摘要
投资组合策略问题是金融领域经久不衰的一个课题,将人工智能技术用于金融市场是信息技术时代一个重要的研究方向。目前的研究较多集中在股票的价格预测上,对于投资组合及自动化交易这类决策性问题的研究较少。文中基于深度强化学习算法,利用深度学习的BiLSTM来预测股价的涨跌,以强化学习的智能体进行观测,更好地判断当期情况,从而确定自己的交易动作;同时,利用传统的投资组合策略来建立交易的预权重,使智能体可以在自动化交易的过程中进行对比,从而不断优化自己的策略选择,生成当期时间点内最优的投资组合策略。文章选取美股的10支股票进行实验,在真实的市场模拟下表明,基于深度强化学习算法的模型累计收益率达到了86.5%,与其他基准策略相比,收益最高,风险最小,具有一定的实用价值。
The problem of investment portfolio strategy is an enduring topic in the financial field,and the application of artificial intelligence techniques in financial markets is an important research direction in the information technology era.Current research is more focused on price prediction of stocks,and less on decision-making problems such as investment portfolio and automated trading.Based on the deep reinforcement learning algorithm,the BiLSTM of deep learning is used to predict the rise and fall of stock prices,and the reinforcement learning agents is used to observe and better assess the current situation,so as to determine one′s own trading actions.Intelligent agents can comparison during automated trading processes by using traditional investment portfolio strategy to establish pre weights for transactions,so as to continuously optimize their strategy choices and generate the optimal investment portfolio strategy at the current time point.10 stocks from the US stock market are selected for experiments.Under real market simulations,the results show that the cumulative return of the model based on deep reinforcement learning algorithm can reach 86.5%.In comparison with other benchmark strategies,it has the highest return and the lowest risk,and has a certain practical value.
作者
杨旭
刘家鹏
越瀚
张芹
YANG Xu;LIU Jiapeng;YUE Han;ZHANG Qin(College of Economics and Management,China Jiliang University,Hangzhou 310018,China;College of Business,Zhejiang Wanli University,Ningbo 315100,China)
出处
《现代电子技术》
北大核心
2024年第6期154-160,共7页
Modern Electronics Technique
基金
国家社会科学基金项目(18BGL224)。
