Importance sampling for Value-at-Risk computations under factor model

Please use this identifier to cite or link to this item: http://ir.lib.ncu.edu.tw/handle/987654321/71911

Title:	Importance sampling for Value-at-Risk computations under factor model
Authors:	呂駿杰;Lu,Chun-Chieh
Contributors:	統計研究所
Keywords:	重要性採樣;因子模型;投資組合風險值;多元分布;Importance sampling;factor model;portfolio VaR;multivariate distribution
Date:	2016-06-29
Issue Date:	2016-10-13 14:05:58 (UTC+8)
Publisher:	國立中央大學
Abstract:	風險價值(VaR)，它被定義為一個給定的時間範圍內利潤和損失分佈的條件分位數。雖然蒙地卡羅模擬是評估投資組合風險價值最有效的方法，這種方法主要的缺點在於需要大量的計算需求。所以在本文中，我們考慮在因子模型之下的重要性採樣。此外，我們對風險因子建模在多元常態分配和多元t分配。在此假設之下，引用upper bound 和optimal tilting 兩種方法來尋找新測度的最佳解。最後，比較重要性採樣和蒙地卡羅的相對效率。我們可以看出重要性採樣顯著地比蒙地卡羅來的更好。;Value-at-Risk (VaR), which is de fined as the conditional quantile of the pro fit-and-loss distribution for a given time horizon. Although the Monte Carlo simulation is the most powerful method to evaluate portfolio VaR, a major drawback of this method is that it is computationally demanding. So in this paper, we consider the efficient importance sampling method under factor model. Furthermore, we model the risk factors with multivariate normal distribution and multivariate t distribution. Among this assumptions, we introduce upper bound method and optimal tilting method to find the alternative measure Q . In the end, we give the relative efficiency of importance sampling method and Monte Carlo method. We show that the importance sampling method is significantly better than Monte Carlo method.
Appears in Collections:	[Graduate Institute of Statistics] Electronic Thesis & Dissertation

Files in This Item:

File	Description	Size	Format
index.html		0Kb	HTML	421	View/Open

社群 sharing

Loading...