A new prediction NN framework design for individual stock based on the industry environment

Abstract

There is a research gap in accurately predicting an individual stock’s finances from industry environment factors. Therefore, to predict trading strategies for a target stock’s closing price, this study constructed a prediction module and an environment module for a hybrid variational mode decomposition and stacked gated recurrent unit (VMD-StackedGRU) model, with individual stock information input into the prediction module and industry information input into the environment module. The results from the U.S. banking industry generalization tests proved that the proposed model could significantly improve prediction performances and that the environment module did not play an important role and was not equal to the prediction module. The hybrid neural network framework was a new application for financial price predictions based on an industry environment. Profitable trading strategies and accurate predictions can be valuable in hedging against market volatility risk and in assuring significant returns for investors and investment institutions.