Advances in Dynamic Risk Assessment and Portfolio Optimization: Integrating Quantitative Models with Behavioral Insights

Abstract

Dynamic risk assessment and portfolio optimisation have undergone profound transformations in recent years due to the rise of data-driven models, adaptive allocation strategies, and insights from behavioural finance. Traditional approaches, which largely assume rational investor behaviour, static risk preferences, and normally distributed returns, are increasingly insufficient
in capturing the complexity and heterogeneity of real-world financial markets. Modern frameworks leverage machine learning, real-time data analytics, and scenario-based simulations to model time-varying risk, portfolio sensitivity, and non-linear asset interactions. Simultaneously, behavioural finance research has provided tools to measure investor biases, such as overconfidence, loss aversion, and herding, and incorporate these psychological factors into portfolio decisions. Hybrid models that integrate statistical rigour with behavioural realism have emerged, enabling more adaptive and personalised investment strategies. Key contributions include the development of dynamic allocation algorithms, risk forecasting models that account for regime shifts, and explainable AI approaches that improve transparency and trust. Despite these advances, challenges remain, including model interpretability, robustness under extreme market conditions, and ethical considerations in data usage. Future research is likely to focus on creating portfolio systems that are not only adaptive and predictive but also behaviourally informed, interpretable, and capable of providing actionable insights for diverse investor profiles in increasingly complex financial environments.