Abstract

The fluctuation of stock market price is the basis of stock market operation. The big fluctuation of stock market price will cause huge losses to investors. Stock price fluctuation is a highly complex nonlinear system, the price adjustment is not in accordance with the uniform time process, has its own process, has high risk, is very difficult to predict. The stock prediction topic is one of the most explored area of research. However, due to the characteristics of large quantity, non-linearity and long memory of stock price data, the traditional machine learning methods have some problems, such as few model parameters, impractical, time-consuming prediction and so on. Therefore, how to study effective deep learning methods to improve the recognition efficiency and prediction accuracy of forecasting models is an urgent problem to be solved in the field of stock forecasting. Based on the theory of deep learning, this paper constructs a stock price prediction index system according to the characteristics of China’s top enterprises. In this paper, the stock price trend of Build Your Dreams (BYD), the top new energy vehicle company in China, is taken as the main data set, and the Long and Short-term Memory (LSTM) recurrent neural network is used to predict BYD’s stock price. This paper solves the long sequence dependence problem in general recursive neural networks. Using the characteristics of recurrent neural network and stock market, the data are preprocessed by interpolation and wavelet denoising, and then pushed to different LSTM layers. Training and testing were carried out in LSTM network models with different numbers of hidden neurons at the same level. The evaluation indexes and prediction effects were compared to find the appropriate LSTM layers and hidden neurons to improve the prediction accuracy

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call