. A Bayesian analysis would start with a prior that the mean is m ± d and then mix in the sample evidence.2As we have noted earlier, our estimates of the means of the returns mr generally contain a great deal of sample error. The sample errors affect the parameter c0 = mr − c1 · mg. If we have a strong prior reason to believe that the unconditional expected return is equal to m, then we can replace the estimate of the coefficient c0 by . A Bayesian analysis would start with a prior that the mean is m ± d and then mix in the sample evidence.