OBTAINING AND PARAMETERIZING MULTIPERIOD PORTFOLIOS WITH DESIRABLE CHARACTERISTICS UNDER LOGNORMAL RETURNS
提出一个简单算法，在假设投资组合收益服从对数正态分布时，获取最大化期望几何平均收益的投资组合，并推导其风险收益特征，用纽交所股票数据验证。
ABSTRACT The expected geometric mean has been shown to be a valuable criterion in ranking portfolios of assets. For example, under certain conditions the maximum expected geometric‐mean portfolio minimizes the expected time to amass a fixed level of wealth as wealth becomes “large.” This paper develops a simple algorithm for obtaining such portfolios when portfolio returns are assumed to be lognormally distributed. The risk‐return characteristics of portfolios are derived and illustrated using data from a subset of stocks found on the New York Stock Exchange.