Hi to everybody,
I wanted to know if anyone had read and went through the paper by Zhang and Xian "The Implied volatility smirk" ( http://www.ccfr.org.cn/cicf2005/paper/20050126110345.PDF ). I find it an interesting concepts which in one way relates to PCA type of studies but also to the more simplified idea of Samuel (2002) to which Rebonato refers in "Volatility and Correlation 2nd edition".
Did anybody try to do a similar analysis and has some readily available code? I would be interested in comparing this method with the one brought up in Goncalves and Guidolin (http://ideas.repec.org/p/fip/fedlwp/2005-010.html) and look at the RMSE.
Moreover, I wanted to ask if anybody had an idea on which is the best method to test when we can define that a model outperforms the other. Is RMSE enough?
Are there any real applicable tests when it comes to the dynamics of the implied volatility surface? If I am trying for example to build a model which predicts the dynamics of the surface, GIVEN A CHANGE IN THE UNDERLYING - ie look at the point of view of eg Derman in "Regimes of Volatility" - how can I compare effectively the above mentioned models with eg sticky strike and sticky delta?