Long-term day ahead market price prediction - errors

Hi everyone,
please is it possible to forecast day ahead market prices for long term period 10-15 years. If yes, please could you recommend me where to find what is the mean absolute error for the long-term day ahead market prediction?

1 Like

Dear Martina,

thank you for your message and interest in AMIRIS.

As scientists we must state that it is virtually impossible to predict electricity price dynamics over a long-term period. Day-ahead market prices are influenced by a multitude of variables, including but not limited to:

  • Weather, climate and generation from renewable energy sources
  • Electricity demand (load)
  • Composition and availability of the (conventional) power plant fleet
  • Fuel prices (e.g. oil, gas, hydrogen)
  • Policies, support schemes, regulations

Over extended timeframes of decades, the abovementioned factors are challenging to quantify. Thus, any model will have a hard time to precisely predict (i.e. hourly) electricity prices.

However, if more abstract indicators suffice (e.g. annual average prices), one might derive some estimators from an extended scenario analysis with a wide parameter variation study. AMIRIS (among other models) can then be applied to estimate these indicators.

For backcasts, and short-term forecasts, where good estimators for the abovementioned factors exist, AMIRIS can perform quite nicely with an MAE of around 5 €/MWh.

For insights into historical performance, we recommend referring to our recently updated AMIRIS examples, which offer a comprehensive analysis of the German market zone from 2015 to 2019, benchmarking against actual historical prices.

Christoph and Felix

1 Like