Stock long-term trend prediction method based on CNN-LSTM

Abstract

The invention discloses a CNN-LSTM-based stock long-term trend prediction method. The method comprises the steps: preprocessing a stock data set of a stock market; carrying out data segmentation on the preprocessed stock data set to obtain a first stock sample point data set; performing convolution and pooling operation on each first sample point in the first stock sample point data set by adopting a pre-trained CNN network model to obtain a second stock sample point data set; performing time sequence analysis on the second stock sample point data set by adopting an LSTM network model, and generating a feature vector of a preset dimension; adopting a full connection layer with a preset length and a softmax classifier to classify and score the feature vectors; according to the classification score, deducing the prediction value of the long-term rise and fall amplitude of the stock and the prediction result of the long-term trend of the stock. According to the method, the advantages of high running speed, high accuracy and lower prediction standard deviation in the long-term running trend of the stock are shown through a large amount of data, and the disadvantage of simplification ofshort-term prediction of the stock is avoided.

Description

technical field [0001] The invention relates to the technical field of stock trend forecasting, in particular to a CNN-LSTM-based stock long-term trend forecasting method. Background technique [0002] The origin of the stock market can be traced back to 1602 when the Dutch bought and sold the stocks of the Dutch East India Company, and the formal stock market first appeared in the United States. With the promotion and use of communication technology and electronic equipment, stock trading is becoming more and more international. At present, the global stock market has reached an unprecedented scale, and its fluctuations have a significant and far-reaching impact on the global economy. Stock prices can not only reflect the current political situation and macroeconomic trends, but also sensitively reflect industry prospects, corporate capital supply and demand, and market supply and demand, which can provide managers with a basis for decision-making. Therefore, based on rel...

AI Technical Summary

Problems solved by technology

[0004] At present, most of the application research on CNN and LSTM models at home and abroad focus on short-term price fitting tasks; most of the features used are only historical stock prices and MACD sets, which do not fit well with long-term trend prediction

Images