Research On Quantitative Investment Model Based On Convolutional Neural Network

In recent years,with the emergence of artificial intelligence such as AlphaGo,deep learning is gradually known and triggered a new upsurge of learning.At the same time,China's quantitative investment industry is also developing rapidly,but there is still a large gap with foreign quantitative investment industry.In essence,quantitative investment is to predict the return of financial products such as stocks or futures by modeling financial data,which is similar to in-depth learning,so it is feasible to combine the two organically.By applying deep learning to quantitative investment,on the one hand,it provides a new research idea and modeling method for quantitative investment,on the other hand,it helps investors to predict stock price more accurately,reduce investment risk and obtain investment income.The quantitative investment strategy studied in this paper uses the in-depth learning method of artificial intelligence,which is widely used in financial mathematics.The convolution neural network in this method is a kind of feedforward neural network with convolution calculation and depth structure.It has the ability of representation learning and can classify the input information according to its hierarchical structure.The research object of this paper is from January 1,2008 to December 31,2016 In order to use the data more effectively,this paper divides the data into five training sets and five test sets by means of window scrolling.Then,the convolution neural network is used in this paper Compared with SVM support vector machine,logistic regression and LSTM neural network,this paper constructs the corresponding logistic regression quantitative investment strategy of Shanghai Composite Index Based on convolution neural network.By comparing the model accuracy of convolution neural network with SVM,logistic regression and LSTM neural network,we find that the prediction effect of convolution neural network is better than other 3 In addition,based on the convolution neural network,this paper constructs a quantitative investment strategy for the Shanghai Composite Index.During the backtesting period from 2017 to 2018,the strategy obtains 124.32%of the total return,with an annualized return of 24.32%,sharp ratio of 2.80 and a maximum withdrawal of-3.30%.The performance is better than the one-time investment strategy as a comparison.Therefore,the convolution neural network designed in this paper is an effective quantitative investment model.The possible innovations of this paper are as follows:first,there are not many researches on the application of convolutional neural network to the prediction of stock price and the quantitative strategy.At present,there is no research on the application of convolutional neural network to the prediction of stock price rise and fall in China,so this paper is an exploration and innovation in this field.Secondly,a new feature framework method is proposed.Firstly,the realized skewness is added as a parameter in this paper,which is not used by other research institutes.Secondly,the LSTM cyclic neural layer is added to the traditional convolution neural network,which fully displays the advantages of two kinds of deep learning models.Finally,this paper abandons other research methods to construct convolution network The data of stock market is transformed into K-line graph or other three-dimensional image as input),and one-dimensional convolutional neural network is used,and the data used is not three-dimensional image,but 5*5 data array.Of course,there are some shortcomings in this paper.One is the problem of model parameters.This paper does not discuss the prediction results of super parameters such as the number of convolution kernels,the size of convolution kernels,the number of neurons in the full connection layer,etc.,because various Super parameters may affect each other,thus affecting the accuracy of the model;the second is the number of samples.Due to the particularity of the stock market,the size of the daily average data is also limited;third,the problem of strategy selection.The quantitative strategy of this paper is not perfect,the control group is relatively simple,there is still a lot of room for improvement.