Research On Stock Price Prediction And Stock Selection Strategy Based On CBAM-LSTM-Attention Hybrid Model

As an important part of the financial market,the stock market provides investors with investment channels.With the development of stock trading technology,stock trading is becoming more and more convenient and more and more participants.By February 25,2022,the number of investors in China’s securities market has exceeded 200 million.Although there are a large number of investors participating in the stock market,most of them do not have rich experience in stock trading.In the process of stock trading,they are easy to be affected by the wrong subjective emotions,leading to the phenomenon of chasing winners and selling losers,which will not only bring economic losses to investors.It is also bad for the long-term health of the stock market.Therefore,the purpose of this paper is to predict the future stock price by using technical means,and build investment strategies based on this,and provide investment suggestions for investors.To a certain extent,this avoids the behavior of stock investors,and also indirectly promotes the healthy development of the stock market.First of all,this paper expounds the theories related to stock price predictability and introduces the theories related to deep learning.Secondly,with the CSI 300 component stocks as the research object,24 characteristic data are constructed based on the volume and price data of stocks,and the deep learning model is used to build the stock price prediction model.The stock price prediction model in this paper is a fusion of multiple deep learning models.First,the training data is put through a convolutional neural network containing attention mechanism,then the convolutional neural network is used to extract the features of the input data by its ability to extract the spatial dimension features,and then a short and long memory recursive neural network is used to further extract the features of the data in the time dimension.Connect another layer of Attention network with attention mechanism.Finally,the trained stock price prediction model is used to construct the investment strategy,and the strategy is backtested and the effect analysis.The results show that:(1)the CBAM-LSTM-Attention stock price prediction model constructed in this paper has a good effect on the test set,indicating that the model is effective in predicting the stock price;(2)The stock strategy based on the CBAM-LSTMAttention stock price prediction model can achieve better performance in the back test,indicating that the stock strategy based on the model is effective;(3)The training data used to construct the stock price prediction model in this paper is the historical stock trading data that can be collected in the market.The empirical results show that the stock price prediction model is effective in predicting the stock price,and the strategies based on the stock price prediction model have good performance,indicating that the stock market in our country is not weak and effective or not completely weak and effective.