Does anyone have experience about Heston model's calibration?

raywin

New Member
Hello Every

Does anyone have experience about Heston model's calibration? I just followed nimalin moodley 's paper. But I can not get the same results as page 29. Did anyone do that before? Thanks a lot!


Ray
 

Bastian Gross

German Mathquant
Hello Ray,

a friend of mine has compute a Heston-Calibration-program in C++.
But I do not know his experience with the nimalin moodley 's paper.

I ask him for further information.


But in the meantime here is a paper about Heston Calibration and a Matlab-Code:
 

Attachments

raywin

New Member
Yes, I try that code. But I think that is not really correct. The main idea is correct but there are some problems. But I can not get the result in page 29 of that paper.
 

raywin

New Member
Thanks. I saw this code before in some web. Currently I just want to use mablab build-in function lsqnonlin to do calibration.
 

raywin

New Member
The idea is not really hard but there are some trivial work to do. I am researching that. Hope crack it asap!
 

mixumus

New Member
Dear all,
I wonder whether it is logical to look at the term structure of the parameters in Heston model.
I mean do you think it is logical to calibrate Heston model to a spesific maturity on a spesific day(and doing this seperately for every maturity) and looking at evolution of the term structure of parameters by following the same procedure.
Do you have any experience about this?
Your help will be very appreciated.
 

shinichi

New Member
Dear all,
I wonder whether it is logical to look at the term structure of the parameters in Heston model.
I mean do you think it is logical to calibrate Heston model to a spesific maturity on a spesific day(and doing this seperately for every maturity) and looking at evolution of the term structure of parameters by following the same procedure.
Do you have any experience about this?
Your help will be very appreciated.
Not sure if I understand well your question. You could use Heston with piecewise constant parameterization to calibrate a given smile surface.
 

shinichi

New Member
The idea is not really hard but there are some trivial work to do. I am researching that. Hope crack it asap!
what's your problem exactly? I haven't seen any result on page 29, except a graph of error calibration with no numerical output.

Calibration of Heston is quite tricky though. It's not as easy as in Moodley's. You should use Gauss-Lombarto and try to use Lipton-Lewis to save time of computation as you'll have to compute only 1 integral instead of 2.

Hope it helps.
 

MagicWeasel

New Member
You should use a global optimization technique (e.g. SA) to calibrate the model since you could end up in a local minima when using nonlinear least squares method. The optimization problem is not convex.
 

Bastian Gross

German Mathquant
The Levenberg-Marquardt Method should also be recommendable.
 

EricFleming

New Member
post the question in nuclear phynance
 

WIL Recruiter

New Member
You should use a global optimization technique (e.g. SA) to calibrate the model since you could end up in a local minima when using nonlinear least squares method. The optimization problem is not convex.
using fminsearch in Matlab will do the trick

I got good calibration results not only for Heston, but also for the more advanced Heston-Nandi (2000) model
 

Ahmed-amager

New Member
hey every one
.I am doing my master thesis in volatility forecating but I have a problem. I want to calibrate heston model as discribed in the following articles but I really cant find any thing (any code or help). I am trying hard but I am not good at programming so its tough. Maybe u or any body else on this site help me with the co de ? thanks in advance:)



http://www.wilmott.com/pdfs/040502_shu.pdf

http://www.google.com/url?sa=t&rct=j&q=pricing s&p 500 index options with heston’s model&source=web&cd=2&ved=0CC4QFjAB&url=http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.201.590&rep=rep1&type=pdf&ei=KoOiT8WkB82k4AStnJWuCQ&usg=AFQjCNE2oGOjaOBlfPX9VQJLTt8tIqVFew&cad=rja
 
Top