Research And Implementation Of Stock Prediction Model Based On Deep Learning

With the rapid development and improvement of China's stock market,lots of companies listed on the stock market.Listed companies can raise funds in the public markets.Institutional investors and individual investors expect to invest in stocks or listed companies,which have room for growth,and they expect to get maximum profit from investment.However,stock investments carry some degree of risk.If the investments are made blindly,they will bring huge losses to investors.Therefore,it is necessary to analyze the factors that affect the movements of stocks in order to make scientific investments.In this thesis,this thesis studies the problem of building a stock prediction model based on machine learning and deep learning.First,we proposed a novel news acquisition method based on Selenium framework,making it easier for news sites built with JavaScript AJAX technology to be crawled.Secondly,this thesis studies previous methods for predicting stock price trends,such as the use of technical analysis methods and fundamental analysis methods.Since news events have a significant impact on stock prices,this thesis explores the impact of multi-category news events on stock price movements.Finally,technical indicator features and multi-category news event features are used for building deep neural networks to predict stock prices trends.And support vector machines are used for comparison experiment.This thesis finds that deep neural networks with technical indicator features and multi-category news event features in the stock price trend prediction problem can achieve good results.The experiments results are analyzed and the possible improvements are proposed.