Regime-Adaptive Continual Learning for Portfolio Management

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Pan, Chaofan, Ren, Lingfei, Xiong, Linbo, Li, Yonghao, Wei, Wei, Yang, Xin
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910273068072960
author Pan, Chaofan
Ren, Lingfei
Xiong, Linbo
Li, Yonghao
Wei, Wei
Yang, Xin
author_facet Pan, Chaofan
Ren, Lingfei
Xiong, Linbo
Li, Yonghao
Wei, Wei
Yang, Xin
contents Financial markets are inherently non-stationary, exhibiting frequent regime shifts and structural changes that render traditional Portfolio Management (PM) approaches ineffective. Existing remedies, such as rolling-window retraining and naive online fine-tuning, are hindered by high computational costs and insufficient knowledge utilization, respectively, resulting in low returns and limited adaptability. Continual learning (CL) offers a promising paradigm by enabling trading agents to accumulate and transfer knowledge across sequential tasks. In this paper, we propose \textbf{Re}gime-aware \textbf{C}ontinual \textbf{A}daptive \textbf{P}ortfolio management (\textbf{ReCAP}), a novel framework that integrates CL into PM to address the challenges of dynamic financial environments. ReCAP employs an adaptive regime detection module to segment historical market data into variable-length regimes, enabling regime-specific learning of policy vectors and the construction of a policy library. During continual trading, a regime-gate module adaptively combines policy vectors from the library based on the current market state, facilitating rapid adaptation to newly detected regimes. Only the regime-gate and the current regime's policy vector are continually updated to preserve useful knowledge effectively. Extensive experiments on five real-world datasets demonstrate that ReCAP consistently outperforms popular baselines, achieving superior returns in long-term investment horizons and rapid adaptation to regime shifts.
format Preprint
id arxiv_https___arxiv_org_abs_2606_00143
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Regime-Adaptive Continual Learning for Portfolio Management
Pan, Chaofan
Ren, Lingfei
Xiong, Linbo
Li, Yonghao
Wei, Wei
Yang, Xin
Portfolio Management
Artificial Intelligence
Financial markets are inherently non-stationary, exhibiting frequent regime shifts and structural changes that render traditional Portfolio Management (PM) approaches ineffective. Existing remedies, such as rolling-window retraining and naive online fine-tuning, are hindered by high computational costs and insufficient knowledge utilization, respectively, resulting in low returns and limited adaptability. Continual learning (CL) offers a promising paradigm by enabling trading agents to accumulate and transfer knowledge across sequential tasks. In this paper, we propose \textbf{Re}gime-aware \textbf{C}ontinual \textbf{A}daptive \textbf{P}ortfolio management (\textbf{ReCAP}), a novel framework that integrates CL into PM to address the challenges of dynamic financial environments. ReCAP employs an adaptive regime detection module to segment historical market data into variable-length regimes, enabling regime-specific learning of policy vectors and the construction of a policy library. During continual trading, a regime-gate module adaptively combines policy vectors from the library based on the current market state, facilitating rapid adaptation to newly detected regimes. Only the regime-gate and the current regime's policy vector are continually updated to preserve useful knowledge effectively. Extensive experiments on five real-world datasets demonstrate that ReCAP consistently outperforms popular baselines, achieving superior returns in long-term investment horizons and rapid adaptation to regime shifts.
title Regime-Adaptive Continual Learning for Portfolio Management
topic Portfolio Management
Artificial Intelligence
url https://arxiv.org/abs/2606.00143