A Futures Curve Is a Price, Not a Promise
Commodity futures curves are often read as forecasts of where commodity prices will be later. The useful question is narrower and more demanding: what information does a futures price add about a future spot price, and what does it fail to promise? This distinction matters to institutions, including central banks, that use futures as a guide to market expectations. It also matters to anyone tempted to treat a market quote as a settled fact about the future.
A futures curve is a set of prices for contracts with different maturities. Each quote is the price at which buyers and sellers agree today on a contract for a specified future date. The future spot price is the price observed in the physical market when that date arrives. Those two prices are related, but they are not the same object viewed at different times. One is a negotiated price today; the other is a later outcome shaped by events that have not yet occurred.
That sounds obvious until a curve becomes politically or economically important. When an oil supply disruption, a harvest threat, or a geopolitical conflict changes the shape of the curve, observers often ask what the market “expects” will happen. A steep curve can look like a confident prediction. A shallow curve can look like reassurance. But the curve is not a public vote on one future price. It is the result of trading, risk transfer, physical constraints, and uncertainty interacting in the present.
The distinction is especially important because the price can be informative without being a neutral summary of every participant’s beliefs. A producer and a consumer may both have good reasons to trade even when neither knows the eventual spot price. One may be reducing exposure to a fall; the other may be protecting against a rise. Their willingness to transact can reveal something about the value of protection, the availability of inventory, or the urgency of physical supply. It can also move the contract away from the price that, in hindsight, would have been the best numerical forecast. Information and bias can therefore coexist.
A simple example makes the point without requiring a prediction. Suppose immediate physical supply is unusually valuable because inventories are low. The spot market may reflect that immediate scarcity, while the futures market reflects a different balance between the cost of carrying supply, the value of having it now, and the risks that conditions will change. The relationship between today’s spot price and a later futures quote can then describe a market under pressure. It does not settle which of those pressures will remain when the contract matures.
The apparent promise
The Federal Reserve’s recent examination of commodity futures begins with exactly this problem. Futures prices are used to infer market expectations, yet a futures contract is frequently bought or sold to hedge risk rather than to make a pure bet on the price that will eventually settle. A producer may sell futures to reduce exposure to a future fall in prices. A consumer may buy them to protect against a future rise. Their agreement produces a price, but the existence of that price does not tell us that either participant believes the eventual spot price will equal it.
This is why a futures price can be informative without being an unbiased forecast. Trading decisions can incorporate information about supply, demand, inventories, and future conditions. They can also incorporate the cost of carrying a commodity, the value of having the physical commodity available, and the terms on which participants transfer risk. The quote may therefore tell us something important about the state of the market while still being systematically above or below the spot price that later appears.
The distinction becomes easier to hold if three objects are kept separate:
- The current futures price: what the market is charging today for a specified maturity.
- The future spot price: what the commodity actually trades for when that future date arrives.
- A forecast judgment: how well the futures price predicts that later outcome compared with a chosen alternative.
The third object is an evaluation, not a definition. A futures price does not become a forecast merely because someone compares it with the later spot price. Nor does it become useless information merely because the comparison reveals bias. A quote can help a forecaster rank possible outcomes or identify a changing market state without supplying the exact number that later proves correct. Treating all three objects as synonyms erases the very information that careful analysis is trying to recover.
The diagram captures the central correction. Today’s spot price and current market conditions help define the setting in which a futures quote is formed, but the quote does not mechanically determine the later spot price. It must instead be judged along separate dimensions: incremental information, unbiasedness, and forecasting difficulty. The three judgments can point in different directions. A contract may improve on a simple baseline, fail an unbiasedness test, and still be associated with a period in which every forecast is unusually difficult.
Three questions hidden in one quote
The first question is comparative: does the futures price forecast the later spot price better than a simple alternative? A quote can add value if it improves on a forecast that simply carries today’s price forward, extends recent price movements, or assumes that price changes tend to reverse. This is a claim about relative performance. It says that one method has smaller errors than another under a stated design.
The comparator matters. If today’s price is already a strong short-term baseline, a futures price need not predict a dramatic move to add value; it only needs to improve the estimate relative to holding today’s price constant. If the alternative extends a recent surge, futures may add value by incorporating reasons that the surge will not persist. If the alternative assumes mean reversion, futures may look different because the market is pricing a physical condition that is not captured by a generic reversal pattern. The same futures quote can therefore look better or worse depending on the question it is asked to answer.
The second question is statistical: is the futures price an unbiased predictor? Unbiasedness is stricter than “better than a baseline.” A forecast can beat a weak alternative and still miss in a systematic way. If the futures price is usually too high in one kind of market or too low in another, its average relationship with later spot prices may fail an unbiasedness test even if it contains useful information. Conversely, a failure to reject an unbiasedness condition would not show that every individual forecast is close; it would describe a particular average relationship under a particular sample and regression.
The third question concerns the environment itself: how hard is the market state to forecast? A large gap between futures and spot may be a sign that the curve is carrying more information about unusual conditions. It may also be a sign that the market is facing a supply shock, a scarcity, or a geopolitical risk whose outcome is unusually difficult to predict. A method can therefore be better than its alternatives precisely when all methods are making larger mistakes. Relative improvement is not the same as confidence.
Curve direction and curve magnitude introduce another distinction. When futures prices stand above spot, the relationship is commonly called contango; when they stand below spot, it is called backwardation. Those labels describe the current relationship. They do not, by themselves, say which direction the future spot price must take. The size of the gap can matter for the information available in the curve, but it does not transform the quote into a guarantee. Direction tells us how the two prices are arranged now; magnitude tells us how far apart they are. Neither substitutes for understanding why they are arranged that way.
The practical danger is not that people use futures prices at all. The danger is asking one number to answer several different questions at once. “The market expects oil to be lower next year” may mean that the one-year futures price is below today’s spot price. It may mean that traders have priced a high convenience value for physical supply now. It may mean that hedgers are willing to pay for protection in a particular direction. Or it may be an informal forecast claim that has not been tested against any benchmark. These interpretations are not interchangeable, and each implies a different kind of evidence.
What the evidence must establish
A serious evaluation therefore needs more than a chart of a curve and a later comparison with spot. It needs a benchmark and a defined horizon. The recent Federal Reserve analysis compares futures with three simple backward-looking approaches. One treats today’s price as the best estimate of the future. Another extends recent price momentum. A third assumes that recent movements tend to return toward a longer-run pattern. Each benchmark embodies a different view of how prices evolve without using the futures curve.
This matters because the phrase “the futures market predicts” can hide the alternative being rejected. If futures beat a no-change forecast, that is meaningful, but limited. If they beat a recent-trend forecast, the result says something about the danger of assuming that a shock will persist. If they beat a mean-reversion forecast, the evidence faces a different comparison. The result is strongest when the comparator, period, commodity, and horizon are named. Without those terms, “better forecasting” can sound like a general property when it is really a relationship between two specified methods.
The same discipline applies to the shape of the data. A futures contract three months from maturity is not answering the same question as one twelve months away. Energy, agricultural commodities, and industrial metals do not share identical storage conditions or market structures. A large curve deviation may be common in one market and rare in another. A result observed in an extreme bin may therefore be informative about that state without describing the ordinary experience of every commodity. Sparse observations also make it important not to read an empty comparison cell as evidence that a market never enters that state.
The Federal Reserve note’s reported evidence is designed around this conditional question: when the futures price is meaningfully different from spot, does it improve on simple alternatives? Its conclusion is not that every futures quote is a reliable point forecast. It is that the curve can contain useful information in particular states, while the formal evidence also cautions that futures generally do not produce unbiased forecasts.
That two-part conclusion is more demanding than either of the usual slogans. “Futures are useless” overlooks the possibility that the curve contains information that backward-looking prices miss. “Futures are the market’s expectation” overlooks hedging, physical-market economics, bias, and the difficulty of unusual states. The evidence calls for a third formulation: futures prices are conditional signals whose meaning depends on how they were formed, what they are compared with, and how uncertain the underlying market is.
Before a curve can be interpreted as a signal, its price-forming forces have to be made visible. Storage and carrying costs, inventories relative to consumption, the value of immediate physical availability, and the purposes of the traders in the market all influence the quote. Liquidity can matter as well, especially for contracts far from the present. The next step is therefore not to guess a direction from contango or backwardation, but to understand what the curve is pricing in the first place.