Analysis of Influence Mechanism of Company Stock Price Based on LASSO-CNN Neural Network

Abstract

To study the correlation between stock prices and financial indicators of Chinese listed companies, this paper selected the relevant data represented by the CSMAR database for research. This paper extracted 12 financial indicators that could reflect solvency, profitability, business capability, economic growth capability, and cash flow processing ability as the input layer and used the Least Absolute Shrinkage and Selection Operator (LASSO) algorithm to reduce the data dimension to five indicators. Finally, we established Convolutional Neural Networks (CNN), trained the neural network, and got the simulation results of the company's stock price. The results show that the model has good robustness, and the fitting accuracy of the stock price is 97%. Among the selected indicators, the turnover rate of accounts payable and the growth rate of net profit have the greatest impact on the fluctuation of stock price, and the importance after standardization is 33.56% and 35.26%, respectively, which provides some suggestions for the prediction and analysis of stock price.