Dynamic Multi-Objective Optimization of Decarbonized Asset Portfolios under Geopolitical Uncertainty Using Hybrid Metaheuristic Algorithms and Recurrent Deep Learning

Tavakoli, Elham; Nouri, Reza

doi:10.25843/JIFM.2025.8563.29755

Dynamic Multi-Objective Optimization of Decarbonized Asset Portfolios under Geopolitical Uncertainty Using Hybrid Metaheuristic Algorithms and Recurrent Deep Learning

Document Type : Original Article

Authors

Elham Tavakoli

Reza Nouri

Master of Financial Management, Islamic Azad University, Sanandaj Branch, Sanandaj, Iran

10.25843/JIFM.2025.8563.29755

Abstract

The transition to a low-carbon economy has introduced “decarbonized assets” as a new financial class. Managing portfolios of these assets is challenging due to geopolitical uncertainty, climate policy volatility, and the dynamic nature of energy and carbon markets. These complexities require adaptive and data-driven optimization approaches capable of handling non-stationary environments.This paper proposes a dynamic multi-objective optimization framework combining metaheuristic algorithms (NSGA-II, PSO, Differential Evolution) with deep recurrent neural networks (LSTM/GRU) for forecasting and decision-making. The model is designed to capture both temporal dependencies and stochastic market behavior. It simultaneously optimizes portfolio return, carbon intensity, and geopolitical risk exposure.
Results indicate that the proposed hybrid approach provides more stable and efficient portfolio solutions compared to the classical Markowitz framework, particularly in highly volatile and uncertain market conditions. The findings highlight the importance of integrating artificial intelligence with evolutionary optimization for sustainable financial decision-making.

Keywords

Multi-objective optimization

Decarbonized assets

Geopolitical uncertainty

Hybrid metaheuristic algorithms

Recurrent deep learning