Short-term forecasting of electricity pricesa hybrid methodology based on fundamental and statistical analysis

Resumo

The ongoing evolutions that are exhibited in the electricity markets of today have brought up a new competitive environment in which traders and practitioners must adapt their strategies and look for support for the decision-making when operating in the market. This is a result of a combination of several factors, which are the increasing renewable penetration in power systems, ongoing regulatory changes, varying weather conditions, volatile fuel costs, and also the global financial instability. Due to the resulting uncertainty as well as the non-stationary and volatile behaviour that is present in electricity market prices, market agents and practitioners are resorting to electricity price forecasting models for several purposes, some of which are highly relevant in short-term contexts, such as risk management. In order to consider all relevant drivers of electricity prices for said purposes, fundamental-statistical hybrid forecasting models may be utilised. On the one hand, fundamental models are responsible for simulating the market clearing while considering the operation of the market, physical features of the power system, regulatory effects, etc. On the other hand, statistical models take advantage of linear and non-linear trends that are assumed to be repetitive, such as intraday effects, seasonal patterns, etc. However, this particular hybrid model has been mostly utilised in longer-term applications, where they have proved beneficial. Therefore, employing this method in short-term applications would involve different and/or additional considerations that must be ascertained.