QuantLib › quantlib-users

Gaussian 1D Models

Classic

List

Threaded

4 messages Options

cheng li

Gaussian 1D Models

Hi Pcaspers and teams,

Recently you add a new example called Gaussian1dModels in quantlib/examples. Really appreciate your kind effort to share your expertise on this topic. Currently I am learning from your example and try to understand more…

Some questions that I need your help…

1. Would you like to share with me the reference literature about Gaussian 1D models? I have the 3 volume books written by Anderson and Piterbarg. There are topics on Short rate models and quasi-Gaussan models in volume 2. Which is the most relevant to your codes?

2. As I saw you wirte one line in your example: “The model is a one factor Hull White model with piecewise volatility”. So can I know what is the essential difference between Gaussian 1D models and existing short rate models which are located under folder models/shortrates? Is this a new numerical scheme?

Really expecting to hear from you.

Regards,

Cheng

------------------------------------------------------------------------------
"Accelerate Dev Cycles with Automated Cross-Browser Testing - For FREE
Instantly run your Selenium tests across 300+ browser/OS combos.
Get unparalleled scalability from the best Selenium testing platform available
Simple to use. Nothing to install. Get started now for free."
http://p.sf.net/sfu/SauceLabs
_______________________________________________
QuantLib-users mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/quantlib-users

Peter Caspers-4

Re: Gaussian 1D Models

Hi Cheng,

there is a one factor Hull White and a one factor Markov functional
model in experimental / models, both derived from a common base class
Gaussian1dModel (which essentially requires the implementation of a
numeraire and a zerobond method). Regarding the Markov model there are
references given in the source code. The Hull White model is an
implementation of Piterbarg, formula (10.14), with piecewise reversion
\kappa and volatility \sigma, so you could start reading around this
place in the second volume. The pricing engines in the same folder are
accepting both models (actually any Gaussian1dModel instance).

There is no particular reason for the redundancy with the classes in
the models folder. First there was the Markov model, based on
CalibratedModel. I copied the code and amended it to get a Hull White
model and eventually extracted the common things. Maybe it would have
been wiser to generalize the existing Hull White implementation. Well.

I guess the Gsr class is a bit more flexible compared to HullWhite in
that both reversion and sigma can be made time dependent (although I
think I saw some developments in that direction in some experimental
sister folder, too (?)). Also multiple curve setups are supported (no
stochastic basis though) and you can include a credit spread to price
optional parts of exotic bonds. The pricing engines are all using
numerical integration against the state variable density. I really
want to add FD engines very soon (or you could try maybe ? :-) ). I
have some confidence in the results produced from both models since we
used them to validate corresponding models in a commercial trading
system across a range of products.

Hope that helps.

Thank you
Peter

On 23 May 2014 04:48, cheng.li <[hidden email]> wrote:

> Hi Pcaspers and teams,
>
>
>
> Recently you add a new example called Gaussian1dModels in quantlib/examples.
> Really appreciate your kind effort to share your expertise on this topic.
> Currently I am learning from your example and try to understand more…
>
>
>
> Some questions that I need your help…
>
>
>
> 1. Would you like to share with me the reference literature about
> Gaussian 1D models? I have the 3 volume books written by Anderson and
> Piterbarg. There are topics on Short rate models and quasi-Gaussan models in
> volume 2. Which is the most relevant to your codes?
>
>
>
> 2. As I saw you wirte one line in your example: “The model is a one
> factor Hull White model with piecewise volatility”. So can I know what is
> the essential difference between Gaussian 1D models and existing short rate
> models which are located under folder models/shortrates? Is this a new
> numerical scheme?
>
>
>
> Really expecting to hear from you.
>
>
>
> Regards,
>
> Cheng

cheng li

答复: Gaussian 1D Models

Hi Peter,

Thanks for your explanation. It really helps me a lot to understand the codes.

BTW I have searched on SSRN and find one paper written by you on the topic of 1D GSR. It goes through all the hard calculus and I think it will be beneficial to me.

I am starting digging into your paper and codes. Thanks for your wonderful work.

Regards,
Cheng

-----邮件原件-----
发件人: Peter Caspers [mailto:[hidden email]]
发送时间: 2014年5月24日 2:53
收件人: cheng.li
抄送: QuantLib users
主题: Re: Gaussian 1D Models

Hi Cheng,

there is a one factor Hull White and a one factor Markov functional model in experimental / models, both derived from a common base class Gaussian1dModel (which essentially requires the implementation of a numeraire and a zerobond method). Regarding the Markov model there are references given in the source code. The Hull White model is an implementation of Piterbarg, formula (10.14), with piecewise reversion \kappa and volatility \sigma, so you could start reading around this place in the second volume. The pricing engines in the same folder are accepting both models (actually any Gaussian1dModel instance).

There is no particular reason for the redundancy with the classes in the models folder. First there was the Markov model, based on CalibratedModel. I copied the code and amended it to get a Hull White model and eventually extracted the common things. Maybe it would have been wiser to generalize the existing Hull White implementation. Well.

I guess the Gsr class is a bit more flexible compared to HullWhite in that both reversion and sigma can be made time dependent (although I think I saw some developments in that direction in some experimental sister folder, too (?)). Also multiple curve setups are supported (no stochastic basis though) and you can include a credit spread to price optional parts of exotic bonds. The pricing engines are all using numerical integration against the state variable density. I really want to add FD engines very soon (or you could try maybe ? :-) ). I have some confidence in the results produced from both models since we used them to validate corresponding models in a commercial trading system across a range of products.

Hope that helps.

Thank you
Peter

On 23 May 2014 04:48, cheng.li <[hidden email]> wrote:

> Hi Pcaspers and teams,
>
>
>
> Recently you add a new example called Gaussian1dModels in quantlib/examples.
> Really appreciate your kind effort to share your expertise on this topic.
> Currently I am learning from your example and try to understand more…
>
>
>
> Some questions that I need your help…
>
>
>
> 1. Would you like to share with me the reference literature about
> Gaussian 1D models? I have the 3 volume books written by Anderson and
> Piterbarg. There are topics on Short rate models and quasi-Gaussan
> models in volume 2. Which is the most relevant to your codes?
>
>
>
> 2. As I saw you wirte one line in your example: “The model is a one
> factor Hull White model with piecewise volatility”. So can I know what
> is the essential difference between Gaussian 1D models and existing
> short rate models which are located under folder models/shortrates? Is
> this a new numerical scheme?
>
>
>
> Really expecting to hear from you.
>
>
>
> Regards,
>
> Cheng

Winkausyar Wanranto

Re: 答复: Gaussian 1D Models

M bm Aa

—
Sent from Mailbox

On Sun, May 25, 2014 at 8:59 PM, cheng.li <[hidden email]> wrote:

Hi Peter,

Thanks for your explanation. It really helps me a lot to understand the codes.

BTW I have searched on SSRN and find one paper written by you on the topic of 1D GSR. It goes through all the hard calculus and I think it will be beneficial to me.

I am starting digging into your paper and codes. Thanks for your wonderful work.

Regards,
Cheng

-----邮件原件-----
发件人: Peter Caspers [mailto:[hidden email]]
发送时间: 2014年5月24日 2:53
收件人: cheng.li
抄送: QuantLib users
主题: Re: Gaussian 1D Models

Hi Cheng,

there is a one factor Hull White and a one factor Markov functional model in experimental / models, both derived from a common base class Gaussian1dModel (which essentially requires the implementation of a numeraire and a zerobond method). Regarding the Markov model there are references given in the source code. The Hull White model is an implementation of Piterbarg, formula (10.14), with piecewise reversion \kappa and volatility \sigma, so you could start reading around this place in the second volume. The pricing engines in the same folder are accepting both models (actually any Gaussian1dModel instance).

There is no particular reason for the redundancy with the classes in the models folder. First there was the Markov model, based on CalibratedModel. I copied the code and amended it to get a Hull White model and eventually extracted the common things. Maybe it would have been wiser to generalize the existing Hull White implementation. Well.

I guess the Gsr class is a bit more flexible compared to HullWhite in that both reversion and sigma can be made time dependent (although I think I saw some developments in that direction in some experimental sister folder, too (?)). Also multiple curve setups are supported (no stochastic basis though) and you can include a credit spread to price optional parts of exotic bonds. The pricing engines are all using numerical integration against the state variable density. I really want to add FD engines very soon (or you could try maybe ? :-) ). I have some confidence in the results produced from both models since we used them to validate corresponding models in a commercial trading system across a range of products.

Hope that helps.

Thank you
Peter

On 23 May 2014 04:48, cheng.li <[hidden email]> wrote:

> Hi Pcaspers and teams,
>
>
>
> Recently you add a new example called Gaussian1dModels in quantlib/examples.
> Really appreciate your kind effort to share your expertise on this topic.
> Currently I am learning from your example and try to understand more…
>
>
>
> Some questions that I need your help…
>
>
>
> 1. Would you like to share with me the reference literature about
> Gaussian 1D models? I have the 3 volume books written by Anderson and
> Piterbarg. There are topics on Short rate models and quasi-Gaussan
> models in volume 2. Which is the most relevant to your codes?
>
>
>
> 2. As I saw you wirte one line in your example: “The model is a one
> factor Hull White model with piecewise volatility”. So can I know what
> is the essential difference between Gaussian 1D models and existing
> short rate models which are located under folder models/shortrates? Is
> this a new numerical scheme?
>
>
>
> Really expecting to hear from you.
>
>
>
> Regards,
>
> Cheng

------------------------------------------------------------------------------
"Accelerate Dev Cycles with Automated Cross-Browser Testing - For FREE
Instantly run your Selenium tests across 300+ browser/OS combos.
Get unparalleled scalability from the best Selenium testing platform available
Simple to use. Nothing to install. Get started now for free."
http://p.sf.net/sfu/SauceLabs
_______________________________________________
QuantLib-users mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/quantlib-users

------------------------------------------------------------------------------
Time is money. Stop wasting it! Get your web API in 5 minutes.
www.restlet.com/download
http://p.sf.net/sfu/restlet
_______________________________________________
QuantLib-users mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/quantlib-users