Quantitative Trading Strategy For On-site Options Based On LSTM-LightGBM

With the Shanghai-Shenzhen 300 stock index option contract——the first index option product listed on my country's capital market,officially listed on the China Financial Futures Exchange,more stock options will appear in my country's capital market in the future.At the same time,in recent years,machine learning has been used in more and more investment channels,and the development of quantitative investment has become more and more rapid.Therefore,how to help investors use quantitative investment methods to obtain stable income in the turbulent stock market option market has become a problem that the capital market will face.Relevant literature and historical experience show that the important basis for using individual stock options to make a profit is to accurately judge the volatility trend of the underlying stock.However,non-linear prediction methods based on simple statistics and traditional stock price technical analysis are used to predict stock prices.Most of the prediction results are lagging and the prediction effect is not ideal.Wavelet analysis is very effective for decomposing financial time series such as stock price fluctuations that are too sharp and the trend is not obvious.The predecessors in the academic world performed wavelet decomposition and denoising on the original stock price series,and used the ARIMA model to fit and predict the low-frequency coefficients.The dynamic gray prediction or SVR model is used to fit and predict the high-frequency coefficients.Finally,the method of reconstructing the coefficients to obtain the predicted stock price improves the accuracy of the stock price prediction.This paper proposes a prediction method based on wavelet transform,LSTM and LightGBM to predict the underlying asset price of options.As an example,the Shanghai 50 ETF option transaction uses the "db4" wavelet function to perform three-level decomposition of the Shanghai and Shenzhen 300 and the Shanghai 50,which represent the market,to obtain high-frequency data and low-frequency data.After denoising the soft threshold of high-frequency coefficients,LightGBM is used respectively.Rolling prediction of high-frequency data and LSTM for low-frequencydata.The result of reconstruction of the predicted values of the two is the final prediction result of the model,that is,the predicted value of the underlying asset price of the option.This paper conducts an empirical analysis of Shanghai Stock 50 and Shanghai and Shenzhen 300 to verify the feasibility of the model proposed in this paper.At the same time,it compares the prediction effect of the industry mainstream model with the model proposed in this paper to verify the validity of the model proposed in this paper.As an example,the prediction result of Shanghai Securities 50 is 82%,which is 1% higher than the mainstream model in the next 7 days.The accuracy of the forecast for the next 15 days is 78%,which is 16% higher than the mainstream model.The prediction accuracy rate for the next 30 days is 74%,which is20% higher than the mainstream model.Therefore,the improvement of the prediction method in this paper is effective.This article divides the market based on the volatility trend of the Shanghai and Shenzhen 300,selects different market options trading strategies according to different markets and builds the corresponding option portfolio.The specific division is as follows: December 1,2013 to June 25,2015 is a bull market,June 25,2015 to December 31,2015 is a bear market,and January 1,2016 to January 29,2018 is a shock The city is a bear market from January 29,2018 to December 2019.The specific strategies are as follows: the strategy of shocking the market: selling put options with the highest price on the day,holding them until the exercise date,expiration dates,and rolling transactions every day;the bull market strategy: buying a first real-value call option At the same time,sell two first-stage false-value call options;the strategy of the bear market: buy one first-stage real-value put option and sell one first-stage false-value put option at the same time.The results of the strategy are as follows: The annualized return of the shock market is 8.92%;the annualized return of the bull market is 47.96%;the optimal performance of the annualized return of the bear market is 7.08%.