Research On Option Pricing Of Carbon Emission Rights Based On EGARCH-SVR And LSTM-BS

The excessive emission of greenhouse gases and the closely related issue of climate change are increasingly being taken seriously and causing concern since the 1990 s.Major countries in the world are working hard for global climate governance and greenhouse gas reduction.Carbon emission rights trading is an environmental-economic allocation mechanism that can effectively control global carbon emissions while maintaining economic development.Carbon emission rights options,as an important derivative in the carbon trading market,can play the role of discovering price of carbon emission rights,help to effectively manage the risks in the market,and ensure the stability and health of the entire carbon trading market.Therefore,it is necessary to study the pricing of carbon emission rights option.This thesis selects the closing price data of the EUA option Dec23-100 Call and its corresponding EUA futures Dec23 from the ICE exchange,and establishes two models,GARCH-SVR and LSTM-BS,for empirical analysis.In the first model,using the historical data of the past 18 months as the training set,the volatility of carbon emission rights futures is fitted by a EGARCH model,and the Hurst index is calculated by R/S method.Then the parameters in B-S(Black-Scholes)formula under fractal Brownian motion(volatility,risk-free interest rate,futures closing price,expiration date,etc.)are modeled by a SVR model,and then predict the closing price of options for the next 10 trading days.In the second model,using the historical data of the past 18 months as the training set,implied volatility is solved by B-S formula inversely,and LSTM model is used to fit and predict implied volatility for the next 10 trading days.Then it is substituted into B-S formula to obtain closing price of options for next 10 trading days.This thesis uses three error evaluation indicators,MAPE、MAE and MSE,to measure prediction effect.The results prove that errors of both models in this thesis are lower than those of traditional B-S model,while LSTM-BS model performs better.The main contributions of this thesis are: firstly,pricing put options using a combination of the traditional pricing formula of European options and machine learning methods;secondly,combining the traditional pricing formula of European options,modeling and predicting implied volatility as the core,and using deep learning methods to price carbon options.This not only enriches the methods of pricing carbon options,but also provides a basis for rational decision-making by investors and market risk management by managers.At the same time,it can also provide a reference for China to carry out carbon emission rights option trading in the future.