Stock Price Prediction Based On Stack Autoencoder Neural Network

Since the birth of the stock market,it has become an indispensable part of the financial industry,and has always been highly concerned by investors and financial managers.The prediction of stock prices has become the focus of countless researchers and investors.There are many complicated and non-linear factors affecting the stock market,so it is of great theoretical significance and application value to develop a more accurate stock price prediction model.With the advent of the era of big data,computer parallel computing capabilities have developed rapidly,and deep learning in machine learning has become the frontier of applications in the financial field.This article uses the stack autoencoder neural network SAE in deep learning to predict stock prices.First,the stock price time series is extracted by an autoencoder with an unsupervised learning mechanism,and then the autoencoder is trained one by one by using the layer-by-layer greedy training algorithm.Supervised fine-tuning the network to the propagation algorithm.In terms of empirical analysis,the daily data of the Shanghai Stock Index is used as an experimental sample,and it is compared with the ARMA time series prediction model and the BP neural network model.The experimental results show that the stack autoencoder neural network model performs better on various prediction performance indicators and has higher prediction accuracy.In order to further verify the predictive ability of the model,this article re-selects two Shanghai Stock Index daily data with the same time period but different sequences for empirical comparison analysis.The experimental results show that the stock price prediction model based on the stack autoencoder neural network works well and has great practical value and application prospects.