August 27th, 2010, 6:03 am
without having seen the article I would say it's simply a statement about dVega/dSpot and dVega/dVol. the former tends to be highest for ~20% delta RR; at least for vol and skew levels typical in EQ deriv indices. given the latter is traded e.g. via a symmetrical (skinny-) fly centered around ATM, both do somewhat depend on the model, because of how you shape a smile. 7-parameter models will give a slightly different fit to the "true" smile than 9-parameter models etc.can you post a link to the article?