Equilibrium investment under dynamic preference uncertainty

Battle Count: 66

The paper is highly relevant to quantitative portfolio management as it addresses how evolving risk preferences affect optimal investment strategies. The preference-hedging component provides actionable insights for dynamic asset allocation when investor risk tolerance changes over time. However, the theoretical nature, single-asset assumption, and reliance on exogenous preference dynamics limit direct implementation in production trading systems. The framework is more applicable to strategic asset allocation and wealth management than high-frequency trading.

The paper involves advanced stochastic calculus (Ito diffusions, change of measure, conditional dynamics), equilibrium control theory for time-inconsistent problems, extended HJB systems, coupled nonlinear PIDEs indexed by a continuum of terminal states, and physics-informed neural network training. The numerical solution requires solving a forward-backward coupled system where the equilibrium policy depends on the PIDE solution and vice versa. The dimensionality and continuum of conditioning arguments present significant computational challenges. Implementation requires expertise in both mathematical finance and deep learning.

The paper provides detailed mathematical derivations, explicit parameter settings for numerical experiments (r=0.02, mu_S=0.07, sigma_S=0.2, sigma_Y=0.04, rho=0.6, exp(y0)=2), pseudocode for the neural network training algorithm (Algorithm 1 and 2), and comprehensive tables of equilibrium policies. However, no code repository is provided, and the neural network architecture details (layers, activation functions, hyperparameters beyond learning rate) are not fully specified. The theoretical framework is fully self-contained with proofs in appendices.