Abstract
We examine possible accuracy gains from forecast averaging in the context of interval forecasts of electricity spot prices. First, we test whether constructing empirical prediction intervals (PI) from combined electricity spot price forecasts leads to better forecasts than those obtained from individual methods. Next, we propose a new method for constructing PI--Quantile Regression Averaging (QRA)--which utilizes the concept of quantile regression and a pool of point forecasts of individual (i.e. not combined) models. While the empirical PI from combined forecasts do not provide significant gains, the QRA-based PI are found to be more accurate than those of the best individual model--the smoothed nonparametric autoregressive model.
Highlights
Since the deregulation of electricity markets in the 1990s, electricity spot price forecasting has attracted a lot of attention
This paper examines possible accuracy gains from forecast averaging in the context of interval forecasts of electricity spot prices
While there is a significant number of studies on the use of forecast combinations for constructing interval forecasts of economic and financial variables, to our best knowledge, there are no publications where this approach would be tested on the extremely volatile electricity spot price data
Summary
Since the deregulation of electricity markets in the 1990s, electricity spot price forecasting has attracted a lot of attention. We address the above mentioned unresolved issue of constructing PI from combined electricity spot price forecasts We do this by constructing empirical PI—as in Weron and Misiorek (2008)—for two methods of forecast averaging, i.e. simple average and least absolute deviation (LAD), that proved to be robust and accurate in our recent point forecasting study (Nowotarski et al 2014). We propose a new method for constructing prediction intervals using the concept of quantile regression and a pool of point forecasts of individual (i.e. not combined) time series models. The latter can be viewed as a natural extension of LAD averaging to an arbitrary quantile. The first prediction of these models is made for February 11, Matlab code is available from http://ideas.repec.org/s/wuu/hscode.html
Sign-in/Register to view highlights & summary 
Published Version (
Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call