Machine Learning-Driven Credit Portfolio Optimization: Balancing Risk, Return, And Default Correlation in Volatile Markets

  ABSTRACT : Financial institutions face significant challenges in managing credit portfolios due to rising market volatility, sector connections, and changing borrower behaviors. Traditional credit risk models often struggle to capture dynamic non-linear relationships, rapidly shifting risk factors, and systemic shocks. This study looks at how machine learning (ML) techniques can improve credit portfolio optimization by balancing risk, return, and default correlation in volatile market conditions. The research combines ML models with financial theory, regulatory standards, and ESG (Environmental, Social, and Governance) factors. By using predictive algorithms, dependency modeling, and reinforcement learning, the study creates a framework for predicting credit risk, assessing correlation, and rebalancing portfolios dynamically. The findings show that MLdriven strategies enhance risk-adjusted returns, improve resilience during stressful situations, and increase transparency and fairness in decision-making. This study adds to academic literature by connecting financial theory and AI applications and offering practical advice for risk managers, regulators, and financial institutions