Replying to a Message

Pure Recruitment

Forum Navigation:

login

FORUMS > Book And Research Paper Forum

Replying to Topic: Book on Interest Rate Modelling by Leif Andersen and Vladimir Piterbarg
Created On Thu Sep 10, 09 09:53 AM by piterbarg

piterbarg
Senior Member

Posts: 305
Joined: Oct 2002

Thu Sep 10, 09 09:53 AM

User is offline

Hello

Some of you will know that Leif Andersen and I have been working on a book on interest rate modelling (for the last 5 years, in fact). Well, the book is not quite ready -- but we are getting close, at about 1000 pages with 100 or so to go. So in preparation we have launched a website andersen-piterbarg-book.com where you can find more information (a table of contents, a sample chapter, etc) or register for future updates.

Vladimir

PS Thank you Paul for letting me post this

rsneevas
Member

Posts: 137
Joined: Jul 2002

Thu Sep 10, 09 11:45 AM

User is offline

I was sceptical until I saw the contents of this book given the vast literature in this area. The coverage is amazing and quite unique. I will be eagerly waiting for this book.

Will you be covering model risk and risk management of exotic IR products?

Good luck with this monumental work.

-------------------------
Nivas

piterbarg
Senior Member

Posts: 305
Joined: Oct 2002

Thu Sep 10, 09 12:42 PM

User is offline

Thank you Nivas for kind words.

Part IV of the book is called Risk Management and is dedicated to issues around risk management of exotic IR products. The focus is very much on obtaining stable and robust risk sensitivities, but we do spend some time on general issues such as what is the risk management process, how the desk looks at it, practicalities, etc. We do not focus too much on model risk per se but throughout the book we discuss how to choose models appropriate for given products, how to calibrate them well, and how to assess the impact of such choices.

rsneevas
Member

Posts: 137
Joined: Jul 2002

Thu Sep 10, 09 01:06 PM

User is offline

Excellent.

I am interested in the model appropriateness, how relevant the model dynamics in pricing exotic instruments (forward dynamics) and risk measures. It looks like that you will be covering these as well. Also, any additional information about the real world challenges (like consideration by desk) will put things in perspective.

Well, I will be buying this book. One more question (which is quite obvious): When can we expect this book?

-------------------------
Nivas

Edited: Thu Sep 10, 09 at 01:06 PM by rsneevas

piterbarg
Senior Member

Posts: 305
Joined: Oct 2002

Thu Sep 10, 09 03:24 PM

User is offline

We expect to finish writing it before the year end. When it will actually be available will depend on how we decide to publish it. If we self-publish -- an option we are mulling over -- it should be quick, so I would expect it to hit the shelves, so to speak, before the end of 2009 or early 2010. If we go with a traditional publisher, that would take 6 to 12 months, I would guess.

Cuchulainn
Senior Member

Posts: 16621
Joined: Jul 2004

Thu Sep 10, 09 03:37 PM

User is offline

Quote
Originally posted by: piterbarg
We expect to finish writing it before the year end. When it will actually be available will depend on how we decide to publish it. If we self-publish -- an option we are mulling over -- it should be quick, so I would expect it to hit the shelves, so to speak, before the end of 2009 or early 2010. If we go with a traditional publisher, that would take 6 to 12 months, I would guess.

Does end 2009 include copy editing, proof reading and camera-ready activities?

-------------------------
www.datasimfinancial.com

piterbarg
Senior Member

Posts: 305
Joined: Oct 2002

Thu Sep 10, 09 03:44 PM

User is offline

that's the plan. most of it has been happening in parallel anyway. But, of course, as I am sure you know, these things seem to always take a lot longer than you expect :-)

Cuchulainn
Senior Member

Posts: 16621
Joined: Jul 2004

Thu Sep 10, 09 03:51 PM

User is offline

Quote
Originally posted by: piterbarg
that's the plan. most of it has been happening in parallel anyway. But, of course, as I am sure you know, these things seem to always take a lot longer than you expect :-)

Indeed,

-------------------------
www.datasimfinancial.com

quantmeh
Senior Member

Posts: 4004
Joined: Apr 2007

Thu Sep 10, 09 07:42 PM

User is offline

nice.

will there be downloadable sample codes?

any specific coverage of mortgage loans in FI section?

piterbarg
Senior Member

Posts: 305
Joined: Oct 2002

Fri Sep 11, 09 09:25 AM

User is offline

no, there are no plans to have software code or cover mortgage markets

Cuchulainn
Senior Member

Posts: 16621
Joined: Jul 2004

Fri Sep 11, 09 10:08 AM

User is offline

Will the authors discuss 3 factor models (e.g. FX swaptions) using PDE/FDM with all whistles and bells, i.e. boundary conditions and truncation, extreme values of rho, correlation etc. It is at that stage becomes important because readers can test the algos. Plan B is to describe the algos in excrutiating detail so that you can program it yourself.

my 2 cents.

-------------------------
www.datasimfinancial.com

Edited: Fri Sep 11, 09 at 10:12 AM by Cuchulainn

piterbarg
Senior Member

Posts: 305
Joined: Oct 2002

Fri Sep 11, 09 10:32 AM

User is offline

we ignore FX (focus on interest rate models) so no PRDs but we do go into an excruciating level of detail and all bells and whistles on the models we consider with an intention that people, having read the book, can write up production quality implementation. this is for both PDE and MC models (and "closed form" too, for single- and multi-rate vanilla derivatives)

Cuchulainn
Senior Member

Posts: 16621
Joined: Jul 2004

Fri Sep 11, 09 10:35 AM

User is offline

Sounds very good.

-------------------------
www.datasimfinancial.com

quantmeh
Senior Member

Posts: 4004
Joined: Apr 2007

Fri Sep 11, 09 02:57 PM

User is offline

Quote
Originally posted by: piterbarg
no, there are no plans to have software code

that's too bad. this would have been a great differentiator from other books on IR modeling.

piterbarg
Senior Member

Posts: 305
Joined: Oct 2002

Sat Sep 12, 09 01:12 PM

User is offline

I don't think you learn by reading other people's code -- you learn by writing your own

Cuchulainn
Senior Member

Posts: 16621
Joined: Jul 2004

Sat Sep 12, 09 07:17 PM

User is offline

Quote
Originally posted by: piterbarg
I don't think you learn by reading other people's code -- you learn by writing your own

Vladimir,
I think you do learn from reading code (and especially if it has been designed with some kind of UML blueprints), even if it has not been documented it may contain design nuggets that you can apply in new situations. I compare it to the building industry and architecture - you create CAD drawings which describe the product you have made or are about to make. BTW I learned C++ and design patterns - not from Gamma's GOF book (it had not been published, 1992) - but from having worked on a C++ risk system for equities. Unfortunately, documentation was Spartan ==> pattern mining.

The second issue - you learn by doing - is certainly true but this is not the main issue here, I think. I suspect Javabean (and possibly others) would like to run their models with their own parameters. Of course, you can write all the code yourself but 1) it takes longer, 2) you may not have the time or expertise to write it and 3) why write it if someone else has done it as part of the service. In particular, university students can benefit enormously from such a service.

In general, paper-based algorithms are the second-last step in the process before one maps them to code. Then we see how they really work and how robust they are. The algos are almost good enough, almost. but the proof is when the numbers come out. I have found that the code is a great judge of how good an algorithm really is. Algos nearly always work on paper, but in the end it is run in C++. We are not sure if the algo is correct until it has been tested on a computer.

I am sure your book will be very succcessful; seconding JB, providing source code (or even pseudo code like Clewlow and Strickland approach which is very good imo) would make it truly unique.

-------------------------
www.datasimfinancial.com

Edited: Sat Sep 12, 09 at 07:48 PM by Cuchulainn

spursfan
Senior Member

Posts: 725
Joined: Oct 2001

Sun Sep 13, 09 06:21 PM

User is offline

Good try, Daniel - but if the book has code, they might be justified in adding a zero or two to the price

Cuchulainn
Senior Member

Posts: 16621
Joined: Jul 2004

Sun Sep 13, 09 06:43 PM

User is offline

Quote
Originally posted by: spursfan
Good try, Daniel - but if the book has code, they might be justified in adding a zero or two to the price

I think this is what the market is looking for. Even '101' sanitized examples. But no code means that everyone will code the models from scratch. And new discussions on Wilmott. Given that all agree that quants spend [60,80]% of their time progamming/developing/coding it seemed to me a good idea to give them code to play with.

-------------------------
www.datasimfinancial.com

Edited: Sun Sep 13, 09 at 07:02 PM by Cuchulainn

mj
Senior Member

Posts: 3031
Joined: Dec 2001

Mon Sep 14, 09 12:49 AM

User is offline

writing good code would make their book more popular, no doubt. It would also delay the launch for about another 3 years...

-------------------------
Quant Job Interview Questions and Answers now available on lulu.com and amazon.com

quantmeh
Senior Member

Posts: 4004
Joined: Apr 2007

Mon Sep 14, 09 02:04 AM

User is offline

Quote
Originally posted by: mj
writing good code would make their book more popular, no doubt. It would also delay the launch for about another 3 years...

dont they have their stuff implemented already?! dont they run some kind of code to test what they write about?

Cuchulainn explained it perfectly. if it's not the code, then it could be pseudo code. for instance, Knuth's books come with machine code, it cant be run anywhere(*), but it's so useful.

FORUMS > Book And Research Paper Forum

Forum Navigation:

© All material, including contents and design, copyright Wilmott Electronic Media Limited - FuseTalk 4.01 © 1999-2010 FuseTalk Inc.