计算机工程与应用 ›› 2023, Vol. 59 ›› Issue (3): 293-299.DOI: 10.3778/j.issn.1002-8331.2109-0388

• 工程与应用 • 上一篇    下一篇

基于多层级新闻的股价预测与交易策略研究

龙文,田嘉祺,毛元丰   

  1. 中国科学院大学 经济与管理学院,北京 100190
  • 出版日期:2023-02-01 发布日期:2023-02-01

Research on Stock Price Prediction and Trading Strategy Based on Multi-Level News

LONG Wen, TIAN Jiaqi, MAO Yuanfeng   

  1. School of Economics and Management, University of Chinese Academy of Sciences, Beijing 100190,China
  • Online:2023-02-01 Published:2023-02-01

摘要: 构建包括公司、子行业和行业三个层级的综合新闻体系,从新闻层级角度拓展了股价预测任务中所使用新闻的范围,研究多层级新闻体系对股价趋势的预测作用。为了更好地利用各层级新闻,引入了多核学习(MKL)模型。研究发现,三个层级的新闻都能在预测中发挥作用,相比只考虑个股新闻的SVM模型,基于多层级新闻的MKL模型预测准确率提升了10%。在此基础上构建交易策略,模拟交易的结果显示,引入多层级新闻的MKL模型能获得超额收益,表明其在市场交易中具有实践价值。

关键词: 多层级财经新闻, 多核学习模型, 股价趋势, 交易策略

Abstract: The paper builds a comprehensive news system including three levels: company, sub industry and industry to expand the scope of news used in stock price forecasting task from the perspective of news level and study the prediction effect of multi-level news system on stock price trend. In order to make better use of the news at all levels, multi-kernel learning (MKL) model is used. It is found that the three levels of news can play a role in prediction. Compared with the support vector machine (SVM) model considering only individual stock news, the prediction accuracy of MKL model based on multi-level news is improved by 10%. The trading strategy is constructed according to the above model. The simulation results show that the MKL model with multi-level news has important practical value in market trading.

Key words: multi-level financial news, multi-kernel learning model, stock price trend, trading strategy