Post by Sapphire Capital on Oct 24, 2008 18:56:18 GMT 4
A Style Based Market Risk Model for Hedge Fund Portfolios
Jason Zhou
Wachovia Securities
Adam Litke
Wachovia Securities
Michael McLaughlin
Wachovia Securities
October 21, 2008
Abstract:
Hedge funds have become an alternative asset class in recent years. However, how to estimate the risk of hedge fund portfolios remains a challenging problem. In this paper, we developed a market risk model by integrating cluster analysis, extreme value theory (EVT) and copula modeling. The cluster models are based exclusively on fund returns, and thus provide more objective style information than the self-reported styles. The EVT part considers the fat tail of fund returns of clusters, implicitly accounting for jump risk. The copula method is used to consider the dependence between clusters. A t copula is constructed based on the average returns of funds in the clusters, and a Clayton copula is constructed based on returns of all funds in the clusters. The t copula is appropriate for simulations on highly diversified portfolios, while the Clayton copula is suitable for simulations on nondiversified portfolios and also provides a tool for stress tests. Monte Carlo simulations indicate that investing in a broadly diversified hedge fund portfolio is not riskier than investing in a traditional diversified portfolio. The cluster models of the system can also be used as a screening tool to detect style drift of hedge funds.
papers.ssrn.com/sol3/Delivery.cfm/SSRN_ID1287641_code1062950.pdf?abstractid=1156356&mirid=1
Jason Zhou
Wachovia Securities
Adam Litke
Wachovia Securities
Michael McLaughlin
Wachovia Securities
October 21, 2008
Abstract:
Hedge funds have become an alternative asset class in recent years. However, how to estimate the risk of hedge fund portfolios remains a challenging problem. In this paper, we developed a market risk model by integrating cluster analysis, extreme value theory (EVT) and copula modeling. The cluster models are based exclusively on fund returns, and thus provide more objective style information than the self-reported styles. The EVT part considers the fat tail of fund returns of clusters, implicitly accounting for jump risk. The copula method is used to consider the dependence between clusters. A t copula is constructed based on the average returns of funds in the clusters, and a Clayton copula is constructed based on returns of all funds in the clusters. The t copula is appropriate for simulations on highly diversified portfolios, while the Clayton copula is suitable for simulations on nondiversified portfolios and also provides a tool for stress tests. Monte Carlo simulations indicate that investing in a broadly diversified hedge fund portfolio is not riskier than investing in a traditional diversified portfolio. The cluster models of the system can also be used as a screening tool to detect style drift of hedge funds.
papers.ssrn.com/sol3/Delivery.cfm/SSRN_ID1287641_code1062950.pdf?abstractid=1156356&mirid=1