最近在金融實證上的文獻指出，在股價報酬率的聯合分佈具有不對稱的相關性。Copula 提供一個方便的架構去描述不對稱的相關性結構。在這篇論文中，我們比較傳統上對報酬率作的常態分佈假設與使用Copula 建構出比較彈性的多元分佈。在Markowitz 的Mean-Variance 架構下，我們考慮一個風險趨避的投資者配置財富於不同的資產。我們用Copula 去建構高維度報酬率的分佈並使用模擬退火法去選擇最佳的權重。最後我們應用我們的方法於投資組合在台灣的股票市場。 Recent studies in the empirical finance literature have reported asymmetric dependence in the joint distribution of stock returns. Copula provides a convenient framework to describe asymmetric dependence structure. In this thesis, we compare traditional multivariate normal distribution assumption for return and a flexible multivariate distribution using copula. Under Markowitz’s mean-variance framework (Markowitz, 1952), we consider a risk averse investor allocating wealth among different assets. We use copula to construct high-dimensional distribution of return, and propose a simulated annealing algorithm to select the optimal portfolio weights. We apply our approach for portfolio selection in Taiwan stock market.
