计算机工程与应用 ›› 2021, Vol. 57 ›› Issue (24): 296-304.DOI: 10.3778/j.issn.1002-8331.2105-0178

• 工程与应用 • 上一篇    

基于HP-LSTM模型的股指价格预测方法

姚远,张朝阳   

  1. 河南大学 管理科学与工程研究所,河南 开封 475004
  • 出版日期:2021-12-15 发布日期:2021-12-13

Stock Index Price Forecasting Method Based on HP Filter

YAO Yuan, ZHANG Zhaoyang   

  1. Institute for Management Science and Engineering, Henan University, Kaifeng, Henan 475004, China
  • Online:2021-12-15 Published:2021-12-13

摘要:

股指价格时间序列受到长期和短期不同因素的影响,且具有非平稳、非线性等特点,传统计量模型的预测精度较低。为提高预测精度,一些研究将人工神经网络模型用于金融时间序列预测,取得了比传统计量模型更好的效果。提出了一种融合了HP滤波(Hodrick-Prescott Filter)和LSTM神经网络模型的股指价格预测模型,模型使用HP滤波将股指价格时间序列分解为长期趋势和短期波动,利用LSTM神经网络模型分别学习长期趋势和短期波动序列的特征,并分别进行长期趋势和短期波动预测,将预测结果融合得出股指价格预测结果。实验结果表明,提出的HP-LSTM混合模型不仅可以有效捕捉到股指价格时间序列的长期趋势和短期波动的变化规律,提高了股指价格预测精度,并且长期趋势和短期波动都具有相应的经济含义,提高了模型的可解释性。

关键词: 股指价格, 人工神经网络, 长短期人工神经网络(LSTM), HP滤波器, 预测

Abstract:

The time series of stock index price is affected by different long-term and short-term factors, and has the characteristics of non-stationary and non-linear. In order to improve the prediction accuracy, some studies use the artificial neural network model for financial time series prediction, and achieve better results than the traditional econometric model. In this paper, a stock index price forecasting model is proposed which combines HP filter and LSTM. The model first decomposes the stock index price time series into long-term trend and short-term fluctuation by HP filter, and then uses LSTM to learn the characteristics of long-term trend and short-term fluctuation series, And the long-term trend and short-term volatility are predicted respectively. Finally, the forecast results are fused to obtain the stock index price forecast results. The experimental results show that the proposed HP-LSTM hybrid model can not only effectively capture the long-term trend and short-term volatility of stock index price time series, but also improve the prediction accuracy of stock index price. Moreover, both long-term trend and short-term fluctuation have corresponding economic implications, which improves the interpretability of the model.

Key words: stock index price, artificial neural network, Long-Short-Term Memory(LSTM), HP filter, forecast