Exponential Gradient with Momentum for Online Portfolio Selection

作者：

Highlights：

• We propose a new framework, which runs efficiently in terms of both time and space, for online portfolio selection in high-frequency trading.

• Theoretical analysis reveals that our proposed framework can achieve a sublinear regret.

• Experiments show that our proposed framework has good transaction cost scalability.

• Based on the empirical study, the proposed algorithms outperform the relevant algorithms.

摘要

•We propose a new framework, which runs efficiently in terms of both time and space, for online portfolio selection in high-frequency trading.•Theoretical analysis reveals that our proposed framework can achieve a sublinear regret.•Experiments show that our proposed framework has good transaction cost scalability.•Based on the empirical study, the proposed algorithms outperform the relevant algorithms.

论文关键词：Online portfolio selection,High-frequency trading,Algorithmic trading,Online learning

论文评审过程：Received 21 May 2020, Revised 4 September 2021, Accepted 5 September 2021, Available online 14 September 2021, Version of Record 20 September 2021.

论文官网地址：https://doi.org/10.1016/j.eswa.2021.115889