Proceedings of the 5th Management Science Informatization and Economic Innovation Development Conference, MSIEID 2023, December 8–10, 2023, Guangzhou, China

Research Article

Stock Prediction Based on LSTM

Download52 downloads
  • @INPROCEEDINGS{10.4108/eai.8-12-2023.2344733,
        author={Min  Li and Weize  Liao and Jianrong  Huang and Jiebo  Jiang},
        title={Stock Prediction Based on LSTM},
        proceedings={Proceedings of the 5th Management Science Informatization and Economic Innovation Development Conference, MSIEID 2023, December 8--10, 2023, Guangzhou, China},
        publisher={EAI},
        proceedings_a={MSIEID},
        year={2024},
        month={4},
        keywords={stock; lstm; prediction; multivariate},
        doi={10.4108/eai.8-12-2023.2344733}
    }
    
  • Min Li
    Weize Liao
    Jianrong Huang
    Jiebo Jiang
    Year: 2024
    Stock Prediction Based on LSTM
    MSIEID
    EAI
    DOI: 10.4108/eai.8-12-2023.2344733
Min Li1, Weize Liao2, Jianrong Huang2, Jiebo Jiang3,*
  • 1: Guangxi Xijiang Development Investment Group Co., Ltd.
  • 2: Wuzhou University
  • 3: Cangwu County Information Center
*Contact email: 359128664@qq.com

Abstract

Stock market forecasting based on Long Short-Term Memory networks (LSTMs) is a prevalent technique within the realms of Natural Language Processing and time-series analysis. Predicting stock market trends is widely acknowledged to be a highly intricate task, attributed to the multifaceted influences ranging from economic factors, policy changes, to market sentiment. The LSTM's superior capability in handling time-series data has thus made it a popular choice for such forecasting endeavors. This paper selects two stocks, Vanadium Titanium Shares (SZ.000629) and Crystal Technology (SH.603005), and applies both single variable LSTM(SV-LSTM) and multivariate LSTM(MV-LSTM) models to forecast the opening price, the lowest price, and the previous closing price. The results indicate that the MV-LSTM yields more accurate predictions and exhibits relatively smaller errors than its SV-LSTM counterpart.