Page 2 of 2
SABR with rho less than -1
Posted: February 10th, 2010, 7:51 am
by JojoLeBarjo
It's a simple model mis-specification, nothing to worry about. Try another model...
SABR with rho less than -1
Posted: February 10th, 2010, 12:56 pm
by Soorma
QuoteOriginally posted by: JojoLeBarjoIt's a simple model mis-specification, nothing to worry about. Try another model...Can you please be a little more explicit? Which are the other models that can be used ?Thanks
SABR with rho less than -1
Posted: February 10th, 2010, 1:16 pm
by Soorma
First order approximation for backbone leads to backbone = alpha/F^(1-Beta) so if beta > 1 ATM vol increases with increase in F i would like to know if beta greater than one creates any other major distribution changes .
SABR with rho less than -1
Posted: February 12th, 2010, 12:27 am
by Alan
You have to be careful withdS = sigma S^beta dB(t) with beta > 1. This is a classic example of a strictly local martingale -- not a true martingale: E[S(t)] < S(0)See chap. 9 of my book for discussion.
SABR with rho less than -1
Posted: February 12th, 2010, 7:13 am
by Paul
Going back to the subject of rho<-1...Whenever you calibrate you can expect nonsense to happen. Just be thankful that your speed of mean reversion is positive and your mean vol levels are not imaginary!P