Spot is set by physical supply and demand. Everything dated is set by the cost of carrying the good to that date - and when the curve slopes upward, a fund that holds contracts loses money every time it rolls, even with the price standing perfectly still.
12 min read · Level 3
The roll is the clearest example in this whole curriculum of a cost that is genuinely invisible rather than merely overlooked. A vault charges a fee and sends you something. A fund states an expense ratio in its documentation. The roll is not a charge at all: it is the difference between two prices, both of which are correct, paid at a moment when nothing appears to have happened. Nobody debits it and no document totals it, and it is routinely several times larger than the fee that every buyer does compare. What I find worth arguing about is the conclusion people draw when they finally notice it, which is usually that futures-based products are a trap. I do not think that is right. The roll is symmetric - it paid 5.8016% in the backwardated year in this lesson - and the asymmetry is not in the instrument but in who is holding it and why. A manufacturer hedging an input is indifferent to the curve because they want the good on the date. A financial buyer who wanted only the price has accidentally bought a second position they never analysed, and they will meet it either as an unexplained shortfall or as an unearned windfall. The useful discipline is therefore not to avoid the instrument. It is to look at the curve before buying, state the annualised number out loud, and be able to say whether you are willing to hold that position as well as the one you came for. It takes a minute and it is the only piece of commodity diligence with an exact answer.
There are two prices in this class and confusing them is the most expensive error available.
The spot price is what the physical good changes hands for today. It is set by physical supply and demand: how much is being produced, how much is being consumed, what is sitting in warehouses, and what it costs to move it. There is no earnings forecast and no discount rate anywhere in that sentence, which is why this class is priced unlike anything else in the curriculum.
A futures price is what somebody will pay today for delivery on a stated future date, and it is not a forecast. It is arithmetic:
Futures price = spot price + the cost of carrying the good to that date
Carrying costs money: the capital is tied up, and the thing has to be stored and insured. Take Metal M at a spot price of 2,000, with financing at 5.0% a year and storage at 0.6% a year, so 5.6% in total:
Three months 2,000 x (1 + 0.056 x 0.25) = 2,028.00 Six months 2,000 x (1 + 0.056 x 0.50) = 2,056.00 Twelve months 2,000 x (1 + 0.056 x 1.00) = 2,112.00
An upward-sloping curve like that is called contango, and it is the ordinary state of a commodity that is plentiful and expensive to store. It does not mean anybody expects the price to rise. It means it costs 2.8% to hold the metal for six months, so nobody will sell it to you for delivery in six months at today's price.
The opposite shape, backwardation, has the dated price below spot - here 1,972.00 at three months. That happens when the good is scarce right now and people will pay a premium to have it immediately rather than later. It is a statement about present scarcity, not about the future.
Now the consequence, which is where the money is.
A fund holding futures cannot hold them forever, because every contract has a date. To maintain the exposure it sells the expiring contract and buys the next one. As a contract approaches its date, its price converges to spot - it has to, because at delivery they are the same thing.
So with the spot price standing exactly still at 2,000 all year, and a fund rolling a three-month contract each quarter:
Buys the three-month at 2,028.00 It converges to spot at 2,000.00 a loss of 28.00, or 1.3807% Buys the next three-month at 2,028.00 and the same thing happens again
Four rolls: (1 - 0.013807)^4 - 1 = -5.4094%
The metal did nothing. The fund is down 5.4094%, and it tracked its contracts perfectly.
Because that last figure explains something that otherwise looks like incompetence or dishonesty, and is neither: commodity funds that hold contracts can return far less than the price of the commodity they name.
Run it the other way to see that the mechanism is symmetric. In backwardation, with a three-month contract at 1,972.00 converging up to an unchanged spot of 2,000:
Gain per roll 28.00 / 1,972.00 = 1.4199% Four rolls 1.014199^4 - 1 = +5.8016%
Same metal, same flat spot price, and a swing of more than eleven percentage points in a year depending only on the shape of the curve. This effect has a name - roll yield - and it is positive in backwardation and negative in contango.
Three things follow, and they are the practical content of this lesson.
A fund's name tells you what it tracks and not what it holds. A product named after a commodity may hold the metal, or it may hold contracts on the metal. Those two behave identically only when the curve is flat, which it rarely is. The distinction is in the documentation and it is the single most consequential sentence in it.
The tracking gap is not an error and cannot be fixed by a better manager. It is the cost of carrying the good, which somebody has to pay. A physical product pays it as a storage charge you can see; a futures product pays it in the roll, where it is invisible and where it is usually larger. Neither is cheating.
The curve shape is a fact you can look up before buying, not a forecast. The dated prices are published. Reading them takes a minute, tells you the annualised cost of maintaining a futures exposure, and is the one piece of diligence in this class that has a precise, checkable answer.
There is one more reason the two prices diverge, and it is worth knowing because it is the main thing the carry formula leaves out. Somebody who actually needs the physical good - a manufacturer who will stop producing without it - values having it now in a way that a financial holder does not. That benefit is called a convenience yield, it behaves like a negative carrying cost, and it is why a curve can be in backwardation even when storage and financing are expensive. It is also why this class connects to real economic conditions in a way a share price does not: the curve is telling you something about physical scarcity right now, and it is one of the few genuinely informative numbers in commodity investing.
Educational, not personalised financial advice. Nothing in the Investor Academy is a recommendation to buy, sell or hold anything — what is right for you depends on your own circumstances and the rules where you live.
Two investors who agreed completely about Metal M and disagreed about nothing except which product to buy.
The year begins with spot at 2,000, the curve in contango at the carry rate above, and it ends with spot at 2,000. Nothing happened to the metal.
Investor one bought a physically backed product. It holds allocated metal and charges 0.40% a year.
Price effect unchanged Fee -0.40% Outcome about -0.40%
Investor two bought a futures-based product. It rolls a three-month contract quarterly and charges 0.75% a year.
Price effect unchanged Roll -5.4094% Fee -0.75% Outcome -6.1188%
Both products are honest, both did exactly what their documentation said, and both track Metal M. The difference is 5.7 percentage points in a year in which the commodity did nothing at all, and it is entirely the curve.
Now the year the curve was the other way. Same flat spot price, but the market in backwardation:
Investor one about -0.40% the fee is the fee Investor two +5.8016% compounded with the 0.75% fee, so +5.0081%
The futures product now beats the physical one by more than five points, again with the metal doing nothing. Neither investor did anything differently. The roll yield swung from -5.4094% to +5.8016%, a swing of 11.2110 percentage points, and neither of them chose it.
Which is why the honest way to read a futures-based commodity product is that it gives you two exposures rather than one: the price of the commodity, and the shape of its curve. If you want only the first, the physical route delivers it and charges you visibly for the privilege. If you are comfortable with both, the curve can work for you as easily as against you - but it should be a position you know you hold.
And before buying either, the curve is published. Looking at it takes a minute and answers precisely which of these two years you are walking into today.
Assuming a futures price is a forecast. It is the spot price plus the cost of carrying the good to that date. A twelve-month price of 2,112.00 against a spot of 2,000 says storage and financing cost 5.6% a year; it says nothing whatever about where the price will be.
Buying a product named after a commodity without checking what it holds. Metal and contracts on metal behave identically only when the curve is flat. In the contango illustrated here the difference was 5.4094% in one year with the spot price unchanged.
Calling the tracking gap an error. It is the cost of carrying a physical good, and somebody always pays it. A physical product charges it visibly as storage; a futures product pays it in the roll, invisibly, and usually more.
Reading contango as bullish and backwardation as bearish. It is the reverse of a forecast: contango generally means the good is plentiful and costly to store, and backwardation generally means it is scarce right now and people will pay to have it immediately.
Treating the roll as a one-directional cost. The same mechanism produced -5.4094% in contango and +5.8016% in backwardation, a swing of 11.2110 points. A futures product is an exposure to the curve as well as to the price, and the curve can pay you.
Checking the curve after buying. The dated prices are published, reading them takes a minute, and the answer is precise - which makes this the cheapest piece of diligence available anywhere in this subject.
Spot, three months, six months, twelve months. Whether the line slopes up or down, and by how much, is published and takes a minute to read. That single observation tells you the annualised cost or benefit of holding this exposure through contracts.
A twelve-month price of 2,112.00 against a spot of 2,000 is 5.6% a year. A three-month price of 2,028.00 is 1.3807% per roll, which is -5.4094% over four rolls. Stating it as a percentage is what makes it comparable with the fee, the storage charge and everything else in lesson two.
Allocated metal, an unallocated claim, or futures contracts. This is the sentence that determines whether the roll applies to you at all, and it is usually one line in a document most buyers never open.
At the contango here, the physical route costs 0.40% and the futures route costs 6.1188%. In backwardation the futures route gains 5.0081% while the physical route still costs 0.40%. Run both before choosing, because the answer reverses and you are choosing the exposure, not the outcome.
The shape changes, and with it the cost of a position you may not otherwise be reviewing. A holding that made sense in backwardation can be quietly paying more than 5% a year once the curve turns, with nothing on any statement to announce it.
Before buying any commodity product, when a fund's return has diverged from the commodity it names, when choosing between a physical and a futures-based route, and periodically on any futures-based holding.
Do not use the curve to forecast the price - it is a cost of carry, not a prediction, and reading it as one is the error this lesson exists to prevent. And do not apply these figures to another commodity; storage costs and scarcity differ enormously between them.
Investors who buy a product named after a commodity and then compare its return with the commodity's spot price frequently conclude that the product failed. Complaints follow, and so sometimes do explanations from the provider that are technically complete and entirely unpersuasive, because the buyer's expectation was formed by the product's name rather than by its documentation. The product usually tracked its own stated index to within a fraction of a per cent. What it did not do, and never claimed to do, was replicate the spot price, because holding contracts through a contangoed curve cannot replicate a spot price. The mismatch is set at the moment of purchase, by a decision that took a few seconds and rested on a name, and it is fully knowable beforehand from one line of documentation and one look at a published curve.
A product's name says what it tracks, never what it holds. The sentence saying what it holds is the one that determines your return.
Curve shapes are regularly reported as though they carried a directional message, with backwardation described as a market signalling strength and contango as weakness. The formula points the other way: an upward-sloping curve is what you get when a good is plentiful and expensive to store, and a downward-sloping one is what you get when it is scarce enough right now that buyers will pay to have it immediately. Neither says anything about the price in a year, because both are dominated by financing, storage and present availability rather than by anyone's forecast. The persistent appeal of the directional reading is that a curve looks like a line on a chart moving through time, and lines on charts moving through time are usually predictions. This one is a cost.
Read a curve as a statement about storage and present scarcity, never as a prediction. It is the cost of waiting, not a view about what waiting will bring.
Create a free account to read the rest and apply it to your own investing.