Dr Antonis Alexandridis from the School of Mathematics, University of Kent presented a paper “Non-linear non-parametric temperature modeling in the context of weather derivatives pricing” in the Actuarial and Financial Mathematics Conference in Brussels, Belgium.
The study aims to develop a model that accurately describes the dynamics of the daily average temperature in the context of weather derivatives pricing. Using wavelet networks and wavelet analysis, a non-linear non-parametrical method is proposed for modelling the daily average temperature. The decomposition of temperature time-series is performed by wavelet analysis. Next, a wavelet network is constructed in order to learn the underlying dynamics of the temperature. The valuation of the weather contracts is based on the forecast of temperature evolution given by the trained network. The model was evaluated and compared in-sample and out-of-sample in various locations against models previously proposed in literature. The findings suggest that the proposed method significantly outperforms alternative models and can be used for accurate weather derivative pricing.