tag:blogger.com,1999:blog-14415229.post1206755836976362184..comments2023-10-10T16:55:02.139+02:00Comments on Chase The Devil: Another SVI Initial GuessFabienhttp://www.blogger.com/profile/07288327695801480778noreply@blogger.comBlogger1125tag:blogger.com,1999:blog-14415229.post-44629360318199954092014-08-06T18:53:16.563+02:002014-08-06T18:53:16.563+02:00I'm trying to implement the quasi-explicit cal...I'm trying to implement the quasi-explicit calibration of the Gatheral's SVI inspired model as laid out in the Zeliade wp. This is the algorithm I follow:<br><br>1. Select some initial parameters values for (a, b, rho, m, sigma)<br>2. For the initial m and s, do a constrained optimization - P(m, sigma)<br>3. Finally, using (a*, b*, rho*) from step (2), use simplex (Nelder-Mead) to estimate m* and sigma* by minimizing P.<br><br>Is this enough to calibrate the model? More specifically, do I need to run multiple iterations of the algorithm or the 3 steps sufficient? I'm a bit confused because in step (2), the solution (a*, b*, rho*) depend on the initially chosen (m*, sigma*) and so different choices of initial values of these should yield different solutions for (a*, b*, rho*). <br><br>ImranAnonymousnoreply@blogger.com