A0273
Title: The mosaic permutation test: An exact and nonparametric goodness-of-fit test for factor models
Authors: Asher Spector - Stanford University (United States) [presenting]
Emmanuel Candes - Stanford (United States)
Rina Foygel Barber - University of Chicago (United States)
Trevor Hastie - Stanford University (United States)
Ronald Kahn - BlackRock (United States)
Abstract: Financial firms often rely on fundamental factor models to explain correlations among asset returns and manage risk. Yet after major events, e.g., COVID-19, analysts may reassess whether existing risk models continue to fit well: specifically, after accounting for the factor exposures, are the residuals of the asset returns independent? With this motivation, the mosaic permutation test is introduced, a nonparametric goodness-of-fit test for preexisting factor models. The method can leverage nearly any machine learning technique to detect model violations while provably controlling the false positive rate, i.e., the probability of rejecting a well-fitting model, without making asymptotic approximations or parametric assumptions. This property helps prevent analysts from unnecessarily rebuilding accurate models, which can waste resources and increase risk. To illustrate the methodology, the mosaic permutation test is applied to the BlackRock fundamental equity risk (BFRE) model. Although the BFRE model generally explains the most significant correlations among assets, evidence of unexplained correlations is found among certain real estate stocks, and it is shown that adding new factors improves model fit. The methods in the Python package mosaicperm are implemented.
