Output details

Weather Derivatives Pricing: Modeling the Seasonal Residual Variance of an Ornstein-Uhlenbeck Temperature Process with Neural Network

<29> This is the first study that applies artificial neural networks in modeling the variance of the temperature process in the context of weather derivatives. The seasonal variance exhibit various seasonalities and cycle that were successfully captured and modelled by a nonlinear nonparametric neural network. An optimal architecture of a neural network was constructed using a model identification procedure. The proposed model offered additional accuracy in modeling and pricing of weather derivatives and an understanding of the dynamics that govern the variance in the residuals of a temperature process. Research has been taken up by researchers at the Humboldt-Universität, Berlin.