Re: [Quantlib-dev] SABR/CMS/Vol Interpolation...

[Toyin A. <toy...@ho...>]

However,

I believe I have found 2 other issues with the new CMS pricer.

1) The computation of the mDelta variable looks incorrect as line #284 
creates a leg from spot and the top (numer) expression of line #289 is thus 
huge. For a deal set in-arrears the numer formula should be zero and not 
("PaymentDate - SpotDate") which for a CMS rate starting in 5 years, 
Quarterly fixed resets would equal 5.25. Line #284 needs to pass the CMS 
swaps fixing date and not the YieldCurve's ReferenceDate.

2) Line #214 returns the size of the schedule object which is not the same 
as the number of periods within the underlying swap. One needs to reduce the 
calculation by one.

Also, I do know that this class is unfinished and thus untested, however I 
have an interest in this model as I previously integrated this model into a 
pricing application while I worked for Jerome Lebuchoux (Author of Models 
for CMS Caps - Euro Derivative, Risk 2000) and M Musiela (the M in the BGM 
model).

It's nice to see a clear C++ version of the algorithm.

Toy out...

>From: "Toyin Akin" <toy...@ho...>
>To: qua...@li...
>Subject: Re: [Quantlib-dev] SABR/CMS/Vol Interpolation...
>Date: Mon, 24 Jul 2006 15:45:08 +0100
>
>
>Hi,
>
>My observations concerning point 2) is invalid as I have now associated it
>with equ 2.12
>
>Toy out.
>
>
> >From: "Toyin Akin" <toy...@ho...>
> >To: qua...@li...
> >Subject: [Quantlib-dev] SABR/CMS/Vol Interpolation...
> >Date: Mon, 24 Jul 2006 11:17:33 +0100
> >
> >
> >Hi all,
> >
> >I have a few questions for you...
> >
> >1) Volatility interpolation...
> >
> >Let's assume that one has built a volatility smile based on linear
> >interpolation or a calibrated SABR curve and these volatilities are based
> >on
> >5Y A/MM underlying swap rates. Now some annoying dude comes along and 
>wants
> >to price a swaption based on 5Y S/BB swaps. However only 5Y A/MM vols are
> >quoted in the market. How does one go about adjusting the  5Y S/BB strike
> >rate to that of the 5Y A/MM strike so that an interpolation can be used? 
>Is
> >this enough? Does one need to modify the interpolated volatility also? Or
> >do
> >I tell the dude "we can't price it"!!! :-)
> >
> >2)
> >
> >Concerning the 'unfinished' CMS convexity adjustment class via 
>replication.
> >Looking at the code (line #235 of the conundrumpricer.cpp class.).
> >Shouldn't
> >this be x - forward (rather than x - strike)?
> >
> >3)
> >
> >Also based on the  'unfinished' CMS convexity adjustment class, I noticed
> >that the integration limit for CAPs is set to 1.0 (the strike value will
> >move from the given value all the way up to 1.0). If I am reading this
> >right
> >and one uses a SABR model with mid to high-ish wing values, you can imply
> >volatility values over 100% during the replication!! Is it enough to
> >instead
> >compute some percentage movement from the given strike (ie - 40%)? The 
>same
> >is true for the downward direction.
> >
> >That's all folks...
> >
> >Toy out...
> >
> >
> >
> >-------------------------------------------------------------------------
> >Take Surveys. Earn Cash. Influence the Future of IT
> >Join SourceForge.net's Techsay panel and you'll get the chance to share
> >your
> >opinions on IT & business topics through brief surveys -- and earn cash
> >http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
> >_______________________________________________
> >QuantLib-dev mailing list
> >Qua...@li...
> >https://lists.sourceforge.net/lists/listinfo/quantlib-dev
>
>
>
>-------------------------------------------------------------------------
>Take Surveys. Earn Cash. Influence the Future of IT
>Join SourceForge.net's Techsay panel and you'll get the chance to share 
>your
>opinions on IT & business topics through brief surveys -- and earn cash
>http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
>_______________________________________________
>QuantLib-dev mailing list
>Qua...@li...
>https://lists.sourceforge.net/lists/listinfo/quantlib-dev