later-dated one. The return that comes from that swap, held apart from any move in the underlying, is roll yield. In contango you sell the cheaper expiring contract and buy a pricier deferred one, so the roll works against you. In backwardation you sell high and buy lower down the curve, and the roll works in your favor.

It helps to split a futures position’s total return into three parts that have little to do with each other. First, the spot return, whether the underlying rose or fell. Second, the collateral or financing return, the interest earned on the cash backing the position. Third, roll yield, the structural drag or lift from replacing one contract with the next. A commodity can climb 10 percent on the spot while a long futures position in steep contango still lags it, because the roll quietly handed back part of that gain at every switch. Blending the three together is how people end up blaming a strategy for a loss the term structure caused.

When a continuous chart hides the roll

Now the opening problem comes back into focus. To chart years of futures history on one line, data vendors splice expiring contracts into a continuous series, and there’s no single clean way to do it. Paste raw prices together and every roll leaves a gap, because the expiring contract and the incoming front month traded at different levels. When I build such a series by hand, the switch date shows a tidy step, say a 2.10 jump, that no single contract ever printed. Those artificial jumps corrupt any indicator that measures change across the seam.

The alternative is a back-adjusted series, where old prices are shifted so the roll gaps vanish and the line runs smooth. That smoothing is a choice with side effects. Back-adjustment can push very old prices into negative territory on long histories, and it changes the size of every historical percentage move, so an adjusted price series can report a trend or a drawdown the tradable contracts never delivered. The same roll shows up as price gaps on an unadjusted chart and as distorted magnitudes on an adjusted one. You pick which distortion you can live with, but you need to know which one you picked.

That’s where a backtest can fool its author. Run a trend model or a moving-average signal over a continuous series, and the result depends as much on the splice method as on the market. A strategy that looks profitable on a back-adjusted chart can owe its edge to the construction rather than to any tradable move, and a test that ignores roll yield will overstate what the position would really have returned. Before I trust any long futures backtest, I want the roll rule, the adjustment method, and the data source in writing, because those three choices alone can manufacture or erase an entire trend.

Reading a curve without fooling yourself

A futures curve is a precise, useful picture, and it’s also a narrow one. It describes the relationship among contract prices at a single moment, and that shape can flip from contango to backwardation within days when a supply shock lands. It can’t be read the same way from one market to the next without knowing the contract specification, the expiry convention, the data source, and the construction method behind whatever chart you’re viewing. A curve on gasoline, a curve on the euro, and a curve on an equity index answer to different economics, and the shared vocabulary hides that at the surface.

So treat the curve as a description of carry and expectations, and keep it firmly separate in your head from the spot chart and from the spliced line your platform draws by default. Holding those views side by side is ordinary inter-market analysis work, and it’s what stops a roll artifact from masquerading as a real trend. Before you read direction into a term structure, read the specification first. Learn the pattern. Ride the trend. Keep the gains.

Educational content only. Not investment advice. Trading involves risk. You are responsible for your decisions.