Cost-of-capital assessments with factor models require quantitative forward-looking estimates. We recommend estimating Vasicek-shrunk betas with 1–4 years of daily stock returns and then shrinking betas a second time (and more for smaller stocks and longer-term projects), because the underlying betas are themselves time-varying. Such estimators also work well in other developed countries and for small-minus-big (SMB) and high-minus-low (HML) exposures. If own historical stock returns are not available, peer betas based on market cap should be used. Historical industry averages have almost no predictive power and should never be used.