Study 004 · Energy systems

The grid pays you to consume, until it doesn’t.

In one month, German hourly electricity moved from -€45.87 to €665.82 per MWh. The transition is not only a story of cheaper generation. It is a problem of matching place, weather and time.

Published 13 July 202643,784 observationsGermany · Denmark · Belgium

A grid can have too much electricity at noon and too little at dinner. The distance between those moments is where flexibility acquires value.

Across 731 German hourly prices collected from aWATTar between 13 June and 13 July, 74 hours were negative. Twenty-five exceeded €250/MWh. The average was €107.16, but the average hides the system operators, batteries, factories and households reacting to the extremes.

-€45.87minimum price / MWhGerman hourly series
€665.82maximum price / MWhsame 31-day window
10.1%negative-price hours74 of 731
3.4%hours above €25025 of 731

The same market can signal abundance and scarcity

Negative prices are not free energy for everyone. They are a wholesale signal that supply and demand are temporarily out of balance, filtered through contracts, taxes, network charges and access to flexible equipment. Spikes carry the mirror image: scarce supply, constrained networks or high demand can make the next unit extremely valuable.

The coexistence of both in a single month is the story. A system with more variable generation needs storage, interconnection, responsive demand and better timing—not simply more annual megawatt-hours.

The energy transition’s scarce commodity is increasingly not electricity itself, but electricity in the right place at the right hour.

Denmark shows the physical side of the price story

We paired the price path with 42,765 minute-level Energinet observations. Solar generation averaged 1,055 MW and reached 4,347 MW. In this window, solar output had a modest negative correlation with reported CO₂ emissions (-0.212) and with net exchange (-0.199).

Those relationships are observational, not causal. Weather, wind, demand, imports and plant availability all move simultaneously. Still, they show why one dataset is insufficient: generation explains more when joined to emissions, exchange, load and price on a common timeline.

Belgium’s next three days contain another kind of flexibility

An Elia load forecast covering 288 quarter-hours ranged from 7,943 MW to 11,179 MW—a 3,236 MW spread, equal to 40.7% of the minimum. That expected daily rhythm is the demand surface against which generation and cross-border flows must be balanced.

For a manufacturer, bank or policymaker, the useful analysis goes further: match prices and load to company locations, industrial sectors, weather, procurement, emissions and financial exposure. That is how a grid signal becomes a business-risk signal without pretending one price spike proves distress.

Responsible interpretationThese three datasets overlap in theme, not perfectly in market design or interval. We align them to explain system dynamics, not to claim that Danish solar caused a German price or Belgian forecast.

Analytical plate

A month between surplus and scarcity

The distribution matters more than the average. Every bar represents a different operational condition.

German price states

Share of 731 hourly observations

Negative
10.1%
€0–250
86.5%
Above €250
3.4%

Analysis path

From grid event to exposure

price / loadgenerationweathercompany & sectorfinancial exposure

Evidence ledger

Three systems, one timeline

aWATTar Germany hourly market prices

731 hourly observations; min, max, mean and threshold counts calculated from collected records.

DE-AWATTAR-2C15

Energinet right-now system data

42,765 minute observations; solar, CO₂ and net exchange fields aligned by time.

DK-ENERGINET-RIGHTNOW-2C03

Elia total load forecast

288 quarter-hour forecast observations; minimum, maximum and spread.

BE-ELIAODS002-DF47

Method & limits

Joined for context, not causal proof.

We summarised the latest broad records available for each source, calculated threshold shares and Pearson correlations for contemporaneous Energinet fields, and retained the time range and source ID with every result.

What this cannot showWholesale prices are not retail bills. Cross-country grids have different market rules. Correlation does not isolate solar’s effect, and the Belgian data are forecasts rather than realised load. Firm-level effects require company and financial joins.