wilmott.com

Does anyone have direct experience or know secondhand if the Malliavin Calculus is used in production anywhere for calculating sensitivities for MC simulations of mortgage products, especially CMOs? If so, does Malliavin produce substantially better results than the brute-force bump-revalue method? The stability of prepayment sensitivity (with respect to overall prepayment magnitude, or even sensitivities with respect to key parameters in a prepayment model) is especially interesting. Stability means some sort of error analysis with respect to the output from different MC runs on the same portfolio with different seeds, increasing the number of paths, different random number generators, etc. Even anecdotal stories would be interesting. Thanks.

QuoteOriginally posted by: JanDashDoes anyone have direct experience or know secondhand if the Malliavin Calculus is used in production anywhere for calculating sensitivities for MC simulations of mortgage products, especially CMOs? If so, does Malliavin produce substantially better results than the brute-force bump-revalue method? The stability of prepayment sensitivity (with respect to overall prepayment magnitude, or even sensitivities with respect to key parameters in a prepayment model) is especially interesting. Stability means some sort of error analysis with respect to the output from different MC runs on the same portfolio with different seeds, increasing the number of paths, different random number generators, etc. Even anecdotal stories would be interesting. Thanks.Jan, I've looked, a little, at Malliavin Calculus, as I was working on Monte Carlo simulations of Cliquet options with global caps and floors. It is very often noted that the practical applicability of the method is limited. I've not seen it applied specifically to CMOs or MBSs, in general.

After several days of waiting and the negative response of one mortgage expert, it appears that the Malliavin Calculus is at least not commonly used in production for MBS products, as I suspected. So here is a less specific question:Are there any mortgage quants out there who have tried Malliavin Calculus for MBS, CMO, IO/PO etc. products in any fashion, or who have thought about trying it? Or even, is there anyone out there who knows of any mortgage quants who have tried it or thought about it?

Jan, I'm guessing you've found an application that works

That's one possibility in my mind, however I am still pretty far from trying it....

Here's the scoop...Malliavin Calculus adds no value. It has the same limitations as usual stochastic calculus and BTW for the same reasons. In fact, they're the samereasons most physicists still use perturbative techniques... Lousy math skills !!!N

QuoteOriginally posted by: N... Lousy math skills !!!NAs a physicist, should I be insulted??

As a physicist, should I be insulted?? Alan,Physicists have this thing with hermitian forms, perturbations, etc. Keep in mind that almost all solutions in physics are tessellations. Life isn't quite so simple. N

Hey Jan. I used to be involved in MBS and currenly use Malliavin practically (where it actually helps and is not hunting flies with a mortar). I am not aware of any applications to MBS, and even some of the interesting areas of MBS, prepayment, hedging, I can't think of any practical application of Malliavin, esp any application that can't be attacked from a simpler angle (cf. Neumann problems with Ito.... why?)Newton, you are still a jackass.

Newton, you are still a jackass. C3,Thanks. You make me feel way too comfortable with my lead in mathematics.Do you still think a correlation matrix needs to be symmetric?What do you think the boys at RenTec are doing? You f'n idiot...Newton

QuoteOriginally posted by: NAs a physicist, should I be insulted?? Alan,Physicists have this thing with hermitian forms, perturbations, etc. Keep in mind that almost all solutions in physics are tessellations. Life isn't quite so simple. NNTo me, the physical world is... (not simple --- is-> exists) Mathematics is simply a tool to help explain natural phenomena, it does not drive natural phenomena. Natural phenomena does not fit into theorems and corollaries. Truly, "life isn't quite so simple", but one can overcomplicate it by trying to fit physical observations into hard and fast mathematical rules. That is far from saying that formal mathematics can't be useful. I'm simply saying to explain or understand physical occurences may not need all of the rigorous workings of formal mathematics..... but all this seems to drift from the topic of this thread....Regards,AlanB

I drift further from the topic thread...Reality isn't constrained by our ability to describe it. Period.CC

Hi everybody. Here is the topic of this thread: How can we best get prepayment sensitivities of mortgage products (especially illiquid CMOs/IOPOs ...). This topic is important, since mortgage desks can lose a lot of money due to prepayment problems.Mortgage products are evaluated by brute force using Monte Carlo, incorporating the complex logic of the prepayment model, CMO contract logic, etc. The usual procedure to get the prepayment sensitivity is also brute force: change the prepayment parameters (or at least the overall normalization), and revalue the portfolio. Another theoretical possibility is the Malliavin Calculus, and I am trying to find out if anyone has actually used this for mortgage products and if it works. If in addition someone knows of advanced mathematical techniques that actually and demonstrably work for the specific topic of prepayment sensitivities of mortgage products, I would be happy to learn about them. I will mention that once I watched a real-life episode of a high-powered mathematician who was trying to push what he claimed was a superior technique, but then it turned out he didn't know what a CMO actually was, and his technique was completely irrelevant, inapplicable, and a waste of time.That said, if anyone has some wisdom on this specific topic, please respond. Thanks.----------------

Would not call it advanced, but used to use Poisson intesity for pp times. These were on IO strips.

QuoteOriginally posted by: ClosetChartistI drift further from the topic thread...Reality isn't constrained by our ability to describe it. Period.CCWell said