Auto-suggest helps you quickly narrow down your search results by suggesting possible matches as you type.

Showing results for

Find a Community

- Home
- /
- Risk and Fraud
- /
- SAS Risk Management
- /
- Facing difficulties in Delta Normal Analysis

- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

04-04-2016 12:38 PM

Hello friends,

Need a clarification on delta normal analysis.

I am trying to estimate the portfolio VaR using a delta normal analysis where I have cosidered a variance / covariance matrix (user specified varified) for a multivariate model ( more than one risk factors).

While running the analysis I am getting the following error -

But the same matrix is working fine for a covariance based monte - carlo analysis.

Please help me in solving the issue.

Thanks,

Anindya

- Mark as New
- Bookmark
- Subscribe
- Subscribe to RSS Feed
- Permalink
- Email to a Friend
- Report Inappropriate Content

Posted in reply to Anindya

04-12-2016 10:52 AM

Hello Anindya,

I have consulted with a Principal Research Statistician in the risk organization. He says that when risk complains about an aggregation having missing sensitivities, it is not complaining about the covariance or correlation matrix. Aggregations are subportfolios. The error is telling you that some subportfolio has missing values for its sensitivities. Either the value of a position is missing at perturbed Delta Normal state or the pricing method sets rf.der to missing for some risk factor. An error might seem extreme for this kind of problem, but any missing value invalidates the calculation of VaR for that subportfolio. As lower-level sensitivities percolate to the portfolio level, the portfolio level thus contains the same missing values, which invalidates the calculation of VaR there. We recommend you check your positions for missing values.

I hope this helps,

Michael Harvey