Search for content and authors
 

Stochastic nonlinear models for power markets

Carlo Lucheroni 

Universita' di Camerino, Piazza Cavour 19/f, Camerino 62032, Italy

Abstract

Price dynamics in electric power markets offer time series behaviours that are difficult to model and to interpret, being these markets very different from the better known stock and bond markets. In particular, combined spiking and reversion of prices to a fixed level haven’t found a commonly accepted modelling frame yet. In Phys. Rev. E 76, 056116 (2007) a mechanism based on the interplay between noise, sinusoidal forcing and nonlinearity was proposed to model electricity prices in the Alberta power market. A stochastic continuous time two-dimensional non-autonomous excitable model – the FitzHugh-Nagumo system – in an unusual parameter regime and in a resonating condition with electricity demand was shown to be able to model well the observed price patterns and to match some interesting statistical and deep patterns of the studied time series, among them the strong presence of time scales and the absence of scaling. Stochastic and coherence resonance studies from physics were used to discuss the mathematical workings of the proposed model.

It turns out that other stochastic nonlinear models, in continuous and discrete time, autonomous and non-autonomous, excitable or threshold-based, in one or more dimensions, can be used to model successfully different aspects of power markets price time series. A discussion of these models and their relationship with some features of existing power markets will be presented, taking together economical, financial, engineering and physics-related aspects of the problem.

 

Legal notice
  • Legal notice:
 

Presentation: Oral at International Conference on Economic Science with Heterogeneous Interacting Agents 2008, by Carlo Lucheroni
See On-line Journal of International Conference on Economic Science with Heterogeneous Interacting Agents 2008

Submitted: 2008-03-15 19:29
Revised:   2009-06-07 00:48