The Relevance of the Distributional Form of Common Stock Returns to the Construction of Optimal Portfolios
比较了股票收益服从高斯分布与稳定帕累托-莱维分布两种假设下，构建有效前沿的稳健性，发现高斯假设及其统计方法更优。
In this paper, we compare the robustness in application of the Gaussian assumption of security return distributions to the robustness of the general stable assumption. Using actual stock return data to simulate the “real world,†a stock market is constructed in which stock returns conform to a Gaussian distribution as well as to a stable Pareto-Levy distribution. Using these two sets of stock returns, efficient frontiers are generated under both assumptions of parametric environments. It is shown that the Gaussian assumption, and its incumbent statistical techniques, is preferable to the general stable assumption.