As far as electricity market stakeholders are concerned, it is crucial for electricity price forecasting to be as accurate as possible in order to compete with the major electricity companies. Numerous variables need to be taken into consideration here. A study by the UPV/EHU’s Department of Foundations of Economic Analysis II has confirmed that surges in electricity, that is, significant changes in short-term pricing, are a key component to bear in mind in the models.
Enhanced market price forecasting enabling small electricity companies to be more competitive
A study by the UPV/EHU-University of the Basque Country analyses models designed to improve electricity market forecasting
First publication date: 06/04/2021
It is well known that the electricity market is a lucrative business of major companies. These large companies handle more information than smaller companies in the sector and therefore wield increased market power. "If we succeed in predicting electricity prices as accurately as possible, small companies in the sector may be able to receive more complete information, and that way the market power of the large electricity companies could be slightly reduced and a more competitive electricity market could be achieved," explained Peru Muniain-Izaguirre, researcher in the UPV/EHU's Department of Foundations of Economic Analysis II, and currently a lecturer in Applied Mathematics at the Faculty of Engineering in Bilbao.
“In this work we explored various models to improve electricity market forecasting. When it comes to predicting electricity prices, many variables have to be taken into account. The most important ones are seasonality —prices in summer tend to be lower due to reduced demand, and in winter it is the opposite—, high uncertainty or variability —if the prices vary considerably from one day to the next, it is more difficult to make forecasts, and so each stakeholder in the market is faced with greater uncertainty when choosing its strategy— and surges, which are huge, short-term price changes that are difficult to bear in mind in the models,” pointed out Muniain.
The study focussed on temporary variability and on surges in electricity prices. “Firstly, we tried to predict the uncertainty in electricity prices as accurately as possible by predicting their variability. To do this, we used autoregressive models in which price uncertainty depends on observations of the past,” said the UPV/EHU researcher. At the same time, “we tried to predict electricity prices as accurately as possible. To do this, we built complex models in which many variables are taken into account, but what happens is that when so many variables are used, the estimates are no longer reliable. So estimation methods allowing variables to be automatically selected were used. In other words, the method itself selects the most important variables for each model to predict these electricity prices,” added Muniain.
“Secondly,” said the researcher, “by using different simulations we made probability forecasts to predict a complete price distribution. That way, the prices over the forthcoming few days can be seen fairly reliably. That offers market stakeholders more complete information to enable them to choose their strategies more effectively. So we believe that in the future the use of probability forecasts will become more widespread.”
This work also concluded that electricity prices have a huge autoregressive effect. In other words, the observations of previous days have a great influence when it comes to predicting the prices over the forthcoming days,” explained Muniain. “At the same time, the importance of large, short-term changes was highlighted, in other words, the incorporating of surges into the prediction models. And the fact is that when bearing the surges in mind, we found that the forecasts improved significantly.”
This research was conducted within the framework of the PhD thesis by Peru Muniain-Izaguirre (Erandio, Basque Country, 1991) entitled ‘Forecasting in Electricity Markets/Aurreikuspenak argindar merkatuetan’. It was supervised by Aitor Ciarreta and Ainhoa Zarraga, both of whom lecture in the UPV/EHU’s Faculty of Economics and Business. Part of the research was conducted in collaboration with the University of Duisburg-Essen in Germany.