摘要
随着消费升级与投资需求的不断增长,投资组合管理受到越来越多的关注。金融大数据的发展对该领域的研究提出了更高的要求。本研究基于强化学习提出了一种基于动态交易的智能投资组合优化方法,该方法考虑了风险因素对投资组合管理过程的影响,能依据市场状态和资产信息自动转换投资组合优化模式以应对市场风格变化,通过投资组合内部资产与外部资产池动态交易的形式来实时调整投资组合资产构成及资产配置。通过中国股票市场数据的实证分析,本研究验证了该投资组合优化方法的可行性和有效性。研究发现,依据市场变化和动态交易方式来选择投资组合的资产构成并考虑风险约束是非常必要的,而且引入更多信息对投资组合的优化有积极作用。此外,在投资组合的优化中,考虑下行风险约束比考虑总体风险更有利于实现既定投资风险下的收益最大化。
With the continuous growth of consumption and investment demand,much attention is paid to portfolio management.The development of financial big data puts forward higher requirements for the research in this field.This paper proposes a portfolio optimisation method based on dynamic trading using reinforcement learning.This method considers the impact of risk on the process of portfolio management,and is able to transform the portfolio optimisation model automatically according to the market state and asset information,so as to cope with the change of different market.Moreover,through the dynamic trading between the portfolio assets and external market assets,these models can adjust the composition of portfolio assets and asset allocation in real time.Through the empirical analysis of Chinese stock market data,the feasibility and effectiveness of the intelligent portfolio optimisation method are testified.It is found that it is necessary to change the asset composition of the portfolio and consider the risk constraints according to the market changes and dynamic trading patterns.And introducing more information has a positive effect on the portfolio optimisation.Additionally,for portfolio optimisation,considering downside risk constraints is more beneficial to maximize returns under given investment risks than considering the overall risks.
作者
王舞宇
章宁
范丹
王熙
WANG Wu-yu;ZHANG Ning;FAN Dan;WANG Xi
出处
《中央财经大学学报》
CSSCI
北大核心
2021年第9期32-47,共16页
Journal of Central University of Finance & Economics
基金
国家重点研发计划专项课题“智能服务交易系统理论基础”(项目编号:2017YFB1400701)
中央财经大学博士研究生重点选题支持计划资助项目“智能投资组合管理与交易策略研究”
中央财经大学新兴交叉学科建设项目“金融系统安全与区块链监管科技”。
关键词
投资组合优化
动态交易
风险约束
强化学习
Portfolio optimisation
Dynamic trading
Risk constraints
Reinforcement learning
