Electricity Futures

Electricity breaks the rule every other commodity obeys. You cannot store it — not at any sensible cost — so it has to be made in the same second somebody uses it. Supply and demand cannot be balanced by a warehouse. They have to match continuously, all day, every day.

That does strange things to the price. On a sunny, windy Sunday lunchtime there is more power than anyone needs, and the price goes below zero: producers pay you to take it away. In a cold snap the opposite happens. In Texas in February 2021 the price went from about $30 to the market's ceiling of $9,000 per megawatt-hour, and stayed there for days.

Nobody can trade that. So the real market is built on averages instead. A power future settles against the average price over a whole month, quarter or year of delivery. There are two standard shapes. Baseload covers every hour. Peakload covers only business hours, when demand and prices bunch up. A German "Cal-26 baseload" contract fixes today what the average of every single hour of 2026 will cost.

Everyone along the chain needs this. Power stations sell years of output ahead of time — the same instinct as the carbon allowance hedging story. Factories fix their electricity bill so they can price their own products. Energy suppliers who sold customers a fixed tariff have to buy the power to honour it. And since prices went up roughly tenfold in the 2021–22 energy crisis, so does every finance director who was not paying attention before.

Pricing without storage

For storable commodities, cash-and-carry pins the curve. For power, the future prices expected spot plus a risk premium:

with the expectation built from fundamentals: the merit order (which plant sets the marginal price at each demand level), fuel and carbon costs, renewable buildout, plant retirements, interconnector flows. The marginal-cost anchor makes power a derivative of other commodities: in most European hours a gas plant sets the price, so power ≈ gas/efficiency + carbon×intensity — the spark spread arithmetic from the fuel-switching calculator, running the grid in real time.

The spreads that organise the market

• Spark / dark spread: power price minus gas/coal cost — a gas or coal plant's margin, tradable as a package; "clean" versions subtract carbon.
• Peak/off-peak: the daily shape. Solar buildout has famously hollowed out midday prices (the "duck curve"), turning the old peak into a valley.
• Geographic spreads: national hubs (German, Nordic, French power) diverge with interconnector limits — France's 2022 nuclear outages sent French-German spreads to historic extremes.
• Cap products: futures on price caps/spikes for retailers hedging scarcity events.

The margin-call lesson of 2022

Futures hedging works on margin — and when European power rose 10×, utilities' short hedges (against physical generation they genuinely owned) generated margin calls in the tens of billions before the offsetting physical revenue arrived. Governments extended emergency credit lines to solvent-but-illiquid utilities across Europe. The hedge was right; the liquidity design wasn't.

Modelling an unstorable underlying

Power spot demands its own process family: mean-reverting diffusions with jumps and spikes (fast reversion after scarcity events), regime-switching models (normal vs. scarcity states), and structural "stack" models that simulate the merit order directly. The workhorse reduced form:

with seasonal mean \(\mu_t\) (daily, weekly, annual cycles) and jump process \(N\). Negative prices force arithmetic rather than geometric dynamics. Averaging into delivery periods tames the spikes at the futures level — but option books on hourly shapes live with the full distribution, which is why power optionality (physical assets, cap contracts) is priced with structural models, not Black–Scholes cosplay.

The renewables transformation of the curve

• Cannibalisation: solar depresses prices exactly when solar produces — each added panel lowers the capture price of all panels. Capture rates (achieved/baseload price) are now the key number in renewable finance.
• Shape risk migration: as intermittency grows, value shifts from average price to flexibility — batteries and peakers monetise spreads between hours, and intraday/balancing markets trade what futures can't reach.
• Negative-price economics: subsidised generation with production-based payments bids negative rationally; hours below zero (record-setting each year in Germany) redistribute value to flexible demand and storage.

PPAs — the futures market's shadow twin

Corporate power purchase agreements (a datacenter buying 15 years of a wind farm's output at fixed price) are bespoke long-dated forwards beyond exchange horizons. They price off the futures curve where it exists, plus basis (location, profile/shape, volume risk) modelled off it where it doesn't — the liquid curve is the pricing spine for a contract market now financing most new renewable construction. The AI-datacenter demand wave has turned long-dated power — for two decades a sleepy market — into one of the most-watched curves in commodities.

The formulas above are standard textbook formulations, simplified for teaching. They explain the mechanism — they are not a valuation tool, and they will not reproduce a dealer’s price.