The Guassian Cupola Correlation Model

Gauss

I misspelled copula as well.

Never met one. As a trader, my dictum has always been to never buy what the Street is buying (or selling), as in this case.

This was all laid out in a book entitled "The (Mis)Behavior of Markets" by Mandelbrot in 2004.

but secondandc, Moodys didn't apply the modeling concept until late 2004.

That article is missing half the story.

Basically this article can be summarized with these two sentences:

"Borrowing from his work in actuarial science and insurance and his knowledge of the broken-heart syndrome, he attempted to solve one of Wall Street quants’ most intractable problems: default correlation."
"He decided to use a very standard type of curve – the Gaussian copula, which is better known as a bell curve, or normal distribution – to map and determine the correlation on any given portfolio of assets"

The key part that is missing is what he used as historical market data to feed his model.
Because historical default data is sparse he used historical CDS prices as a proxy (the higher the price, the higher the default probability that was used to price it) but the fatal flaw with that is that the CDS market is only a little over a decade old which correlated nicely with the biggest RE boom in recent history. No wonder rating agencies and IBs were falling over themselves to create triple A CDOs.

It's a bit like LTCM - which points to another flaw in the FT article - they only went back about 5 years for their market data, a period during which the "historical" correlation numbers they were using were flat and consistent. Had they gone back a few years more, they would have seen similar correlation convergence that killed them in the end.

AR, why did you reference Felix Salmon. He wrote the piece that I am referencing but I do not see his name in the FT article.
http://www.wired.com/techbiz/it/magazine/17-03/wp_quant?currentPage=all

I found the reference in another Salmon piece. If I pull off of a source rather than finding it myself I like to give credit when I remember to. I haven't read the wired article yet. Thanks for your explanation. I'm pretty good on the general macro stuff, but TA and elsewhere is a new learning curve for me. I just thought the story was kind of fascinating. Such a little thing, such huge ramifications.