Research on Stock Price Prediction Based on Markov-LSTM Neural Network -Take the New Energy Industry as an Example

Abstract

With the increasingly close combination of the financial field and the real industry, as a new economic growth point, it is of great significance to study its price fluctuation prediction in the financial stock market. Based on this, we propose a neural network prediction method via Markov chain and LSTM neural network. Then we select three stocks as the research object, K-mean clustering algorithm is used to determine Markov chain’s relative states, K was obtained using the cross validation method. The empirical results show that the new method is better than LSTM neural network in the evaluation indexes of average relative error, a posteriori error, mean square error and small robability error of stock price prediction. This shows that under certain conditions, Markov chain-LSTM neural network can make more accurate prediction of stock price, and is more suitable for systems with high volatility and strong randomness. Finally, we discuss some promotion and improvement methods of Markov chain-LSTM neural network, and put forward relevant suggestions for the high-quality development of new energy industry from the perspective of finance.