Reading the Market
Has the Regime Changed?
The real risk is not volatility (or the lack of it).
It is delayed recognition.
Energy markets are too complex to read from raw prices alone. But they are also too important to flatten into a handful of lazy stories.
What we need are better heuristics: patterns, state descriptions, disciplined language that compresses the market without pretending it is simpler than it is.
Energy markets do not move through time as a smooth series of prices. They move through states.
Some days are quiet.
Some are all shape.
Some are tight.
Some are stressed in reserve.
Some are transitional, a few things are moving, but not enough to justify a cleaner read.
Market commentary usually handles those states loosely. One day is “volatile”, another is “tight”, a third is “dislocated”.
The language is impressionistic. Isolated price spikes and duck curves tell us little. Movement is easy to see; change is harder to recognise.
Let’s introduce a framework.
Why Classify?
Because a number on its own is rarely enough.
If day-ahead spread is EUR 114/MWh, is that unusual? If aFRR- is elevated, is that a one-off block effect or part of a broader reserve-pressure state? If FCR is high but day-ahead is not, are we seeing underlying system tightness or narrower reserve strain?
Those are regime questions. A regime classifier is not mainly about prediction. It is about compression.
It takes a larger field of prices and turns it into a smaller, more stable question:
What state is the market expressing?
That answer changes how you read everything else:
whether a spread move is structural or tactical
whether reserve prices are confirming or diverging from the day-ahead story
whether a quiet day is truly quiet or merely transitional
whether a market note should be read as distortion or regime shift
The regime label is not meant to explain the whole world. Fuel headlines and geopolitics sit in the background. But the first task is simpler:
Identify the state the market is actually expressing before attaching a larger story to it.
Keep It Simple
There is an obvious temptation to make a regime classifier too clever.
You can easily pull in:
ID
reBAP
residual load
forecast error
renewables by type
cross-border position
activated balancing volumes
All of those matter.
But the cleaner starting point is narrower and easier to build awareness around. Our first version uses four core layers:
DE-LU day-ahead prices
German aFRR capacity prices
German FCR capacity prices
SPARX (dislocation layer)
That is already enough to produce a meaningful state description without turning every run into an essay-length calculation.
How the Classifier Works
It computes daily features from those four layers.
Day-Ahead (Shape)
Daily average
Daily minimum
Daily maximum
Daily spread
This captures level — but more importantly, shape.
aFRR Capacity (Reserve Stack Intensity)
Positive-side average
Positive-side marginal
Negative-side average
Negative-side marginal
Marginal-over-average premium
Reserve markets are not just about price level. They are about the steepness of the accepted stack.
FCR Capacity (System Backdrop)
Germany average
Germany maximum
Cross-border average (secondary context)
FCR helps distinguish ordinary shape from broader reserve pressure.
SPARX (Expectation vs Settlement Gap)
Raw SPARX level
Percentile rank
Direction of change
SPARX captures the gap between the day-ahead expectation and short-term realised reality.
It is neither a day-ahead surface nor a reserve surface.
It is the dislocation layer.
