Trend Prediction of Stock Index Based on Convolutional Neural Network

ABSTRACT

The emergence of the COVID-19 has a huge impact on the Chinese and American economies, including the fluctuations of stock price in the financial market. It's significantly valuable to search out the rules of index variability under this post-epidemic era. In this paper, we create an improved Convolutional Neural Network to search out the future trend of Shanghai Composite Index and Nasdaq composite index by using the daily data from January 1, 2011 to April 23, 2021, and find out the characteristics through nonlinear test and random lasso algorithm. The empirical results show that the prediction correction determination coefficients of Shanghai Composite Index and Nasdaq composite index reach 0.87 and 0.97 respectively, which shows that it is feasible and effective to use convolutional neural network to predict the stock index. © 2022 IEEE.