Modelling and Pricing the Weather Derivative

Abstract The aim of this paper is to design a temperature foresting model which is able to model the daily temperature with accurate prediction and determine the most appropriate method to price the weather derivatives. In the temperature forecasting model, Fourier series is used to determine the seasonal component of the daily temperature and the Ornstein-Unlenbeck process is used to model the trend and the residual. Wavelet analysis is used to simulate the seasonal variance of residual in the Ornstein-Unlenbeck temperature process. We also used neural network in order to model the seasonal variance of residual. Our model validated on the last ten year of historical temperature collected from Las Vegas McCarran International Airport traded at Chicago Mercantile Exchange. The results of our model showed that neural network provided a significant improvement on modelling the seasonal variance of the residual which lead to a better estimation on the daily temperature. In the pricing method of weather derivatives, four different types of pricing method are used to price several HDDs call option contract. This paper concluded that Monte-Carlo simulation is the most appropriate method to price the HDDs call option. Normal approximation can also be used to price the HDDs call option for in the money and at the money, but not for out of money