The good, the not so bad, and the ugly.

We propose a new methodology to evaluate the skill of fund managers. We assume that a manager’s skill, characterized by alpha, is drawn from one of several distributions based on their ability (e.g., Good, Neutral, or Bad). Each of these distributions has a mean and variance of alphas. The composite distribution of a sample of fund managers is therefore a mixture of Normal distributions. We estimate these distributional parameters by fitting the composite distribution to the sample of alphas. We are also able to calculate the probability a manager is of a given type conditional on their alpha. We illustrate our approach in a sample of hedge funds. We find that about half of the managers in our sample have skill, even though only 20% have alphas that are significantly positive.