Modeling asset returns under time-varying semi-nonparametric distributions
扩展了半非参数密度，在时变条件下用GJR-GARCH建模收益，得到偏矩和预期损失的闭式解，并基于偏度-峰度前沿与Hansen偏斜t分布比较，最后用S&P 100股票进行样本外组合选择。
We extend the semi-nonparametric (SNP) density of We estimate robust tail-indexes for testing the existence of the unconditional higher-order moments. We obtain closed-form expressions of partial moments and expected shortfall under the time-varying SNP density with the GJR-GARCH for modeling returns. A comparative study between SNP and Hansen's skewed-t, based on skewness-kurtosis frontiers, in-sample and backtesting analyses, is also implemented. Finally, we conduct an out-of-sample portfolio selection exercise for the stocks of the S&P 100 index through an equity screening method based on our parametric one-sided reward/risk performance measures and compare with the Sharpe ratio portfolio.