The Research On Temperature Prediction Model And Pricing Of Weather Derivatives

China has a vast territory and a huge span from north to south.It has been deeply affected by different climates from ancient times to the present.The way to deal with weather risks in our country is usually weather insurance for catastrophic risks,but there are very few countermeasures for general weather risks that are less harmful but also have long-term impacts.In response to general weather risks,weather derivatives were born at the end of the 20th century.With the rapid development of weather derivatives abroad,our country has also begun to study weather derivatives.In this thesis,from the perspective of the weather risk management strategy,we systematically introduce the weather derivatives,and select the weather derivatives based on the temperature index for research.Since the accuracy of temperature prediction is the key to the pricing of weather futures and options,we divide the main content into two parts:temperature prediction and derivatives pricing.In the temperature prediction part,firstly,we conduct a statistical analysis on the daily average temperature of Beijing in the past 50 years,and find that the daily average temperature difference in Beijing obeys a normal distribution.Through further research,we find that the daily average temperature shows periodic changes and has mean reversion,and then a time-varying Ornstein-Uhlenbeck(O-U)mean reversion model including a sine function term is established.Secondly,we determine the random term of the addition model according to the principle of time series decomposition,and use the autoregressive moving average(ARMA)model to fit the non-random term of the addition model,and then establish the addition model of time series decomposition.Finally,the advantages and disadvantages of the two models are compared from both qualitative and quantitative perspectives.The results show that both models have good prediction effect on air temperature.However,the prediction error of the time-varying O-U mean reversion model is larger than that of the addition model.In order to reduce the error,we further improve the model and find that the improved volatility calculation model is better than the original volatility calculation model.In the derivatives pricing part,firstly,we establish options and futures pricing models based on the daily average temperature in Beijing.Secondly,combining the time-varying O-U mean reversion model of the improved volatility calculation method with the addition model,we utilize the Monte Carlo simulation to predict the options and futures pricing of Beijing in January and August 2020 and analyze the advantages and disadvantages of models comparing with the actual pricing.The results show that both models are suitable for the pricing of Beijing’s daily average temperature weather derivatives.In contrast,however,the addition model works better for pricing.Finally,the pricing errors of the two models at different baseline temperatures are researched,and the results show that the pricing errors are relatively small when the baseline temperature is 16℃ to 19℃.