It does not establish causality.
They describe co-movement and context; they do not establish causality or predict tomorrow. Ranking a deviation is not explaining a cause.
European power markets
Understand what moved a European power market
A ranked comparison of measurable market conditions against that market's own recent normal, with the observed outcome for the same hours. Observed history, not a forecast and not a cause.
When a European power market moves, the useful question is not only what the price did, but which measurable conditions sat furthest from that market's own recent normal in the same hours — and on what evidence that ranking rests.
PowerUnits answers that question as a panel at the top of the Markets Today workspace surface. It is not a standalone application.
Definition
What a power market driver explainer is
A power market driver explainer is a structured comparison of measurable market conditions against a baseline, so that a reader can see which conditions moved furthest in a chosen period and on what evidence that comparison rests.
PowerUnits ranks eleven defined drivers by how far the period average sat from that market's own preceding 720-hour normal, and shows the observed outcome for the same hours beside the ranking.
The readout reports co-movement and context. It explicitly declines causal attribution and monetary attribution: no calibrated price model stands behind these numbers, so no euro contribution is shown.
Evidence levels
Four levels of explanation, and where this sits
Not every explanation of a market move is the same kind of claim. PowerUnits publishes a four-level ladder so that the current product position is explicit, and so that levels it does not implement are not implied by the ones it does.
| Level | What it claims | PowerUnits status |
|---|---|---|
| 1 — Observed driver change | Reports that a measurable condition changed in the chosen period, without comparing it to that market's own recent normal. | Not the PowerUnits position |
| 2 — Deviation from the market's own normal, ranked | Ranks defined drivers by how far the period average sat from that market's own preceding 720-hour normal, alongside the observed outcome for the same hours. | PowerUnits is here |
| 3 — Statistical attribution with a fitted contribution | Assigns a fitted statistical contribution of each driver to the observed outcome. | Not implemented — PowerUnits reports co-movement, not a fitted attribution |
| 4 — Calibrated monetary contribution | States how many euros per megawatt-hour each driver contributed to the outcome. | Not implemented — no calibrated price model stands behind these numbers, so no euro contribution is shown |
Readout
What a driver readout contains
A driver readout is a dated comparison for one market and one chosen period. The elements below describe what the panel contains; they are not a live market result and are not tied to a specific calendar window on this page.
| Element | What it shows |
|---|---|
| Market and period | The selected European market and the analysis window under comparison. |
| Ranked drivers | Eleven defined drivers ordered by deviation from that market's own 30-day normal. |
| Observed outcome | The measured market outcome for the same hours, shown beside the ranking for context. |
| Euro contribution | Not shown. No calibrated price model stands behind these numbers, so no euro contribution is shown. |
| Withheld drivers | Generation-derived drivers currently withheld under the ADR 037 calibration gate are named rather than silently omitted. |
Method
Deviation from the market's own 30-day normal
For a chosen market and period, PowerUnits compares each defined driver's period average with that market's own preceding 720-hour normal — roughly a 30-day baseline owned by the same market, not an average across all markets.
Drivers are then ranked by how far the period sat from that normal. The observed outcome for the same hours is shown alongside the ranking so the reader can see co-movement and context: they describe co-movement and context; they do not establish causality or predict tomorrow. Observed history, not a forecast and not a cause.
Disclosure
The drivers we withhold rather than show
Under the ADR 037 generation-calibration gate, PowerUnits withholds generation-derived drivers rather than show them uncalibrated. Serving a known-wrong generation series would be worse than withholding it.
The reason is concrete: the v1 aggregation summed sub-hourly intervals instead of averaging them, so hourly generation was overstated by roughly four to five times. Until that calibration is corrected, generation-derived values stay out of the ranking.
Withheld items are named to the user rather than silently omitted. The panel makes clear which drivers can currently be compared and which cannot.
| Withheld item | Status | Reason |
|---|---|---|
| Residual load, residual-load share and residual-load ramp | Withheld | Generation-derived; ADR 037 calibration gate |
| Renewable share | Withheld | Generation-derived; ADR 037 calibration gate |
| Thermal share | Withheld | Generation-derived; ADR 037 calibration gate |
| Wind and solar share of demand | Withheld | Generation-derived; ADR 037 calibration gate |
| Three dependent rule flags tied to the withheld generation series | Withheld | Dependent on withheld generation-derived inputs |
Coverage
Which markets and what freshness
The Driver Explainer panel is available across the 17 European markets in the shared allowlist: Germany, France, the Netherlands, Belgium, Spain, Italy, Poland, Sweden, Austria, Czechia, Finland, Hungary, Slovakia, Romania, Switzerland, Norway and Great Britain.
Data reaches the analysis with a pipeline lag of several hours. That lag is qualitative: it is not an SLA, and the surface is not presented as an immediate feed.
Great Britain is the explicit coverage gap: GB has no rows in the model dataset view, is served by a separate source path, and currently returns an honest empty driver pack rather than an invented ranking.
| Market | Driver pack | Freshness note |
|---|---|---|
| Germany | Available | Pipeline lag of several hours — qualitative only |
| France | Available | Pipeline lag of several hours — qualitative only |
| the Netherlands | Available | Pipeline lag of several hours — qualitative only |
| Belgium | Available | Pipeline lag of several hours — qualitative only |
| Spain | Available | Pipeline lag of several hours — qualitative only |
| Italy | Available | Pipeline lag of several hours — qualitative only |
| Poland | Available | Pipeline lag of several hours — qualitative only |
| Sweden | Available | Pipeline lag of several hours — qualitative only |
| Austria | Available | Pipeline lag of several hours — qualitative only |
| Czechia | Available | Pipeline lag of several hours — qualitative only |
| Finland | Available | Pipeline lag of several hours — qualitative only |
| Hungary | Available | Pipeline lag of several hours — qualitative only |
| Slovakia | Available | Pipeline lag of several hours — qualitative only |
| Romania | Available | Pipeline lag of several hours — qualitative only |
| Switzerland | Available | Pipeline lag of several hours — qualitative only |
| Norway | Available | Pipeline lag of several hours — qualitative only |
| Great Britain | Empty pack — separate source path; no rows in the model dataset view | Honest empty result rather than an invented ranking |
Limitations
What this does not establish
It shows no euro attribution.
No calibrated price model stands behind these numbers, so no euro contribution is shown. A ranked deviation is not a monetary contribution.
It is observed history, not a forecast.
Observed history, not a forecast and not a cause. The panel does not predict the next period.
Some drivers are withheld on purpose.
Generation-derived drivers are withheld under ADR 037 rather than shown uncalibrated. The absence is disclosed, not hidden.
It is a panel, not a product of its own.
The Driver Explainer sits at the top of Markets Today. It is not a standalone application and should not be described as one.
FAQ
Common questions
What drives European power prices?
Measurable conditions such as demand, renewable output, thermal availability, fuel and carbon costs and cross-border exchange can all sit far from a market's own recent normal in the same hours as a price move. PowerUnits ranks those defined drivers by deviation from that normal. The ranking reports co-movement and context; it does not establish which condition caused the price to move.
What is a 30-day normal?
In PowerUnits it is the market's own preceding 720-hour baseline — roughly thirty days of that market's history — against which the chosen period's driver averages are compared. It is not an average across all markets and not a seasonal forecast.
Why does PowerUnits not show how many euros each driver contributed?
Because no calibrated price model stands behind these numbers, so no euro contribution is shown. Showing a euro figure without that model would invent a precision the product does not have.
Does this explain why the price moved?
No. They describe co-movement and context; they do not establish causality or predict tomorrow. Observed history, not a forecast and not a cause.
Which markets are covered, and how fresh is the data?
Seventeen European markets are in the allowlist. Data reaches the analysis with a pipeline lag of several hours — qualitative only, never an SLA, and never described as real-time. Great Britain currently returns an honest empty driver pack because it is served by a separate source path with no rows in the model dataset view.
Why are some drivers missing from the ranking?
Generation-derived drivers are withheld under the ADR 037 calibration gate rather than shown uncalibrated. The v1 aggregation summed sub-hourly intervals instead of averaging them, overstating hourly generation by roughly four to five times. Withheld items are named to the user rather than silently omitted.
Driver Explainer
Understand what moved a European power market
A ranked comparison of measurable market conditions against that market's own recent normal, with the observed outcome for the same hours. Observed history, not a forecast and not a cause.
Open Markets Today