考虑投资者情绪的均值回归在线投资组合策略  

Mean reversion online portfolio strategy with investor sentiment

作　　者：钟衍楠 徐维军[1,2] 于孝建 张卫国[1] Zhong Yannan;Xu Weijun;Yu Xiaojian;Zhang Weiguo(School of Business Administration,South China University of Technology,Guangzhou 510641,China;Greater Bay Intelligent Finance and Risk Management Research Base,Guangzhou 510641,China;School of Economics and Finance,South China University of Technology,Guangzhou 510641,China;不详)

机构地区：[1]华南理工大学工商管理学院,广东广州510641 [2]大湾区数智金融与风险管理研究基地,广东广州510641 [3]华南理工大学经济与金融学院,广东广州510641

出　　处：《系统工程学报》2025年第1期45-60,共16页Journal of Systems Engineering

基　　金：国家自然科学基金资助项目(72271095,71771091);广东省基础与应用基础研究资助项目(2022A1515010224).

摘　　要：利用文本中包含的投资者情绪挖掘股票的均值回归特征,从相对价格层面提出了基于回归特征修正的价格预测方法.将策略微调后的均值波动收益作为优化目标,利用窗口期内相对价格的波动导致的潜在收益控制模型的优化强度,提出了基于回归特征修正和微调惩罚的在线投资组合策略.选取上证100、沪深300、富时A50和恒生指数等国内指数市场数据对策略进行验证,结果表明:本文所提相对价格预测方法能提升现有均值回归策略的收益,所提策略的收益高于现有均值回归策略.采用短期内的投资者情绪作为市场均值回归特征,能更好地捕捉股票变化,提升改进策略的整体绩效表现.Based on the investor sentiment contained in the text,this paper analyzes the stock’s mean reversion feature and proposes a relative price prediction method.An online portfolio strategy is proposed that is based on regression feature correction and fine-tuning penalties,with the mean volatility return generated by the strategy as the optimization objective.The optimization strength is controlled using the potential return caused by the relative price fluctuations during the window period.The empirical results based on the SSE100,CSI300,CHA50,and HSI index market data show that the relative price prediction method proposed in this paper can improve the returns of existing mean reversion strategies.Further,the proposed strategy is better than the existing mean reversion strategies in return.Using short-term investor sentiment as the stock’s mean reversion feature can better capture stock changes and improve the overall performance of the improvement strategy.

关 键 词：在线投资组合 均值回归 投资者情绪:文本挖掘 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]