Seasonality In Commodity Markets

Fourteen winters out of twenty is a 6% coin flip

Start with the statistics, because they decide how much weight a seasonal pattern deserves. Suppose someone tells you natural gas has risen in 14 of the last 20 winters. That sounds compelling. Run it against a fair coin: the probability of getting 14 or more heads in 20 tosses is about 5.8%. It is suggestive and it is nowhere near conclusive, and it is a long way from the evidence you would want before risking money on a calendar.

Weaken it slightly. Twelve years out of twenty, a 60% hit rate, has a one-tailed probability of about 25% from a coin. One in four random series of twenty coin tosses produces a 60% result. If a seasonal chart shows you 60%, it has told you approximately nothing.

And the samples are always small. A seasonal pattern gets one observation a year, so twenty years of data is twenty data points, not the thousands a strategy backtest would use. This is why seasonality is background context rather than a trading signal, which the app version of this lesson says plainly and is worth repeating: these are tendencies drawn from history, not rules about the future.

The physical calendar underneath the pattern

Seasonality in commodities is not numerology. Unlike a currency pair, a commodity is a physical thing that gets grown, extracted, stored and burned on an annual cycle, and the cycle leaves a fingerprint in price.

• Grains follow planting, growing and harvest. Northern hemisphere corn and soybean supply arrives in the autumn, so the market is at its best supplied in October and at its most uncertain during the summer growing window when weather can still change the crop.
• Natural gas is a heating and cooling fuel, so demand peaks in the coldest and hottest weeks and storage is rebuilt in between.
• Refined fuels follow driving and heating seasons, with a specification change between summer and winter petrol blends that creates its own calendar.
• Industrial metals track construction activity, which slows in northern winters and around long holiday periods in major consuming economies.

The mechanism is the part worth learning. If you can state why a pattern should exist in physical terms, you can also tell when this year breaks it. If you cannot, you are trading a chart of averages.

Gas storage as a clock

Natural gas gives the cleanest example. Roughly from April to October, production exceeds demand in the United States and gas is injected into storage. From November to March, withdrawals exceed injections and inventories fall. The weekly storage figure published each Thursday at 10:30 New York time is the market's running scoreboard on that cycle, and it routinely moves price several percent in a minute.

What makes the seasonal tradable is not the calendar but the level. Entering winter with inventories well above the five-year average is a different market from entering it 15% below, even though the date is identical. The calendar tells you what should happen next. The storage number tells you what buffer exists if it does not.