Study On Pricing Options With Deep Learning

As one of the cornerstones of financial derivatives,option is of great significance in the field of risk management with excellent performance in arbitrage and hedging.Traditional no-arbitrage pricing models,including Black-Scholes model,Merton model and Heston model,all adopt stochastic process to fit option pricing process under strict assumptions.However,because of the difference between their hypothesis and the real market environment,they cannot match the option price in the reality accurately.Therefore,starting from the data-driven model,this paper attempts to adopt deep learning algorithm to solve the option pricing problem.This paper investigated the feasibility of pricing European options with timesequencing data processing method and deep learning models,based on Chinese Shanghai 50 ETF options and American S&P 500 options.Four competing models were built to verify the improvement of the 1D-CNN and LSTM models on the option pricing task.Methods like cross-validation and statistical tests were also used to make our experiments more robust.Besides,in order to increase the stability and the interpretability of our pricing models,the ALE method was selected to interpret and analyze the behavior of the deep learning models.The empirical results indicated that,in both 50 ETF option and S&P500 option pricing tasks,the 1D-CNN and LSTM models had significant advantages in forecasting accuracy and robustness under moneyness,trading date or maturity dimension irrespectively.With the help of ALE method,it is proved that the improved performance brought by the 1D-CNN and LSTM models could be attributed to their capability of capturing time-series information and their different emphasis on input features and lags.