Hundreds of factors have been proposed to explain asset returns during the past two decades. In this paper, we develop a Bayesian approach to explore the space of possible linear factor model in this 'factor zoo'. We conduct an extensive rearch for promising models using 83 candidate factors and individual-stock return data. Our results show that (i) only a handful of factors matter to explain individual stocks; (ii) the only factor that is consistently selected over time is the market factor; and (iii) other factors which are occasionally selected are not those in widely used multi-factor models. 

Keywords: Factor selection, Bayesian variable selection, Seemingly Unrelated Regressions​