沪深300股票配对交易策略优化研究——基于LSTM动态参数调整和市值筛选方法被引量：1

Research on Optimization of Paired Trading Strategies for Shanghai and Shenzhen 300 Stocks——Based on LSTM dynamic parameter adjustment and market value filtering

原文传递

导出

摘要构建有活力、有韧性的资本市场,对于促进上市企业高质量发展和优化金融市场投融资功能具有重要意义。在传统配对交易策略中,相对固定的交易参数和缺乏灵活性的预筛选规则限制了策略对市场变化的动态响应。为了克服这些限制,本文提出了一种结合市值筛选规则与多股票配对交易框架的优化策略,确保选取的股票能够有效识别低估值优质标的。该策略采用长短期记忆网络(LSTM)算法动态调整交易参数,以提升预测精度。实证分析发现:与固定参数策略相比,本文所构建的优化策略不仅增强了对我国股票市场的价格发现能力,还实现了更高的投资回报。研究拓展了配对交易的理论范畴,为挖掘我国股票市场中的优质及低估值资产提供了模型与算法指导,有助于提振市场信心和改善长期投资预期。 Building a dynamic and resilient capital market is crucial for promoting the high-quality development of listed companies and optimizing the investment and financing functions of the financial market.Traditional pair trading strategies,with their relatively fixed trading parameters and a lack of flexible pre-screening rules,are limited in their dynamic response to market fluctuations.To overcome these limitations,this paper proposes an optimized strategy that integrates market value screening rules with a multi-stock pairs trading framework,ensuring that selected stocks can effectively identify undervalued,high-quality targets.The strategy employs Long Short-Term Memory(LSTM)networks to dynamically adjust trading parameters,enhancing prediction accuracy.Empirical analysis reveals that,compared to strategies with fixed parameters,the optimized strategy developed in this study not only enhances price discovery capabilities in the Chinese stock market but also achieves higher investment returns.This research expands the theoretical domain of pair trading and provides model and algorithmic guidance for uncovering high-quality and undervalued assets in the Chinese stock market,thereby helping to boost market confidence and improve long-term investment expectations.

作者刁海璨张延群 DIAO Haican;ZHANG Yanqun

机构地区中国社会科学院数量经济与技术经济研究所

出处《价格理论与实践》北大核心 2024年第3期149-155,共7页 Price:Theory & Practice

基金国家自然科学基金重大项目“宏观大数据建模和预测研究”(项目编号:71991475) “中国社会科学院经济大数据与政策评估实验室”(2024SYZH004) 中国社会科学院学科建设“登峰战略”资助计划数量经济学优势学科(DF2023YS29)

关键词多支股票配对交易协整检验交易参数市值筛选 LSTM multi-stock pairs trading cointegration test trading parameters market capitalization filtering long short-term memory(LSTM)

分类号 F832.51 [经济管理—金融学] F224 [经济管理—国民经济]

作者简介张延群(1968-),女,内蒙古包头人,德国柏林自由大学经济学博士,中国社会科学院数量经济与技术经济研究所研究员,博士生导师,中国社会科学院大学教授。研究方向:经济预测、大数据模型研究。