Re: [Quantlib-dev] Credit curves

[Simon I. <s.i...@gm...>]

Luigi,

Any further notice on this? If anyone is developing this at present, I'll
wait a bit for a well thought out base class interface - but sometimes it's
best just to go with a simple one.
The simplest (I feel) would be one with the basic functions:

virtual QL_Real NpvNoDefault(const Cashflow& Payment, const Date& StartDate,
const Date& EndDate) const = 0;
virtual QL_Real NpvGivenDefault(const Cashflow& Payment, const Date&
StartDate, const Date& EndDate) const = 0;

You might also want to add another parameter that makes explicit whether a
default on the end date is included.

This interface allows for stochastic default-intensity and stochastic
interest-rates. I've not worked with firm-value models, so don't know how
this interface would hold up. It also specialises for single-name credit -
multiple names would require a different interface (e.g. expected remaining
notional for an ITraxx curve).

However, if anyone is working on this and has a better concept, please let
me know...

Simon



On 9/26/07, Luigi Ballabio <lui...@gm...> wrote:
>
> On Wed, 2007-09-26 at 14:18 +0100, Simon Ibbotson wrote:
> > Hi, is there any developer out there planning to contribute a CDS /
> > bond curves (using reduced form approach)? I wouldn't want to
> > duplicate any work.
>
> There might be a contribution shortly. I don't think that it uses
> reduced form approach, but it should at least define a base interface
> for the default-probability curve. I'll get back as soon as I know for
> sure.
>
> Luigi
>
>
> --
>
> Use every man after his desert, and who shall scape whipping?
> -- Hamlet, Act II, scene II
>
>
>