Font Size: a A A

Empirical Research On Quantitative Strategy Based On LSTM Model

Posted on:2021-05-06Degree:MasterType:Thesis
Country:ChinaCandidate:J W JiFull Text:PDF
GTID:2480306107463714Subject:Master of Finance
Abstract/Summary:PDF Full Text Request
The TMT industry is an information industry which is rapidly rising under the background of high and new technology.Its high growth has always been the focus of the global investment market.However,in this new industry,it is difficult for traditional investment models to effectively evaluate it.In a high-growth emerging industry,it is worth exploring to build an effective investment portfolio to obtain excess returns.At the same time,with the rapid development of artificial intelligence,deep learning,which is good at processing complex non-linear data,has been successful in various fields.This article attempts to apply the advantages of deep learning to the unpredictable stock data of the TMT industry,and construct a quantitative stock selection strategy that can obtain excess returns.The basic idea of the strategy is to predict the stock price through the multi-factor stock index,and construct a portfolio that meets the expectations,so as to obtain good investment returns.The strategy focuses on studying how to improve the predictive ability of multi-factor models for stock price information.Deep learning deep neural networks(DNN)and long-term short-term memory neural networks(LSTM)are introduced as prediction models.The accuracy of the model training results determines the accuracy Whether a quantitative strategy can achieve reliable excess returns.In this paper,through repeated testing,we find models and parameters suitable for A-share TMT industry stock data,and find that the LSTM model with Re LU activation function,eight nodes,and single layer performs best in predicting the underlying data.And it has been confirmed that the LSTM model with time series processing ability is significantly better than the traditional DNN model in processing stock data.Finally,in the backtesting of the quantitative model,the return on investment strategy,Sharpe rate,maximum drawdown,etc.were analyzed,and it was found that the LSTM quantitative investment strategy based on the prediction of stock price fluctuations performed well and outperformed all TMT industries.The quantitative strategy of the stock pool structure provides the LSTM model with empirical evidence in the A-share TMT industry.
Keywords/Search Tags:quantitative investment, long-and-short-term memory neural networks, deep neural networks, TMT industry
PDF Full Text Request
Related items