How Day-Ahead Electricity Markets Work

A day-ahead market is an auction, held once each day, that sets a price and a schedule for every hour of the following day. Sellers submit offers saying how much electricity they will produce at what price, buyers submit bids saying how much they will take and up to what price, and a computer finds the price at which the two sides match in each hour. That price then settles the trades, and the resulting schedule tells the grid operator what to expect.

Why the market runs a day early

Electricity has to be produced at the moment it is used, but the machines that produce it cannot appear at the moment they are needed. A large thermal plant takes hours to warm up and has a minimum time it must stay on once started. Water in a reservoir used today cannot be used tomorrow. Even flexible plants have costs attached to starting and stopping.

So the industry settles on a two step arrangement. Most of tomorrow’s electricity is traded today, giving everyone time to prepare, and the remaining adjustments are made in the hours and minutes before delivery. The day-ahead auction is the anchor of that arrangement. It is where the bulk of the volume trades, and its prices are the reference that contracts, hedges and public commentary all point at.

Inside the auction

Every participant submits orders before a fixed deadline, usually around the middle of the day. After that deadline nothing can be changed.

A sell order says: in hour 14, I will produce up to 400 megawatts if the price is at least 42 per megawatt hour. A buy order says: in hour 14, I will take 300 megawatts if the price is no more than 90. Participants can submit many such orders in steps, building a curve rather than a single point.

For each hour the algorithm stacks all the sell orders from cheapest to most expensive, producing a rising supply curve, and stacks all the buy orders from highest willingness to pay downwards, producing a falling demand curve. Where the two cross is the clearing price and the cleared volume. Every sell order below the crossing point is accepted and every buy order above it is accepted. Orders on the wrong side are rejected.

Then comes the part that surprises people new to the subject: everybody accepted trades at that single crossing price. A wind farm that offered at zero and a gas plant that offered at 80 both receive the clearing price if the clearing price is 85. This is called uniform pricing, and the reasoning behind it is explained in our explainer on merit order and marginal pricing.

Real auctions add complications on top of this picture. Plants that cannot run for a single hour in isolation can submit block orders covering a run of consecutive hours, accepted or rejected as a whole. Some orders carry conditions about recovering start up costs. These make the problem harder to solve, which is why the results take a little time to publish after the deadline.

Hours, and increasingly quarter hours

Traditionally each auction period was one clock hour. That was a reasonable match to how conventional plants were scheduled, but it is a poor match to how wind, solar and demand actually move, which is continuously.

Markets have therefore been moving to shorter periods, with products of fifteen minutes alongside or instead of hourly ones. A finer period lets the price follow the real shape of a sunrise or a wind ramp instead of averaging it away, and it reduces the mismatch the grid operator has to clean up within the hour.

The same shift is visible in the data we publish. Grid operator feeds are reported hourly on our grid zone pages, which is enough to see the shape of a day but not the swings inside an hour.

Coupling: many zones, one calculation

A single country’s auction in isolation would waste the interconnectors between countries. If one area clears at 30 and its neighbour at 90, power should flow from the first to the second until either the difference disappears or the cable is full.

Market coupling does exactly that, automatically. Instead of each area clearing alone, the coupled areas are cleared in one calculation, with the available transfer capacity on every border entered as a constraint. The algorithm maximises the total value of the accepted trades across the whole coupled region, subject to those limits.

The results follow a simple pattern. When a border has spare capacity, the two zones on either side clear at the same price, because the algorithm has moved power until they met. When the border is full, the prices separate, and the size of the gap shows how much the missing capacity is worth in that hour.

This is why maps of European prices show large blocks of identical colour on calm ordinary days and a patchwork on stressed ones. It also means cross border capacity is allocated without anyone trading it separately. It is bundled into the energy auction itself, which is why the arrangement is called implicit allocation.

What happens after the auction closes

The day-ahead result is a plan, not reality. Forecasts move, plants fail, and demand surprises.

Intraday markets run continuously through the delivery day so that participants can buy or sell the difference between their day-ahead position and their latest expectation. A wind operator whose forecast has dropped buys back the shortfall. A plant that has tripped does the same.

Whatever remains unbalanced at the moment of delivery is the grid operator’s problem, and it is solved through the balancing market, where the operator pays for output to be raised or lowered in real time. Those costs are charged back to the parties whose positions turned out to be wrong.

American markets are organised slightly differently. A regional market such as PJM also runs a day-ahead auction, but it commits and dispatches individual units against a detailed model of the transmission network, and produces prices at thousands of individual points rather than one price for a large zone. The principle is the same. The resolution is finer.

What this site publishes

World Power Monitor does not yet publish European wholesale prices. Our Europe page and country pages carry annual and monthly electricity statistics compiled by Ember, and our grid zone pages carry hourly operating data for United States grids. The methodology page sets out which layer each figure comes from.