標題: | 以相對熵的重要抽樣法計算投資組合信用風險 Importance Sampling via The Cross-Entropy for Portfolio Credit Risk |
作者: | 楊哲宇 韓傳祥 李明佳 Yang, Che-Yu Han, Chuan-Hsiang Li, Ming-Chia 應用數學系所 |
關鍵字: | 投資組合信用風險;重要抽樣;相對熵;蒙地卡羅;條件抽樣;portfolio credit risk;importance sampling;cross-entropy;Monte Carlo method;conditional sampling |
公開日期: | 2016 |
摘要: | 我們將用重要抽樣法去提升投資組合風險機率計算問題的效率,利用
最小相對熵和某些限制式去決定在重要抽樣下使用的機率測度。在此篇
論文,我們考慮常態分配與學生-t 分配的兩種模型並且與Glasserman,
Chan and Kroese,Scott 所提出的三種方法比較。數值結果顯示,雖然
我們的方法精準度略低,但執行我們的重要抽樣法所需的計算時間比起
來相對少很多。 We use the importance sampling to increase the efficiency of estimating the probability of the portfolio credit risk and make use of the cross-entropy and some constraint to decide the importance sampling measure. In this thesis, we consider the normal copula and t copula and compare with Scott’s, Glasserman’s, and Chan and Kroese’s methods. From the numerical results, our importance sampling takes lesser computational time than others’ though our method losses a little accuracy. |
URI: | http://etd.lib.nctu.edu.tw/cdrfb3/record/nctu/#GT070352218 http://hdl.handle.net/11536/138633 |
Appears in Collections: | Thesis |