Proceedings of the 4th International Conference on Economic Management and Model Engineering, ICEMME 2022, November 18-20, 2022, Nanjing, China

Research Article

Stock Price Prediction Based on Decision Trees, CNN and LSTM

Download1036 downloads
  • @INPROCEEDINGS{10.4108/eai.18-11-2022.2327160,
        author={Ruotong  Li and Moying  Ma and Nan  Tang},
        title={Stock Price Prediction Based on Decision Trees, CNN and LSTM},
        proceedings={Proceedings of the 4th International Conference on Economic Management and Model Engineering, ICEMME 2022, November 18-20, 2022, Nanjing, China},
        publisher={EAI},
        proceedings_a={ICEMME},
        year={2023},
        month={2},
        keywords={stock price; forecasting; decision trees; cnn; lstm},
        doi={10.4108/eai.18-11-2022.2327160}
    }
    
  • Ruotong Li
    Moying Ma
    Nan Tang
    Year: 2023
    Stock Price Prediction Based on Decision Trees, CNN and LSTM
    ICEMME
    EAI
    DOI: 10.4108/eai.18-11-2022.2327160
Ruotong Li1, Moying Ma2,*, Nan Tang3
  • 1: Department of electronic commerce, South China University of Technology
  • 2: Department of Economy, Xiamen University
  • 3: School of Marine Sciences, Sun Yat-sen University
*Contact email: 15220182202589@stu.xmu.edu.cn

Abstract

Stock price forecasting plays an important role in quantitative transaction and financial analysis. In this paper, three well-known machine learning approaches, decision tree, LSTM, and CNN, are implemented in order to realize stock price prediction. After introducing each model's background, principle, and method, the historical closing price of China Merchants Bank stock is used as the training set data. After the establishment of the three models, the comparison between the predicted value and the real value on the test set is demonstrated. According to the results, the prediction feasibility is verified to a certain extent. Furthermore, it is concluded that the forecast effect in the short term is better than that in the long term. These results shed light on the effectiveness limitations of machine learning used in quantitative trading.