Incremental Reinforcement Learning for Portfolio Optimisation

Portfolio optimisation is a crucial decision-making task. Traditionally static, this problem is more realistically addressed as dynamic, reflecting frequent trading within financial markets. The dynamic nature of the portfolio optimisation problem makes it susceptible to rapid market changes or fina...

Full description

Saved in:
Bibliographic Details
Main Authors: Refiloe Shabe, Andries Engelbrecht, Kian Anderson
Format: Article
Language:English
Published: MDPI AG 2025-06-01
Series:Computers
Subjects:
Online Access:https://www.mdpi.com/2073-431X/14/7/242
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849733601358249984
author Refiloe Shabe
Andries Engelbrecht
Kian Anderson
author_facet Refiloe Shabe
Andries Engelbrecht
Kian Anderson
author_sort Refiloe Shabe
collection DOAJ
description Portfolio optimisation is a crucial decision-making task. Traditionally static, this problem is more realistically addressed as dynamic, reflecting frequent trading within financial markets. The dynamic nature of the portfolio optimisation problem makes it susceptible to rapid market changes or financial contagions, which may cause drifts in historical data. While reinforcement learning (RL) offers a framework that allows for the formulation of portfolio optimisation as a dynamic problem, existing RL approaches lack the ability to adapt to rapid market changes, such as pandemics, and fail to capture the resulting concept drift. This study introduces a recurrent proximal policy optimisation (PPO) algorithm, leveraging recurrent neural networks (RNNs), specifically the long short-term memory network (LSTM) for pattern recognition. Initial results conclusively demonstrate the recurrent PPO’s efficacy in generating quality portfolios. However, its performance declined during the COVID-19 pandemic, highlighting susceptibility to rapid market changes. To address this, an incremental recurrent PPO is developed, leveraging incremental learning to adapt to concept drift triggered by the pandemic. This enhanced algorithm not only learns from ongoing market data but also consistently identifies optimal portfolios despite significant market volatility, offering a robust tool for real-time financial decision-making.
format Article
id doaj-art-9d76da15712d49c6b9d513d435cde377
institution DOAJ
issn 2073-431X
language English
publishDate 2025-06-01
publisher MDPI AG
record_format Article
series Computers
spelling doaj-art-9d76da15712d49c6b9d513d435cde3772025-08-20T03:08:00ZengMDPI AGComputers2073-431X2025-06-0114724210.3390/computers14070242Incremental Reinforcement Learning for Portfolio OptimisationRefiloe Shabe0Andries Engelbrecht1Kian Anderson2Department of Industrial Engineering, Stellenbosch University, Stellenbosch 7600, South AfricaDepartment of Industrial Engineering, Stellenbosch University, Stellenbosch 7600, South AfricaComputer Science Division, Stellenbosch University, Stellenbosch 7600, South AfricaPortfolio optimisation is a crucial decision-making task. Traditionally static, this problem is more realistically addressed as dynamic, reflecting frequent trading within financial markets. The dynamic nature of the portfolio optimisation problem makes it susceptible to rapid market changes or financial contagions, which may cause drifts in historical data. While reinforcement learning (RL) offers a framework that allows for the formulation of portfolio optimisation as a dynamic problem, existing RL approaches lack the ability to adapt to rapid market changes, such as pandemics, and fail to capture the resulting concept drift. This study introduces a recurrent proximal policy optimisation (PPO) algorithm, leveraging recurrent neural networks (RNNs), specifically the long short-term memory network (LSTM) for pattern recognition. Initial results conclusively demonstrate the recurrent PPO’s efficacy in generating quality portfolios. However, its performance declined during the COVID-19 pandemic, highlighting susceptibility to rapid market changes. To address this, an incremental recurrent PPO is developed, leveraging incremental learning to adapt to concept drift triggered by the pandemic. This enhanced algorithm not only learns from ongoing market data but also consistently identifies optimal portfolios despite significant market volatility, offering a robust tool for real-time financial decision-making.https://www.mdpi.com/2073-431X/14/7/242portfolio optimisationreinforcement learningincremental learningrecurrent neural networkslong short-term memory networks
spellingShingle Refiloe Shabe
Andries Engelbrecht
Kian Anderson
Incremental Reinforcement Learning for Portfolio Optimisation
Computers
portfolio optimisation
reinforcement learning
incremental learning
recurrent neural networks
long short-term memory networks
title Incremental Reinforcement Learning for Portfolio Optimisation
title_full Incremental Reinforcement Learning for Portfolio Optimisation
title_fullStr Incremental Reinforcement Learning for Portfolio Optimisation
title_full_unstemmed Incremental Reinforcement Learning for Portfolio Optimisation
title_short Incremental Reinforcement Learning for Portfolio Optimisation
title_sort incremental reinforcement learning for portfolio optimisation
topic portfolio optimisation
reinforcement learning
incremental learning
recurrent neural networks
long short-term memory networks
url https://www.mdpi.com/2073-431X/14/7/242
work_keys_str_mv AT refiloeshabe incrementalreinforcementlearningforportfoliooptimisation
AT andriesengelbrecht incrementalreinforcementlearningforportfoliooptimisation
AT kiananderson incrementalreinforcementlearningforportfoliooptimisation