Trend representation based log-density regularization system for portfolio optimization

作者：

Highlights：

• A novel trend representation is proposed for portfolio optimization.

• The system directly exploits time variable to represent price trend patterns.

• The system is robust to a small error of the price prediction.

• Experimental results show that the proposed system is effective and efficient.

摘要

•A novel trend representation is proposed for portfolio optimization.•The system directly exploits time variable to represent price trend patterns.•The system is robust to a small error of the price prediction.•Experimental results show that the proposed system is effective and efficient.

论文关键词：Trend representation,Log-density regularization,Ridge regression,Portfolio optimization

论文评审过程：Received 4 May 2017, Revised 13 October 2017, Accepted 18 October 2017, Available online 20 October 2017, Version of Record 21 December 2017.

论文官网地址：https://doi.org/10.1016/j.patcog.2017.10.024