3. Contango and backwardation: what it costs to keep the position alive

Derivatives 1: Futures

A future has a sell-by date. That sentence, so obvious it hardly gets said, is the source of one of the market's quietest losses. If your horizon is longer than the contract you have bought, you have no choice but to roll the position over and over, and every roll has a price that almost nobody looks at before going in.

The futures curve and its two shapes

For the same underlying, several expiries trade at once: next month, the one after, six months out, a year out. If you plot those prices in order you get the futures curve, which can only have two shapes:

• Contango: the far expiries trade above the near ones. It is the normal shape when the cost of carry dominates, that is, when funding and storage cost more than the asset yields.
• Backwardation: the far ones trade below the near ones. It appears when having the physical asset in hand has a value of its own, because it is scarce or because it is needed right now. That benefit is called the convenience yield, and it subtracts from the cost of carry just as a dividend does.

They are only labels for two slopes. What matters is what each of them does to you while you hold the position.

Rolling the position: what it is exactly

Suppose you are long the March contract and want to stay exposed in April: before March expires, you sell that contract and buy the April one. That trade is called a roll, and it is not an administrative formality: it is a purchase and sale at market prices with an economic trail.

In contango, you sell the cheap contract and buy the dear one, so with the same money you control less material than before. In backwardation the opposite happens: you sell the dear one and buy the cheap one, and you end up controlling somewhat more.

It is worth being precise, because this is where almost everybody gets confused. The roll is not a commission nor an instant loss that shows up in your account: on the day of the switch, selling at 80 and buying at 81.20 does not take money away from you. What it does is move the bar. From that moment on you need the price to rise to 81.20 just to stand where you were. Repeat that twelve times a year and the bar has moved a very long way.

The drag, with numbers

Let us put figures on it. The near contract on a commodity trades at 80 dollars and the next one at 81.20. The difference is 1.20 on 80, that is, 1.5% per roll.

If you roll every month for a year, that 1.5% is applied twelve times, one on top of another: they do not add up to 18%, they compound. The result is that you end up controlling around 83.6% of the material you controlled at the start, or to put it another way, a drag of close to 16.4% a year.

Read the consequence carefully: if the spot price ends the year exactly where it started, you have lost around 16%. Not through commissions, not through a bad manager, not through bad luck. Through the shape of the curve.

And in reverse: if the near contract traded at 80 and the next one at 78.80, the same arithmetic works in your favor. Here it is best not to assume symmetry, because there is none: each roll leaves you 80 divided by 78.80, that is 1.52% more material, and compounded twelve times that comes to almost 20% of gain with spot flat, against the 16.4% you were losing in the opposite case. Compounding twelve times amplifies more on the upside than on the downside, and that detail explains why commodity series in prolonged backwardation perform far better than a quick calculation suggests. Over long horizons, the curve weighs more than the underlying's own movement.

Why an exchange-traded product can lose money for years

Here is the application that costs the private investor the most money. Many exchange-traded commodity funds do not store barrels or tons: they hold futures and roll them according to a published calendar. Their return is not spot's, but spot's plus the accumulated effect of those rolls.

When the market spends years in contango, a product like that can lose money steadily even though the commodity has risen over the same period. The most quoted case is the largest American exchange-traded oil fund: from its launch in 2006 until the end of 2019, the benchmark barrel finished at levels of the same order as at the start, while the fund had lost around 80% of its value. The exact figures depend on which dates you compare, but that is the magnitude, and it does not come from fees: it comes from thirteen years of rolling in contango.

That product is not broken: it does exactly what its prospectus says. What happens is that the prospectus describes an exposure to futures and the buyer thinks he is buying barrels.

A real case: April 20, 2020

The extreme case came on April 20, 2020. The WTI oil contract expiring in May, one day from ceasing to trade, closed at around minus 37.63 dollars a barrel. A negative price: you had to pay for somebody to take your oil.

The mechanism is purely physical. WTI settles with delivery at Cushing, Oklahoma, and Cushing's tanks were practically full because demand had collapsed during the lockdowns. Anybody still long the May contract faced receiving real barrels with nowhere to put them, and against that, any price at all was better than the logistical disaster.

Note that the spot price of crude elsewhere in the world was never negative. What was negative was one specific contract, with a saturated delivery point, in its final hours of life.

The consequences were expensive. The products rolling that contract fled toward distant expiries at the worst possible moment and with the curve in a brutal contango, and some changed their rules on the hoof. Several brokers also discovered that their systems did not even accept negative prices: one of the large American intermediaries admitted losses of the order of a hundred million dollars from client accounts left in the red.

The module's conclusion fits in a line: in futures, keeping the position alive is not free, and the bill arrives with every roll.