SERVING THE QUANTITATIVE FINANCE COMMUNITY

 
User avatar
rdiour01
Topic Author
Posts: 9
Joined: January 31st, 2011, 4:09 pm

Heston Initial Parameters

September 9th, 2011, 3:17 pm

Does anyone know how to go about "guessing" the initial parameters for calibrating the Heston model?Are there "correct" guesses, or do you just punch in any value?Thanks in advance!
 
User avatar
frolloos
Posts: 1621
Joined: September 27th, 2007, 5:29 pm
Location: Netherlands

Heston Initial Parameters

September 9th, 2011, 4:33 pm

have you tried searching the forum for 'heston calibration'? just a thought.
Last edited by frolloos on September 8th, 2011, 10:00 pm, edited 1 time in total.
 
User avatar
frenchX
Posts: 5911
Joined: March 29th, 2010, 6:54 pm

Heston Initial Parameters

September 9th, 2011, 6:07 pm

pierre gauthier wrote a paper which exactly answers your question.Fitting the Smile, Smart Parameters for SABR and Heston
Last edited by frenchX on September 8th, 2011, 10:00 pm, edited 1 time in total.
 
User avatar
Peniel
Posts: 59
Joined: April 8th, 2006, 9:46 am

Heston Initial Parameters

April 4th, 2012, 4:07 pm

Regarding the Gauthier/Rivaille paper, they fix kappa "at a typical market level" (kappa=2 in the article).The value of kappa is important as it is used to imply the others parameters.Does anyone have an idea about how to imply this market level?
 
User avatar
Alan
Posts: 10386
Joined: December 19th, 2001, 4:01 am
Location: California
Contact:

Heston Initial Parameters

April 4th, 2012, 11:15 pm

It's a relatively unstable parameter -- don't sweat it -- just start the solver at 2 as suggested.
 
User avatar
Peniel
Posts: 59
Joined: April 8th, 2006, 9:46 am

Heston Initial Parameters

April 5th, 2012, 9:21 am

Thanks Alan for your answer.My concern is that if I choose a "wrong" kappa, I'll get a set parameters for a local minimum of the problem.Then I would apply Levenberg-Marquardt using this local minimum as initial values and the algorithm would stick to it.I would be interested in knowing what people do in real life to get stable parameters.* Do they just know it because they actually calibrate the model (meaning that as they calibrate, say, everyday and they know where to expect each parameter for a specific underlying)* Do they apply a specific procedure (like above)? or just global then local calibration?
ABOUT WILMOTT

PW by JB

Wilmott.com has been "Serving the Quantitative Finance Community" since 2001. Continued...


Twitter LinkedIn Instagram

JOBS BOARD

JOBS BOARD

Looking for a quant job, risk, algo trading,...? Browse jobs here...


GZIP: On