Quantifying and modeling price volatility in the Dutch intraday electricity market

This paper aims to provide a solid basis for the quantification and modeling of price volatility in the Dutch intraday electricity market. It analyzes price volatility through realized volatility, which is adapted from foundations in quadratic variation theory. Realized volatility is then estimated using differing multivariate linear regression and random forest regression models. We build these models around features pulled from quadratic variation theory, market fundamentals, liquidity, and information asymmetry. Furthermore, we assess the impact of features within the models using permutation feature importance and recursive feature elimination. The models leverage a multi-year dataset from EPEX SPOT containing completed trades of hourly products as well as other complementary data sources. The results of the paper include recommendations for future price volatility research within intraday electricity markets, mainly: (i) strive to utilize order book data to have a clearer idea of how prices settle and true bid–ask spreads, and (ii) increase model robustness by combining modeling efforts to assess DA, ID and balancing market impacts on price. This paper aims to benefit multiple stakeholders namely, academic researchers, industry participants, and European regulators, by providing a structured view on price volatility quantification and estimation for internationalized intraday electricity markets.