Contango and backwardation

One commodity, many prices 

A commodity does not have a price. It has a set of prices, one for each date on which it can be delivered. The price for immediate delivery is the spot price, and alongside it a market quotes a price for delivery next month, the month after, and so on out along a listed strip. Plotted against their delivery dates, those prices form a curve, and the shape of that curve is a separate piece of information from the level of the prices in it.

Two names describe the two possible shapes. A curve in contango slopes upward: each successive delivery month costs more than the one before it. A curve in backwardation slopes downward: the nearest months are the dearest, and delivery further out costs less. The terms are old market usages rather than technical constructions, and they describe the relationship between dates and nothing else. A market in contango can be rising, falling or flat, and so can a market in backwardation.

Why a curve normally slopes upward 

The explanation for contango is arithmetic rather than psychological. Somebody who buys a physical commodity today and holds it until a later date incurs three costs: storage, insurance and the funding of the capital tied up in the goods. A buyer indifferent between taking delivery now and taking delivery later will pay more for the later date by roughly the sum of those costs, because that is what the seller has to spend to bridge the gap. The sum is called the cost of carry, and it is the natural width of a contango.

The relationship is enforced by arbitrage rather than by convention. If the later month were priced above spot by more than the full cost of carrying the goods, a participant with access to storage and funding could buy the physical commodity, sell the later contract, store the goods and lock in the difference. That trade is available to anyone with a warehouse and a credit line, so the gap does not usually widen much beyond the cost of the trade. Where it does, the binding constraint is normally that storage capacity is physically full, which is exactly the condition under which deeply oversupplied energy markets have produced extreme curve shapes.

Why a curve sometimes slopes downward 

Backwardation is harder to explain by arithmetic, because a negative cost of carry is not physically possible. The standard explanation is the convenience yield: a benefit that accrues from actually holding the physical good rather than a claim on it later. A refinery with crude in its tanks can keep running through an interruption. A manufacturer with metal on site does not stop its line. That benefit has a value, and when scarcity is acute the value of having the good now exceeds the cost of storing it, so nearby prices trade above later ones.

Backwardation is therefore the market's characteristic shape when physical supply is tight, and it tends to be steepest at the front of the curve, where the scarcity is most acute, flattening further out where the market expects conditions to have normalised. That last clause is where a genuine analytical disagreement lives, and it is set out below rather than resolved.

The roll, where curve shape becomes a real cost 

A futures contract has a last trading day. An exposure that is meant to persist beyond it cannot simply continue; the expiring contract is closed and an equivalent position is opened in a later month. That transfer is the roll, and its economics are set entirely by the shape of the curve. In contango the later contract costs more than the one being closed, so maintaining the same exposure requires paying up, and the position is worse off by the difference before the price of the commodity has done anything at all. In backwardation the later contract is cheaper, and the transfer is a credit on the same terms.

The final row is the practical content of the whole subject. Under a persistent contango, an exposure held continuously through several rolls requires the spot price to have risen by the accumulated roll cost simply to break even, and that requirement exists independently of any view about the commodity. This is the documented reason that long horizon holdings in continuously rolled commodity products have at times diverged substantially from the spot price of the commodity they track, and it is the first thing to check before comparing a chart of spot prices with the record of an instrument that rolls.

How a CFD inherits the curve 

Contracts for difference on commodities are written in two forms and the curve reaches both, by different routes. A dated instrument references a specific delivery month and inherits its last trading day: on the stated date it is closed at the prevailing price or rolled into the following contract, and the date is published in the instrument's specifications rather than chosen by either party. Where a roll occurs, the price gap between the two months appears as a step in the quoted price, and providers ordinarily apply a corresponding cash adjustment so that the step itself does not create or destroy value in an open position.

A continuous instrument has no stated expiry and quotes a single ongoing price, but the underlying exposure still has to be moved between delivery months, so the economics of the roll are applied instead as an adjustment while the position remains open. The two designs differ in presentation rather than in substance: in neither case does the curve go away, and in both cases the cost or credit accrues to the party holding the exposure. A position held past the daily cut off additionally carries a financing adjustment for as long as it stays open, which is a separate line from the roll.

The practical consequence is that the horizon over which a commodity exposure is held changes what it costs, and does so through a mechanism that has nothing to do with the direction of the price. A position closed within a session is unaffected by the curve. A position carried across several delivery months is exposed to it repeatedly, and in a steep contango that exposure can be a substantial share of the position's value over a year.

Why a curve is not a forecast 

The most common misreading of a futures curve is to treat the price of a distant month as the market's expectation of where the spot price will be on that date. It is not. The distant price is where a trade can be agreed today for that date, and it is bounded by arbitrage against storage and funding costs rather than by anybody's opinion. An upward sloping curve in a storable commodity is the normal state of affairs and says nothing about direction, which is why a headline describing contango as a bullish forecast is describing the wrong thing.

There is a genuine and unresolved argument underneath this, and a reader will meet both sides of it. One tradition holds that a futures price is an unbiased estimate of the future spot price, so any systematic difference between them should not persist. Another, older tradition holds that hedgers with physical exposure are structurally willing to accept a worse price in order to transfer risk, so the price of a distant contract carries a persistent premium or discount that compensates whoever takes the other side. The empirical evidence supports each account in different commodities and different periods, which is why the disagreement has not resolved, and why any statement that a curve shape reliably predicts anything should be treated as a claim requiring evidence rather than as an established fact.