LSTM artificial neural network prediction of stock prices in China

• 발행기관: 한국동북아경제학회
• 저자명: Boye, LI Sangwook, KIM
• 간행물 정보: 『동북아경제연구』東北亞經濟硏究 第32卷 第2號, 61~84쪽, 전체 24쪽
• 주제분류: 경제경영 > 경제학
• 파일형태: PDF
• 발행일자: 2020.08.30

영문 초록

This paper combines factor analysis of traditional statistics and LSTM artificial neural network in deep learning to make stock price prediction, and finally carries out back measurement. Firstly, financial indexes of companies in CSI 300 stocks are selected for factor analysis. In this paper, a total of 9 financial indicators are selected and 4 factors(solvency factor, growth competency factor, market value factor, profitability factor) and total score are extracted and total score. After that, each factor is ranked and the data of the top 5 stocks with the highest score of each factor is passed into LSTM artificial neural network as stock selection criteria for fitting and testing. Finally, the five stocks with the highest degree of fit were backtested. In the backtest, the LSTM was used as the core to predict the situation of the current stock on the second day. According to the situation of the second day, the price of the first day was compared, and the strategy of buying low and selling high was implemented. Through the above strategy in the backtesting period, the highest one yield of 20.9%, the rest of the backtesting stocks are more than Shanghai and shenzhen 300 benchmark. indicating the usefulness of LSTM artificial neural network in stock prediction.

목차

Ⅰ. Introduction

Ⅱ. Literature review

Ⅲ. Factor analysis and LSTM

Ⅳ. Empirical analysis

Ⅴ. Conclusion

Reference

키워드

China Stock price CSI 300 Factor analysis LSTM Back-testing JEL 분류번호: G1, N2