A constrained portfolio trading system using particle swarm algorithm and recurrent reinforcement learning

Saud Almahdi, Steve Yang

Research output: Contribution to journalArticle

Abstract

This study extends a recurrent reinforcement portfolio allocation and rebalancing management system with complex portfolio constraints using particle swarm algorithms. In particular, we propose to use a combination of recurrent reinforcement learning (RRL) and particle swarm algorithm (PSO) with Calmar ratio for both asset allocation and constraint optimization. Using S&P100 index stocks, we show such a system with a Calmar ratio based objective function yields a better efficient frontier than the Sharpe ratio and mean-variance based portfolios. By comparing with multiple PSO based long only constrained portfolios, we propose an optimal portfolio trading system that is capable of generating both long and short signals and handling the common portfolio constraints. We further develop an adaptive RRL-PSO portfolio rebalancing decision system with a market condition stop-loss retraining mechanism, and we show that the proposed portfolio trading system outperforms the benchmarks consistently especially under high transaction cost conditions.

Original languageEnglish (US)
Pages (from-to)145-156
Number of pages12
JournalExpert Systems With Applications
Volume130
DOIs
StatePublished - Sep 15 2019

Fingerprint

Trading systems
Reinforcement learning
Portfolio rebalancing
Portfolio constraints
Retraining
Asset allocation
Benchmark
Optimal portfolio
Portfolio allocation
Objective function
Transaction costs
Stock index
Market conditions
Reinforcement
Management system
Mean-variance
Efficient frontier
Sharpe ratio

Cite this

@article{08ed62b1ecc74cc1b49a6ab1104ffa38,
title = "A constrained portfolio trading system using particle swarm algorithm and recurrent reinforcement learning",
abstract = "This study extends a recurrent reinforcement portfolio allocation and rebalancing management system with complex portfolio constraints using particle swarm algorithms. In particular, we propose to use a combination of recurrent reinforcement learning (RRL) and particle swarm algorithm (PSO) with Calmar ratio for both asset allocation and constraint optimization. Using S&P100 index stocks, we show such a system with a Calmar ratio based objective function yields a better efficient frontier than the Sharpe ratio and mean-variance based portfolios. By comparing with multiple PSO based long only constrained portfolios, we propose an optimal portfolio trading system that is capable of generating both long and short signals and handling the common portfolio constraints. We further develop an adaptive RRL-PSO portfolio rebalancing decision system with a market condition stop-loss retraining mechanism, and we show that the proposed portfolio trading system outperforms the benchmarks consistently especially under high transaction cost conditions.",
author = "Saud Almahdi and Steve Yang",
year = "2019",
month = "9",
day = "15",
doi = "https://doi.org/10.1016/j.eswa.2019.04.013",
language = "English (US)",
volume = "130",
pages = "145--156",
journal = "Expert Systems with Applications",
issn = "0957-4174",
publisher = "Elsevier Limited",

}

A constrained portfolio trading system using particle swarm algorithm and recurrent reinforcement learning. / Almahdi, Saud; Yang, Steve.

In: Expert Systems With Applications, Vol. 130, 15.09.2019, p. 145-156.

Research output: Contribution to journalArticle

TY - JOUR

T1 - A constrained portfolio trading system using particle swarm algorithm and recurrent reinforcement learning

AU - Almahdi, Saud

AU - Yang, Steve

PY - 2019/9/15

Y1 - 2019/9/15

N2 - This study extends a recurrent reinforcement portfolio allocation and rebalancing management system with complex portfolio constraints using particle swarm algorithms. In particular, we propose to use a combination of recurrent reinforcement learning (RRL) and particle swarm algorithm (PSO) with Calmar ratio for both asset allocation and constraint optimization. Using S&P100 index stocks, we show such a system with a Calmar ratio based objective function yields a better efficient frontier than the Sharpe ratio and mean-variance based portfolios. By comparing with multiple PSO based long only constrained portfolios, we propose an optimal portfolio trading system that is capable of generating both long and short signals and handling the common portfolio constraints. We further develop an adaptive RRL-PSO portfolio rebalancing decision system with a market condition stop-loss retraining mechanism, and we show that the proposed portfolio trading system outperforms the benchmarks consistently especially under high transaction cost conditions.

AB - This study extends a recurrent reinforcement portfolio allocation and rebalancing management system with complex portfolio constraints using particle swarm algorithms. In particular, we propose to use a combination of recurrent reinforcement learning (RRL) and particle swarm algorithm (PSO) with Calmar ratio for both asset allocation and constraint optimization. Using S&P100 index stocks, we show such a system with a Calmar ratio based objective function yields a better efficient frontier than the Sharpe ratio and mean-variance based portfolios. By comparing with multiple PSO based long only constrained portfolios, we propose an optimal portfolio trading system that is capable of generating both long and short signals and handling the common portfolio constraints. We further develop an adaptive RRL-PSO portfolio rebalancing decision system with a market condition stop-loss retraining mechanism, and we show that the proposed portfolio trading system outperforms the benchmarks consistently especially under high transaction cost conditions.

UR - http://www.scopus.com/inward/record.url?scp=85064433484&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85064433484&partnerID=8YFLogxK

U2 - https://doi.org/10.1016/j.eswa.2019.04.013

DO - https://doi.org/10.1016/j.eswa.2019.04.013

M3 - Article

VL - 130

SP - 145

EP - 156

JO - Expert Systems with Applications

JF - Expert Systems with Applications

SN - 0957-4174

ER -