The Study Of Stock Market Prediction Based On Deep Learning Networks

In the application field of statistics and finance,the stock market is a nonlinear system with a complex relationship between variables.How to accurately predict its future price or trend is a subject worth studying.The research content of this paper is how to apply the technology of deep learning to the prediction of the stock market.The data selects the stock sequence data of two different markets,high frequency and low frequency.The main work includes the following three aspects:Firstly,the pre-processing of the original data,the establishment of dual LSTM deep learning network for stock price prediction,and the comparison of the prediction results with the traditional AR time series model,the comparison of the experimental results confirmed the advantages of the deep learning model in the prediction accuracy.In addition,the application scope of the prediction model is illustrated by analyzing the prediction cases of two stocks.Secondly,the trend of stock rise and fall is predicted.By improving and enhancing the price prediction network,the convolution is added to carry out deep feature learning,and the convolution-double GRU deep learning network is established.The training results were strengthened while the training parameters were reduced.After the experiment was compared with the logistic model,the more accurate prediction results were obtained,and the real-time prediction was supported.Thirdly,two machine learning strategies for reinforcement learning are used to enhance prediction.The soft voting strategy can realize the automatic selection of super parameters,reduce the pre-training process and increase the reliability of prediction.Double threshold classification is used to reduce the risk of misjudgment and the adaptive process is realized.The experimental results show that the enhanced learning results have been significantly improved.