Hedge Funds: The Good, the Bad, and the Lucky
提出一种基于改进EM算法和技能组正态混合模型的估计方法，用于评估对冲基金绩效，并构建结合估计alpha与技能分布的投资策略，该策略在样本外测试中优于传统方法。
We develop an estimation approach based on a modified expectation-maximization (EM) algorithm and a mixture of normal distributions associated with skill groups to assess performance in hedge funds. By allowing luck to affect both skilled and unskilled funds, we estimate the number of skill groups, the fraction of funds from each group, and the mean and variability of skill within each group. For each individual fund, we propose a performance measure combining the fund’s estimated alpha with the cross-sectional distribution of fund skill. In out-of-sample tests, an investment strategy using our performance measure outperforms those using estimated alpha and t -statistic.