Methodology

Named sources. Documented processing. Clear limits.

Analysts should know where a figure comes from—and what it is not. PowerUnits.io names its sources and keeps historical observations, illustrative stress tests, scenarios, and forecast-readiness evidence distinct.

Data sources

Three foundations behind the workspace

Market series, weather context, and plant records stay on separate layers—so you can trust what each chart is allowed to claim.

ENTSO-E Transparency Platform

Markets & outages

Demand, day-ahead prices, generation by technology, and outage events are processed into versioned hourly market series where coverage exists.

ERA5 weather (Copernicus)

Weather context

Temperature, wind, and solar proxies help interpret demand and renewables. Raw source artifacts remain separate from normalized country-level analysis layers.

Global Energy Monitor (GEM)

Power plants

Unit-level generation asset records that power the map and fleet views—the spatial backbone for European plant intelligence.

How data enters the product

Controlled ingestion, not an open web scrape

Named source interfaces feed versioned processing jobs. Raw artifacts and normalized analysis layers remain separate, with run reports and explicit join keys supporting traceability.

Why that matters

When you present day-ahead prices, residual load, or plant context to a desk or stakeholder, provenance beats mystery. Bounded reads keep the stack auditable and prevent silent blending of incompatible layers.

Simple pipeline story

Fetch from documented sources (ENTSO-E, ERA5, …)
Retain source artifacts according to the source pipeline
Normalize to hourly, per-country analysis layers
Join weather and drivers with explicit keys
Log every run for traceability

Trust boundaries

Four claim types stay distinct

Historical observations are not forecasts, stress tests and scenarios are illustrative, and readiness is not a live model score.

Measured history

Prices, demand, residual load, and related drivers as observed through documented transforms—ideal for explaining what happened. Available only where source coverage supports the selected market and window.

Forecast Readiness

Evidence and caveats on a versioned snapshot of inputs and coverage. Not a forecast or live model-skill leaderboard.

Stress-Test

Stress-Test changes explicit assumptions to explore directional system pressure. It does not calculate traded clearing prices or guaranteed outcomes.

Scenario

Developing multi-horizon paths expose assumptions and drivers. They are illustrative and do not represent a production forecast.

Want a product-level overview? Explore product surfaces →