Research On Stock Forecasting Based On Data Driven

Stock market forecasts provide investors with reliable trading signals to help investors develop long-term,stable and profitable investment strategies.At the same time,stock market forecasts play an important role as an integral part of future automated trading.Data-driven stock market forecasting is more reliable and accurate than traditional trading signal generation methods,so it is a hot research direction in the field of stock market forecasting.With the development of intelligent algorithms and data processing,more and more research extracts stock-related data information from various data sources to predict the stock market.Based on many researches,this paper combines intelligent algorithms with stock forecasting,using a variety of data sources and intelligent algorithms to predict the stock market.For long-term stock price forecasting and stock trend forecasting,this paper combines daily sentiment indicators and technical indicators as data sources,and applies them to the long-short memory neural network(LSTM)model to predict the stock market.For the stock sentiment indicators,applying the two-layer bidirectional long-term memory neural network(BI-LSTM)model based on glove word embedding and attention mechanism extracts stock sentiment indicators from social media;in addition,the dimensionality reduction model based on decision tree(DT)and principal component analysis(PCA)is used to reduce the dimensionality of stock technical indicators and extract the main data information.The experimental results show that the proposed model can significantly improve the accuracy of price forecasting and stock trend forecasting.This forecast provides reliable and cost-effective trading information for long-term investment users.For the short-term stock market forecast,this paper proposes a pattern recognition model based on nasNet to identify the four common stock minute k-line patterns.The results of the identification can be used to automatically identify short-term k-line models,predict reliable trading signals,and help investors make short-term high-yield investment strategies.The k-line pattern recognition result can also be an important part of the future automated trading strategy,freeing investors from heavy observation tasks and providing a stable and efficient automated trading strategy.The experimental results show that the pattern recognition model based on nasNet can well identify the k-line pattern,and the recognition accuracy reaches98.6%.For the stock market forecast,this paper also surveyed the mainstream trading software on the market and found that these trading software did not provide stock market forecasts based on historical data and intelligent algorithms.Therefore,based on the realization of the functions of existing trading software,this paper has written a stock virtual application system based on artificial intelligence algorithm,which aims to provide users with reliable prediction results based on historical data,help users build investment confidence and build a more reliable investment strategy.