Yield Curve Boostraping Improvements
Locked
 
|
So far I've been quite happy with the functionality in the QuantLib library - I'd like to start using it to compare to market pricing. I've found that the curve stripping methodology produces extremely jagged forward curves which result in poor pricing of several market structures.
I'd like to add a curve stripping methodology that I've been using with an internal library at my current work though I definitely need some help in adding such a change to the library. Also, some algorithmic help in dynamically selecting knot points would be much appreciated as well. The basic idea is the creation of a smooth interpolation method for a selected set of knot points for a given curve. This is done using an exponential spline across the term structure. Let me know if anyone can help me with this, I'd be glad to discuss the algorithm further if the support is there. - Luis |
Re: Yield Curve Boostraping Improvements
|
|
Hi Luis
> algorithmic help in dynamically selecting knot points would be > much appreciated as well. some algorithms for instruments selection are defined in QuantLibAddin, see qlRateHelperSelection in qlo/ratehelpers.hpp. They are not in QuantLib as Luigi has preferred to keep them in the application layer out of the analytic library so far. > using an exponential spline across the term structure. This is something I've been planning for a long time now. We have Log-Linear interpolation, but what we really need is a logarithmic adapter of any available interpolation. Then my favorite approach would be monotone-cubic interpolation of log-discounts ciao -- Nando ------------------------------------------------------------------------- This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2005. http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ _______________________________________________ QuantLib-dev mailing list [hidden email] https://lists.sourceforge.net/lists/listinfo/quantlib-dev |
Re: Yield Curve Boostraping Improvements
|
|
In reply to this post by newbie73
On Mon, 2007-10-01 at 07:20 -0700, newbie73 wrote:
> So far I've been quite happy with the functionality in the QuantLib > library - I'd like to start using it to compare to market pricing. > I've found that the curve stripping methodology produces extremely > jagged forward curves which result in poor pricing of several market > structures. Does this depend on the chosen interpolation? > Also, some algorithmic help in dynamically selecting knot points would be > much appreciated as well. The basic idea is the creation of a smooth > interpolation method for a selected set of knot points for a given curve. "Selecting knot points" as in "choosing which instruments to use among the available ones" (which is what Nando referred to) or as in "choosing a set of knots freely, i.e., not necessarily corresponding to instrument maturities"? Later, Luigi -- This gubblick contains many nonsklarkish English flutzpahs, but the overall pluggandisp can be glorked from context. -- David Moser ------------------------------------------------------------------------- This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2005. http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ _______________________________________________ QuantLib-dev mailing list [hidden email] https://lists.sourceforge.net/lists/listinfo/quantlib-dev |
|
|
The choppy forward curves appear regardless of which interpolation method is used. The term structure of 3m or 6m forward rates results in a saw-tooth term structure which deviates from market by huge amounts.
Regarding the knot points, I was referring to picking maturities that do not necessarily correspond with the chosen instruments. The methodology we use here treats the area between each knot point as its own curve (or curve function), so you end up with many separate curves which segment the yield curve which are then combined to ensure smooth forward curve generation. The technique itself is quite old and was developed by Lehman in the late 80s, but works well in most cases I've encountered. - Luis
|
Re: Yield Curve Boostraping Improvements
|
|
You might want to examine the paper by Hagan&West on smooth yieldcurve construction. Their convex monotone spline method gives great results.
Simon
On 10/2/07, newbie73 <[hidden email]> wrote:
------------------------------------------------------------------------- This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2005. http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ _______________________________________________ QuantLib-dev mailing list [hidden email] https://lists.sourceforge.net/lists/listinfo/quantlib-dev |
Re: Yield Curve Boostraping Improvements
|
|
In reply to this post by newbie73
Yes, I confirm the
Hagan-West paper as a key reference.
Luis, is the
method you mention published somewhere ?
Ciao Marco
------------------------------------------------------------------------- This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now >> http://get.splunk.com/ _______________________________________________ QuantLib-dev mailing list [hidden email] https://lists.sourceforge.net/lists/listinfo/quantlib-dev |
|
|
Is the Hagan-West paper titled, "Interpolation Methods for Curve Construction" ? If so, then I just printed a copy of it last week. As far as the method I referred to in my earlier post, it is not published publicly. It was written as internal research at Lehman Brothers in the late 80s, so I have an older printed photocopy laying around that I used as a reference.
- Luis
|
Re: Yield Curve Boostraping Improvements
|
|
Hi!
In a case like this a look at scholar.google.com is strongly recommended. There one finds under Appl. Math. Fin., Vol. 13, No. 2, 89-129 the correct reference to the already published paper. Rgds Frank -------- Original-Nachricht -------- > Datum: Tue, 9 Oct 2007 04:31:37 -0700 (PDT) > Von: newbie73 <[hidden email]> > An: [hidden email] > Betreff: Re: [Quantlib-dev] Yield Curve Boostraping Improvements > > Is the Hagan-West paper titled, "Interpolation Methods for Curve > Construction" ? If so, then I just printed a copy of it last week. As > far > as the method I referred to in my earlier post, it is not published > publicly. It was written as internal research at Lehman Brothers in the > late 80s, so I have an older printed photocopy laying around that I used > as > a reference. > > - Luis > > > > Bianchetti Marco-2 wrote: > > > > Yes, I confirm the Hagan-West paper as a key reference. > > Luis, is the method you mention published somewhere ? > > Ciao > > Marco > > > > > > -----Original Message----- > > From: [hidden email] > > [mailto:[hidden email]] On Behalf Of Simon > > Ibbotson > > Sent: 02 October 2007 15:03 > > To: newbie73 > > Cc: [hidden email] > > Subject: Re: [Quantlib-dev] Yield Curve Boostraping Improvements > > > > > > You might want to examine the paper by Hagan&West on smooth > > yieldcurve construction. Their convex monotone spline method gives great > > results. > > > > Simon > > > > > > > > On 10/2/07, newbie73 <[hidden email]> wrote: > > > > > > The choppy forward curves appear regardless of which > > interpolation method is > > used. The term structure of 3m or 6m forward rates > > results in a saw-tooth > > term structure which deviates from market by huge > > amounts. > > > > Regarding the knot points, I was referring to picking > > maturities that do not > > necessarily correspond with the chosen instruments. The > > methodology we use > > here treats the area between each knot point as its own > > curve (or curve > > function), so you end up with many separate curves which > > segment the yield > > curve which are then combined to ensure smooth forward > > curve generation. > > > > The technique itself is quite old and was developed by > > Lehman in the late > > 80s, but works well in most cases I've encountered. > > > > - Luis > > > > > > Luigi Ballabio wrote: > > > > > > On Mon, 2007-10-01 at 07:20 -0700, newbie73 wrote: > > > > > >> So far I've been quite happy with the functionality > > in the QuantLib > > >> library - I'd like to start using it to compare to > > market pricing. > > >> I've found that the curve stripping methodology > > produces extremely > > >> jagged forward curves which result in poor pricing of > > several market > > >> structures. > > > > > > Does this depend on the chosen interpolation? > > > > > >> Also, some algorithmic help in dynamically selecting > > knot points would be > > >> much appreciated as well. The basic idea is the > > creation of a smooth > > >> interpolation method for a selected set of knot > > points for a given curve. > > > > > > "Selecting knot points" as in "choosing which > > instruments to use among > > > the available ones" (which is what Nando referred to) > > or as in "choosing > > > a set of knots freely, i.e., not necessarily > > corresponding to instrument > > > maturities"? > > > > > > Later, > > > Luigi > > > > > > > > > -- > > > > > > This gubblick contains many nonsklarkish English > > flutzpahs, but the > > > overall pluggandisp can be glorked from context. > > > -- David Moser > > > > > > > > > > > > > > ------------------------------------------------------------------------ > > - > > > This SF.net email is sponsored by: Microsoft > > > Defy all challenges. Microsoft(R) Visual Studio 2005. > > > > > http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ > > > _______________________________________________ > > > QuantLib-dev mailing list > > > [hidden email] > > > > > https://lists.sourceforge.net/lists/listinfo/quantlib-dev > > > > > > > > > > -- > > View this message in context: > > http://www.nabble.com/Yield-Curve-Boostraping-Improvements-tf4548647.htm > > l#a12997264 > > Sent from the quantlib-dev mailing list archive at > > Nabble.com. > > > > > > > > ------------------------------------------------------------------------ > > - > > This SF.net email is sponsored by: Microsoft > > Defy all challenges. Microsoft(R) Visual Studio 2005. > > http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ > > _______________________________________________ > > QuantLib-dev mailing list > > [hidden email] > > > > https://lists.sourceforge.net/lists/listinfo/quantlib-dev > > > > > > > > > > > ------------------------------------------------------------------------- > > This SF.net email is sponsored by: Splunk Inc. > > Still grepping through log files to find problems? Stop. > > Now Search log events and configuration files using AJAX and a browser. > > Download your FREE copy of Splunk now >> http://get.splunk.com/ > > _______________________________________________ > > QuantLib-dev mailing list > > [hidden email] > > https://lists.sourceforge.net/lists/listinfo/quantlib-dev > > > > > > -- > View this message in context: > http://www.nabble.com/Yield-Curve-Boostraping-Improvements-tf4548647.html#a13113782 > Sent from the quantlib-dev mailing list archive at Nabble.com. > > > ------------------------------------------------------------------------- > This SF.net email is sponsored by: Splunk Inc. > Still grepping through log files to find problems? Stop. > Now Search log events and configuration files using AJAX and a browser. > Download your FREE copy of Splunk now >> http://get.splunk.com/ > _______________________________________________ > QuantLib-dev mailing list > [hidden email] > https://lists.sourceforge.net/lists/listinfo/quantlib-dev -- Der GMX SmartSurfer hilft bis zu 70% Ihrer Onlinekosten zu sparen! Ideal für Modem und ISDN: http://www.gmx.net/de/go/smartsurfer ------------------------------------------------------------------------- This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now >> http://get.splunk.com/ _______________________________________________ QuantLib-dev mailing list [hidden email] https://lists.sourceforge.net/lists/listinfo/quantlib-dev |
«
Return to quantlib-dev
|
1 view|%1 views
| Free forum by Nabble | Edit this page |