• Energy Research

This paper develops a new approach to short-term electricity forecasting by focusing upon the dynamic specification of an appropriate calibration dataset prior to model specification. It challenges the conventional forecasting principles which argue that adaptive methods should place most emphasis upon recent data and that regime-switching should likewise model transitions from the latest regime. The approach in this paper recognises that the most relevant dataset in the episodic, recurrent nature of electricity dynamics may not be the most recent. This methodology provides a dynamic calibration dataset approach that is based on cluster analysis applied to fundamental market regime indicators, as well as structural time series breakpoint analyses. Forecasting is based upon applying a hybrid fundamental optimisation model with a neural network to the appropriate calibration data. The results outperform other benchmark models in backtesting on data from the Iberian electricity market of 2017, which presents a considerable number of market structural breaks and evolving market price drivers.

Various power exchanges are nowadays being affected by a plethora of factors that, as a whole, cause considerable instabilities in the system. As a result, traders and practitioners must constantly adapt their strategies and look for support for their decision-making when operating in the market. In many cases, this calls for suitable electricity price forecasting models that can account for relevant aspects for electricity price forecasting. Consequently, fundamental-econometric hybrid approaches have been developed by many authors in the literature, although these have rarely been applied in short-term contexts, where other considerations and issues must be addressed. Therefore, this work aims to develop a robust hybrid methodology that is capable of making the most of the advantages fundamental and the hybrid model in a synergistic manner, while also providing insight as to how well these models perform across the year. Several methods have been utilised in this work in order to modify the hybridisation approach and the input datasets for enhanced predictive accuracy. The performance of this proposal has been analysed in the real case study of the Iberian power exchange and has outperformed other well-recognised and traditional methods.

Traders and practitioners in diverse power exchanges are nowadays being most exposed to uncertainty than ever. The combination of several factors such as renewable generation and regulatory changes calls for suitable electricity price forecasting models that can deal with complex and unusual market conditions. Several authors have proposed combining fundamental approaches with econometric models in order to cover all relevant aspects for electricity price forecasting. This combination has shown positive results for medium-term horizons. However, this approach has rarely been carried out for short-term applications. Moreover, several day-to-day applications in electricity markets require fast responsiveness and accurate forecasts. All of these facts encourage this work s short-term hybrid electricity price forecasting model, which combines a cost-production optimisation (fundamental) model with an artificial neural network (econometric) model. In order to validate the advantages and contributions of the proposed model, it has been applied to a real-size power exchange with complex price dynamics, such as the Iberian electricity market. Moreover, its forecasting performance has been compared with those of the two individual components of the hybrid model as well as other well-recognised methods. The results of this comparison prove that the proposed forecasting model outperforms the benchmark models, especially in uncommon market circumstances. info:eu-repo/semantics/draft