Study on energy market prices developed by INESC TEC

Study on energy market prices developed by INESC TEC

The paper PriceMining - Study on Market Prices using Advanced Techniques of Data Analytics, which came to an end on February, was developed by INESC TEC.

08th April 2019

The paper PriceMining - Study on Market Prices using Advanced Techniques of Data Analytics, which came to an end on February, was developed by INESC TEC’s Centre for Power and Energy Systems (CPES). On the one hand, it aimed to explain the daily market prices while on the other, the volume turnover and the regulation reserve prices. The team that was involved in this study included Carla Gonçalves, Renato Silva Fernandes, Luís Miguel Ribeiro, João Paulo Viana, Jose Villar and Ricardo Bessa.

The project, which was funded by EDP Produção, consisted in applying advanced techniques of data analytics to the historic data of the wholesale electricity market, in order to develop a causal study of the variables that define the price of the daily market, the volume of the tertiary reserve and the corresponding price.

Through segmentation techniques for defining the behavioural patterns and the explanatory variables under linear regression models, LASSO and causal models, the results allowed to improve the chances on the behaviour of those variables and also to find out new data. For example, regarding the price of the daily energy market, the analyses show the relevance of the net consumption in order to explain the price, as well as to show how different production technologies become more or less relevant according to the net consumption value.

Another example is the tertiary reserve case, in which the study concluded that the reverse mobilisation (that is manually carried out by the system operator) can be mostly explained by an estimation of the consumption and production deviation (mostly on the solar and wind technologies), based on the last assigned programme and on the last detected measures that were available at the time. Other variables, which initially would have been considered relevant (such as secondary regulation energy, the system deviation or the ranks of the time scheduling), were of little explanatory value.

The developed methodologies can now be applied to other relevant variables in order to optimise the market strategy of the agents, to reproduce the behaviour of the competition and to improve the offers for those markets. These same technologies can also serve as basis of tools for monitoring and supervising the market behaviour. 

The researchers mentioned in this news piece are associated with INESC TEC.