Beyond Beta Pricing: SDF Selection from Euler-Restricted Traded–Nontraded Factor–Return Models
Date:
Major Revision, Management Science
We develop Beyond Beta Pricing (BBP), a Bayesian generative learning framework for discovering stochastic discount factors (SDFs) in theory-constrained financial environments. BBP reframes SDF selection as probabilistic model discovery over a combinatorial set of factor subsets. Using Bayesian marginal likelihoods as evidence scores, the BBP learner performs automated model discovery to identify the most probable and parsimonious SDF consistent with the data and the no-arbitrage principle. Applied to U.S. equity returns with empirical-Bayes type priors, posterior mass concentrates on SDFs spanned by traded factors. Out-of-sample validation confirms that the BBP-learned SDF delivers superior risk-adjusted performance.
Paper: SSRN link