Research On The Strategy Of Stock Timing In Quantitative Trading

In recent years,Chinese economy grows rapidly,the securities market continues to expand and the stock market has ushered in an unparalleled period of prosperity,which leads to investors facing with the unprecedented opportunities,as well as the high risks and high challenges.Due to the large number of investors and the convenient way of trading,the stock market generates a lot of valuable information.Massive data has been created in each trading day,which constitutes one of the main bodies of investment analysis.In the Chinese Stock Market,quantitative investment strategies based on computer and large data are emerging,and the quantitative model has increasingly become the mainstream tool to guide market investment.At the same time,the nonlinear and complexity of the stock market make the traditional investment strategy difficult to reach people's expectations.It is found that the artificial neural network possesses some characteristics such as strong non-linear approximation ability,self-learning and self-adaptive.Therefore,the superiority of artificial neural network in stock market prediction is becoming more and more obvious.This paper selects a total of 1,000 trading days,15 indicators of the Shanghai Composite Index data,from December 21,2012 to December 30,2016.First of all,the principal component analysis is used to reduce the dimension of input parameters,as the results,5 principal component variables are obtained.Then,the time series ARIMA-GARCH model and the BP neural network prediction model for the Shanghai Composite Index are established.However,some inherent defects of BP neural networks such as slow convergence rate of learning and the uncertainty of network structure restrict its prediction accuracy and speed.Therefore,in this paper,the genetic algorithm is used to optimize the neural network and a neural network model based on genetic algorithm is established.Finally,the relative error is taken as the measurement index,and the model results are compared and analyzed.The results show that the neural network is superior to the ARIMA-GACH model in the prediction of stock price,moreover,the neural network optimized by genetic algorithm is also improved in the speed and precision compared with the general neural network.