Open Conference Systems, 50th Scientific meeting of the Italian Statistical Society

Font Size: 
Prediction interval of electricity prices by robust nonlinear models
Luigi Grossi

Last modified: 2018-05-16


It is well known that volatility of electricity prices
estimated through GARCH-type models can be strongly affected by the
presence of extreme observations. Although the presence
of spikes is a well-known stylized effect observed on electricity markets,
their presence has been often neglected and robust estimators have been rarely applied.
In this paper we try to fill this gap introducing a robust procedure to the study of the dynamics of
electricity prices. The conditional mean of de-trended and
seasonally adjusted prices is modeled though a robust estimator
of SETAR processes based on a polynomial weighting function (Grossi and Nan, 2015), while a robust GARCH is used for the conditional
variance. The robust GARCH estimator relies on the extension of the
forward search by Crosato and Grossi (2017). The robust SETAR-GARCH
model is applied to the Italian electricity markets using data in
the period spanning from 2013 to 2015. The purpose of this application is therefore twice:
first, it is possible to enhance the prediction from point to intervals with associated probability levels,
second, we set up a procedure to detect possible extreme prices which are commonly observed in electricity markets.


1. Crosato L., Grossi L.: Correcting outliers in GARCH models: a weighted forward approach.
Statistical Papers (2017) doi:10.1007/s00362-017-0903-y.

2. Grossi L., F. Nan: Robust estimation of regime switching models. In: Morlini I., Minerva T.,Palumbo F. (eds.) Advances in Statistical Models for Data Analysis, pp. 125–135. SpringerInternational Publishing, Switzerland (2015).

3. Lisi, F., Nan, F.: Component estimation for electricity prices: Procedures and comparisons.
Energy Economics 44, 143–159 (2014).

4. Nowotarski J., Tomczyk J., Weron R.: Robust estimation and forecasting of the long-term
seasonal component of electricity spot prices. Energy Economics 39, 13–27 (2013).

5. Percival, D., Walden, A.: Wavelet Methods for Time Series Analysis. Cambridge University
Press. (2000).

Full Text: PDF