Commodity Calendar Spreads

Thesis (edge)

A calendar spread is long one delivery month and short another in the same commodity. If crude oil rises 5 dollars across the whole curve, you make nothing and lose nothing, because both legs moved together. What you are actually trading is the shape of the curve: whether near-term barrels get more or less expensive relative to barrels three months out.

That shape is not random. It is set by physical reality. When storage tanks are full, nobody wants a barrel today, so the near contract trades cheap relative to the deferred, and the curve slopes upward. When inventories are low and buyers are scrambling, the near contract trades at a premium and the curve slopes downward. The spread is a live read on how tight the physical market is.

There are two ways to trade this. The trend or carry version says the curve shape persists, so lean into it. The mean reversion version says the spread has bounds set by the cost of storage, so fade it when it stretches too far. Both are legitimate, and which one works depends on the commodity and the time frame.

The clean logic behind the mean reversion version is worth stating. The upward slope cannot get steeper than the cost of storing the commodity, because if it did, anyone could buy the physical product, store it, sell the deferred contract and lock in a profit. That gives a real economic ceiling. There is no matching floor on the downside, which is why backwardation can go to extremes that look absurd and keep going.

Where it works (regimes)

It works when the physical market is doing something clear: inventories filling up, a supply disruption, a seasonal squeeze in natural gas heading into winter. In those moments the spread tells a story and it tends to keep telling it.

It fails when a spread that was well behaved for a decade gets caught in a squeeze. Someone with a large physical position stands for delivery, the front month detaches from reality, and a spread that normally moves a few cents moves several dollars. Because spreads are usually calm, traders run them with heavy leverage, and that is exactly the combination that destroys accounts.

Signals

• The spread level itself: front price minus deferred price, quoted consistently.
• Compare that level to the cost of carry: interest plus storage plus insurance. A spread wider than full carry is theoretically bounded and often a fade.
• Compare the spread to its own seasonal history. Natural gas in October is not the same market as natural gas in April, and a signal that ignores the calendar will be badly wrong.
• Inventory data where available, since it is the underlying driver rather than a proxy for it.
• The spread's own momentum, which tends to work in energy, where tightness builds and persists.

Portfolio construction

Size by the volatility of the spread, not the outright. This is the single most common mistake. A crude oil calendar spread might move a tenth of what the outright contract moves on a normal day, which tempts people into holding ten times the contracts. That works until a squeeze, when the spread suddenly moves more than the outright.

Diversify across commodities and across parts of the curve. A book of eight spreads in different commodities is a very different risk profile from four spreads in energy, which is really one bet on energy inventories.

Never let a spread's near leg run into delivery unless you have a physical operation and know exactly what you are doing. Roll or close well before first notice day.

Risk model

Model the squeeze, not the average day. Use expected shortfall rather than volatility, and use it on the spread's worst historical moves rather than its typical ones.

Set a leverage cap in risk terms and enforce it regardless of what the exchange margin says. Exchanges give large margin offsets for calendar spreads because the two legs usually move together. That is a statement about normal conditions and it is not a risk limit.

Costs & implementation

Deferred contracts are much less liquid than the front month. The screen may show a tight market, but the size behind those quotes is small. Where the exchange lists the spread as its own tradeable instrument, use it, since you get a single fill with no leg risk. Where it does not, executing two legs separately exposes you to slippage between the fills, and that slippage can exceed your expected edge on a tight spread.

Capacity is modest. This is a strategy that works well at moderate size and degrades quickly when you try to scale it.

Failure modes

• Sizing by margin instead of by risk, then getting destroyed in a squeeze.
• Holding the near leg too close to delivery.
• Ignoring seasonality and fading a spread move that is simply the normal seasonal pattern.
• Assuming deferred contract quotes are tradeable in size.
• Treating a spread as low risk because it is usually quiet.

Our Notes & Suggestions

This is the most technically demanding strategy in the trend and carry family, and the one where a small operational mistake causes the biggest loss. Start with a small number of liquid spreads and learn the physical market behind them. The traders who consistently make money here understand storage, refinery schedules and crop cycles, not just the price series.

Backtest against the physical data, not just prices. If your signal is supposed to be about inventories, check that it actually correlates with inventories. If it does not, you have built something else and you do not know what.

Above all, respect the tail. The spread's daily volatility is a comfortable, reassuring, misleading number, and sizing off it is how people blow up in this trade.