Price Drivers
The model inputs with the largest effect on the average day-ahead price forecast.
Bidding Zone
Drivers are per zone. Pick one to inspect.
Current Drivers
Pushing prices up
Pushing prices down
Click a driver to compare its day-ahead level with its seasonal average.
How model effects are calculated
Integrated gradients explain the joint neural price model by measuring changes from a reference input set to the current regional inputs. The reference uses standardized training means. Effects are measured in EUR/MWh and explain the raw forecast. They are read from the model's own calculation and add back up to its forecast, rather than being estimated afterwards from the results.
One model forecasts every zone together, so a driver can come from another zone: an input labelled with a zone prefix, such as FI | Wind power, is that zone's condition affecting the price shown here. Neighbouring supply and demand genuinely move an interconnected market, and the model weighs them directly rather than through a separate context input.
This page averages each input's effect across the current day-ahead delivery day. The labels Strongly and Moderately compare each displayed effect with the strongest current driver. They are relative rankings for this forecast, not fixed price thresholds. Inputs with less than a third of the strongest effect are left out, so a quiet day lists fewer drivers. Closely related inputs may be combined into one driver by adding their model effects.
The underlying values explain the model's calculation, not necessarily a causal effect on the market. Seasonal comparisons shown when inspecting a driver are provided as supporting market context and are separate from model attribution.