Mean–variance portfolio optimization with deep learning based-forecasts for cointegrated stocks
作者：
Highlights：
• Novel method for investment using machine learning models.
• Portfolio formation by mean–variance and stationary analysis for predictive finance.
• Integrating pairs trading and A-LSTM model shows significant results.
• Considers holding period of the portfolio of several days, weeks, and one month.
• Proposed portfolio outperforms conventional mean–variance portfolio.
摘要
•Novel method for investment using machine learning models.•Portfolio formation by mean–variance and stationary analysis for predictive finance.•Integrating pairs trading and A-LSTM model shows significant results.•Considers holding period of the portfolio of several days, weeks, and one month.•Proposed portfolio outperforms conventional mean–variance portfolio.
论文关键词：Mean–variance portfolio optimization,Pairs trading,Mean-reverting spread prediction,Stationary portfolio,Attention-based LSTM network
论文评审过程：Received 2 September 2021, Revised 22 January 2022, Accepted 26 March 2022, Available online 12 April 2022, Version of Record 18 April 2022.
论文官网地址：https://doi.org/10.1016/j.eswa.2022.117005
