Mean-reverting no-arbitrage additive models for forward curves in energy markets

Abstract

In this paper we present an additive no-arbitrage model for energy forward markets capable to exhibit mean-reversion. The model naturally incorporates term structures for both the mean-reversion level and the volatility of forward prices and it is able to reproduce the seasonalities empirically observed in gas and power markets. We also present a method to estimate the model parameters, based on quadratic variation/covariation for the volatility and on constrained maximum-likelihood estimation for the mean-reversion speed and level. We apply this technique to time series of Phelix Base forward products.