📝 SummarXiv

Abstract

Existing approaches to asset-pricing under model-uncertainty adapt classical utility-maximization frameworks and seek theoretical comprehensiveness. We move toward practice by considering binary model-risks and by emphasizing 'constraints' over 'preference'. This decomposes viable economic asset-pricing into that of model and non-model risks separately, leading to a unique and convenient model-risk pricing formula. Its parameter, a dynamically conserved constant of model-risk inference, allows an integrated representation of ex-ante risk-pricing and bias such that their ex-post impacts are disentangled via well-known anomalies, Momentum and Low-Risk, whose risk-reward patterns acquire a fresh significance: peak-reward reveals ex-ante risk-premia, and peak-location, bias.