In competitive electricity markets, in addition to the uncertainty of exogenous variables such as energy demand, water inflows, and availability of generation units and fuel costs, participants are faced with the uncertainty of their competitors' behavior. The analysis of electricity price time series reflects a switching nature, related to discrete changes in competitors' strategies, which can be represented by a set of dynamic models sequenced together by a Markov chain. An input-output hidden Markov model (IOHMM) is proposed for analyzing and forecasting electricity spot prices. The model provides both good predictions in terms of accuracy as well as dynamic information about the market. In this way, different market states are identified and characterized by their more relevant explanatory variables. Moreover, a conditional probability transition matrix governs the probabilities of remaining in the same state, or changing to another, whenever a new market session is opened. The model has been successfully applied to real clearing prices in the Spanish electricity market.

A functional time series is the realization of a stochastic process where each observation is a continuous function defined on a finite interval. These processes are commonly found in electricity markets and are gaining more importance as more market data become available and markets head toward continuous-time marginal pricing approaches. Forecasting these time series requires models that operate with continuous functions. This paper proposes a new functional forecasting method that attempts to generalize the standard seasonal ARMAX time series model to the L ² Hilbert space. The structure of the proposed model is a linear regression where functional parameters operate on functional variables. The variables can be lagged values of the series (autoregressive terms), past observed innovations (moving average terms), or exogenous variables. In this approach, the functional parameters used are integral operators whose kernels are modeled as linear combinations of sigmoid functions. The parameters of each sigmoid are optimized using a Quasi-Newton algorithm that minimizes the sum of squared errors. This novel approach allows us to estimate the moving average terms in functional time series models. The new model is tested by forecasting the daily price profile of the Spanish and German electricity markets and it is compared to other functional reference models.

This paper proposes a methodology for determining the optimal bidding strategy of a retailer who supplies electricity to end-users in the short-term electricity market. The aim is to minimize the cost of purchasing energy in the sequence of trading opportunities that provide the day-ahead and intraday markets. A genetic algorithm has been designed to optimize the parameters that define the best purchasing strategy. The proposed methodology has been tested using real data from the Spanish day-ahead and intraday markets over a period of two years with a significant cost reduction with respect to trading solely in the day-ahead market.

This paper provides a focus upon forecasting electricity prices in the medium term (from a few weeks to several months ahead) in which accurate estimates of tail risks, e.g., at the 1%, 5%, 95%, and 99%, are important. Medium term forecasting and risk analysis are important for operational scheduling, fuel purchasing, trading, and profit management. We extend the research on hybrid forecasting methods, which link detailed fundamental price formation models, using optimization techniques and market equilibrium considerations, with econometric recalibration to the time series data. This paper is innovative in its use of quantile regression to undertake the recalibration and provide accurate risk estimates. It is shown that probabilistic outputs from the fundamental model add value over expected value inputs to the quantile regressions and that if the fundamental model is itself well specified to diurnal variation through the inclusion of relevant explanatory variables such as demand or climatic conditions, then it is not necessary to undertake the quantile regressions separately for each hour of the day. A real application of the proposed methodology is successfully tested on the Spanish electric power system, in which the high penetration of intermittent wind generation creates extreme price risks. The hybrid method outperforms the more conventional fundamental model, making particular use of wind generation data in the quantile recalibrations.