EURO 2025 Leeds
Abstract Submission

3063. Efficient behaviorally biased portfolio decisions preconditioned by asset interconnectedness

Invited abstract in session WC-61: Behavioural studies in cognate domains 2, stream Behavioural OR.

Wednesday, 12:30-14:00
Room: Maurice Keyworth G.31

Authors (first author is the speaker)

1. Gordon Dash
Finance and Decision Sciences, University of Rhode Island
2. Nina Kajiji
Computer Science and Statistics, University of Rhode Island, and The NKD Group, Inc.

Abstract

Investors continue to seek efficient asset allocation models that incorporate environmental, social, and governance (ESG) considerations while accounting for investor behavioral biases. This study integrates a machine learning-based asset return forecasting framework and layered behavioral portfolio optimization goals to enumerate efficient multi-asset portfolios. Our approach explicitly accounts for investor biases presented in the literature. Building on prior research on asset return generation, we propose an optimized multi-objective neural network framework to predict asset risk premia while extracting conditional pricing information through cross-asset information sharing and nonlinear dynamics. The ML output sets the foundation for a hierarchical multiobjective portfolio optimization model to enumerate efficiently diversified asset allocations with behavioral objectives. Simulation results reinforce and extend existing research on portfolio sustainability, network theory, and financial return interconnectivity. Overall, our findings confirm the presence of a hump-shaped ESG efficiency frontier and produce enhanced Sharpe ratios while minimizing global minimum variance portfolio drawdowns. The ML framework combined with multiobjective behavioral optimization provides new insights into managing financial market complexity amid cognitive dissonance, particularly when investors face balancing ‘green’ and ‘brown’ asset diversification strategies.

Keywords

Status: accepted

Back to the list of papers