Why does January gas cost more than October gas, every single year?
Some curve shapes repeat on the calendar. They are not anomalies, they are not forecasts, and they are not free money — they are the storage model applied to a commodity whose demand or supply arrives in a lump.
Natural gas: a demand lump
Heating demand peaks in winter. Production is comparatively flat. The gap is bridged by storage — gas is injected into underground facilities roughly from April to October and withdrawn roughly from November to March.
So the curve carries a repeating sawtooth. An illustrative strip:
- Oct $2.80 · Nov $3.20 · Dec $3.70 · Jan $3.95 · Feb $3.85 · Mar $3.40 · Apr $2.90 · … · following Jan $3.85
Read the October-to-January gap correctly. It is not a forecast that gas will be more expensive in January. It is the cost of taking gas that exists in October and making it exist in January: financing the inventory, renting the storage capacity, plus whatever the market will pay for winter security. Traders who buy October gas, pay to store it and sell January gas compete that spread down toward the true cost of seasonal storage. The shape is the price of the service.
The single most-watched calendar spread in that market is March against April — the last withdrawal month against the first injection month. It prices whether storage will make it to the end of winter, it can move violently on a cold forecast, and it has earned the nickname the widow-maker honestly.
Agriculture: a supply lump
Grains invert the logic. The crop arrives all at once at harvest and is consumed all year.
Immediately after harvest, inventories are at their annual maximum, convenience yield is near zero, and the curve typically sits at or near full carry from the harvest month outward — the market paying elevator operators to store the crop. As the marketing year runs down and stocks deplete, carry compresses and the curve flattens.
Then there is a feature with no analogue in energy: the old crop / new crop boundary. The US corn marketing year turns on 1 September, so July — the last month of the outgoing year — is a claim on last year's harvest, while December is a claim on a harvest that has not happened yet. September itself straddles the turn, which is why the clean comparison is July against December. They are not two dates for the same pile of grain — they are claims on two different piles.
Illustratively: July corn at $4.85 and December corn at $4.55. That is an inversion of $0.30, and no carry relationship explains it. Carry arithmetic only ever caps how far December can sit above July; it has nothing to say about an inversion, and across this boundary the two prices are not even reading off the same supply. What the spread says is that the remaining old crop is tight while the incoming one is expected to be adequate. Old-crop grain can of course be physically carried from July into December — it exists, and it stores — but doing so pays away the 30 cents and five months of elevator charges, which is precisely what an inverted spread is telling storage operators not to do.
The literacy point
Seasonality is predictable, repeating, and priced. That last word matters. Knowing that January gas is dearer than October gas is not an insight, because everyone knows, and the spread already contains it. What varies — and what the market actually trades — is whether this winter's storage is adequate, whether this crop is large, and how the spread moves relative to what was already priced.
It also means a naïve seasonal chart is a trap. A repeating shape in a curve is not the same as a repeating pattern in returns.
Try it now
- Five years of continuous natural gas is below. Notice that the seasonal curve shape is largely invisible in it — because any single line, stitched front-month or spot assessment, is always standing at the near end and never looking down the strip. The seasonality you can see here is last year's; the seasonality on a curve is next year's.
- Now open the exchange's settlement page and read the next 18 months of gas contracts in order. The sawtooth is immediate and obvious, and nothing on the chart above could have told you it was there.
- Do the same for a grain, and locate the old-crop/new-crop boundary. Say out loud why the spread across that boundary is not a carry relationship.