QuantLib › quantlib-users

Re: Advice requested: getting into computationalfinance

Posted by cuchulainn on
URL: http://quantlib.414.s1.nabble.com/Advice-requested-getting-into-computational-finance-tp7422p7433.html

>>Regarding dr. duffy's book, its a good starting point and worth the effort to go through the book. 
As a first introduction to the field in general its reasonable. "Europe in 14-days". A tour de force through the field, but each time it gets interesting one reads "for further details see..." As already mentioned it does not really address the methods in option-pricing.
 
I think you are referrring(?) to the "Introduction to  C++ for financial engineers", which is a basic introduction. I have 2 others books as well, one on more advanced topics in C++ and PDE and a book on PDE/FDM.
 
//
On the Crank Nicolson, here is a thread showing how it not works for the Greeks
 
 
And for the Heston model CN needs to be replaced by Richardson (see Shephard's thesis, it's on my site).
 
regards

Daniel Duffy

From: chrilly [mailto:[hidden email]]
Sent: Fri 10-04-2009 21:47
To: Satish Vangipuram
Cc: Daniel J. Duffy; ksvanhorn; [hidden email]
Subject: Re: [Quantlib-users] Advice requested: getting into computationalfinance

>>There are many people who study quant finance but only a small proportion get jobs as a quant.  This was true even before >>the financial crisis and is  more true in present conditions.
The times are generally not the best for starting a new job/project. But the quant business is probably even worse than most other fields. I know of an university which thinks about a "rescue-plan" for students which have done e.g. for 3 years their finance-studies. To ease the change to other more promising directions. Or at least to change the name of the course. Its not only that the demand for finance has evaporated. The field has also got a very bad reputation (at least in some European countries). Its almost like having a PhD for drug-trafficing.  
 
>>Regarding dr. duffy's book, its a good starting point and worth the effort to go through the book. 
As a first introduction to the field in general its reasonable. "Europe in 14-days". A tour de force through the field, but each time it gets interesting one reads "for further details see..." As already mentioned it does not really address the methods in option-pricing.
 
>>The comments by chrilly are true but can be worked around with the help of forums etc. I would ask this question "IF dr. >>duffy's book is not good, what are the other options". 
The negative answer, which books are even worse, is easier:
Justin London, Modeling Derivatives in C++. One of the worst books I have ever seen.
D.Tavella, C.Randall: Pricing Financial Instruments: The Finite Difference Methods. The are some interesting glimps, but overall the book reads like the authors haven written it on one weekend.
Somewhat better is: L.Clewlow, C.Strickland: Implementing Derivatives Models.
Wilmott has in Vol-III some explanations and code. But the quality is also low. Generally there are zillions of Financial books but the overall quality is rather low. Its was like the market. A lot of hot overpriced air.
Its better to read general classical introductions/books to the field. Although this books are for my taste also too mathematical and too less practical. E.g.
J.W.Thomas: Numerical Partial Differential Equations.
K.W.Morton, D.F.Mayers: Numerical Solution of Partial Differential Equations.
Quantlib has also some classes (E.g. Crank-Nicolson).
 
>>It is an art to put together code from different source to get something working.  Sure there is some hardwork involved in >> but thats the way it is with c++. 
Yes.
 
Chrilly
 
 
----- Original Message -----
From: [hidden email]
To: [hidden email]
Cc: [hidden email] ; [hidden email] ; [hidden email]
Sent: Friday, April 10, 2009 7:52 PM
Subject: Re: [Quantlib-users] Advice requested: getting into computationalfinance

well, well, well.. there are some strong comments here.
however the advice is probably a good one ie to think a few times before jumping in. 

There are many people who study quant finance but only a small proportion get jobs as a quant.  This was true even before the financial crisis and is  more true in present conditions.

Regarding dr. duffy's book, its a good starting point and worth the effort to go through the book.  The comments by chrilly are true but can be worked around with the help of forums etc. I would ask this question "IF dr. duffy's book is not good, what are the other options".  It is an art to put together code from different source to get something working.  Sure there is some hardwork involved in it but thats the way it is with c++. 

Just my thoughts


On Fri, Apr 10, 2009 at 11:38 AM, chrilly <[hidden email]> wrote:
First I would question if all this mathematical stuff is really needed (I have a PhD in Statistics). There is a huge gap between the academic-financial and the real-world. Most models are mathematics pour la mathematics. There are a lot of would-be Einsteins in the field and they want to show how good they are in Ito-Calculus et al.
 
I can not judge the quality of the distance-learning material. I have only read Daniel J.Duffy: Finite Difference Methods in financial engineering. I would give this book 2-Amazon stars. The author has put his dusted PhD-Thesis from the shelve. The thesis is obviously about hyperbolic DEs, the option-problems are parabolic. So he speaks most of the time of another problem.
The presented code is terrible. Some parts are commented by the author with
// This part of code is not really optimal but it works 
Unfortunatly even this is not always true.
 
D.Duffy has written also an article in Wilmotts magazine "A Critique of the Crank-Nicolson Scheme". The scheme has - as pointed out by the author - some numeric problems. But its practically a non-problem. The options-models are all junk. The numerical problems of Crank-Nicoloson are negible in comparison to the model-errors. We are not dealing with dynamic fluid problems.
Furthermore one can solve the instability of Crank-Nicolson in a trivial way. Crank-Nicolson is a 1:1 combination of the Explicit and Implicit-Solution method. By changing the weight from 0.5 to e.g. 0.6 (in direction of the implicit method) the instability problems disappear. Theoretically only Crank-Nicolson has 2nd order convergence, the implicit method (and also a factor of 0.6) has 1st order. But the exact formulation is: One can only proove 1st order, in most problems the convergence is almost identical. Besides this, numerical accuracy is not a real topic in junk-models.
 
Chrilly Donninger
 
 
 
----- Original Message -----
Sent: Friday, April 10, 2009 3:42 PM
Subject: Re: [Quantlib-users] Advice requested: getting into computationalfinance

 
we have a range of distance learning courses that just might address your needs.
 
regards
 
Daniel J. Duffy

From: ksvanhorn [mailto:[hidden email]]
Sent: Fri 10-04-2009 06:22
To: [hidden email]
Subject: [Quantlib-users] Advice requested: getting into computational finance


I'm an experienced software engineer with a B.S. in physics, a Ph.D. in
computer science, and a love of mathematics.  I feel like my brain is going
to atrophy if I don't find some way to use my math skills, and I ran across
a description of computational finance in Wikipedia that sounds like it
could be just what the doctor ordered.  The Wikipedia article states that
"knowledge of the C++ programming language, as well as of the mathematical
subfields of: stochastic calculus, multivariate calculus, linear algebra,
differential equations, probability theory and statistical inference are
often entry level requisites".  I'm a bit rusty on my differential equations
and I lack the background in stochastic calculus, but I'm pretty strong on
everything else.

So how would you all recommend I go about getting into this field?  One
constraint I face is that I can't drop everything and go back to school for
two years; I need to keep on making an income.   But I'm pretty good at
studying things on my own.

-- Kevin S. Van Horn

--
View this message in context: http://www.nabble.com/Advice-requested%3A-getting-into-computational-finance-tp22982909p22982909.html
Sent from the quantlib-users mailing list archive at Nabble.com.


------------------------------------------------------------------------------
This SF.net email is sponsored by:
High Quality Requirements in a Collaborative Environment.
Download a free trial of Rational Requirements Composer Now!
http://p.sf.net/sfu/www-ibm-com
_______________________________________________
QuantLib-users mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/quantlib-users

------------------------------------------------------------------------------
This SF.net email is sponsored by:
High Quality Requirements in a Collaborative Environment.
Download a free trial of Rational Requirements Composer Now!
http://p.sf.net/sfu/www-ibm-com

_______________________________________________
QuantLib-users mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/quantlib-users

------------------------------------------------------------------------------
This SF.net email is sponsored by:
High Quality Requirements in a Collaborative Environment.
Download a free trial of Rational Requirements Composer Now!
http://p.sf.net/sfu/www-ibm-com
_______________________________________________
QuantLib-users mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/quantlib-users



------------------------------------------------------------------------------
This SF.net email is sponsored by:
High Quality Requirements in a Collaborative Environment.
Download a free trial of Rational Requirements Composer Now!
http://p.sf.net/sfu/www-ibm-com
_______________________________________________
QuantLib-users mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/quantlib-users
Free forum by Nabble Edit this page