Vanna is called 'shadow gamma' by NNT p 200 Volga is called 'vega convexity' described p184 p 238 and the following ones.
Yes, right. (Although I don't see 'shadow gamma' referenced on pg 200 - but I do recall his mentioning it elsewhere). What's interesting to me is: either the terms vanna and volga weren't in common use in 1997, or NNT chose to use his own terminology. I don't recall which is the case.
My bad. 'Shadow gamma' is explained p138, and vanna is called 'Ddelta/dvol' p200 . DdeltaDvol is an explicit way to describe that vanna is a partial derivative of an option delta wrt volatility. It's easier to remind what it means.
As a former market maker, I can tell you that I paid little attention to my vanna or any of the vega's derivatives. Here is why, unless you have some ULTRA complex strategy they really shouldn't matter too much. Just looking at charts I think you guys can figure out how your vega is changing. Here is an important vega calc that matters. Weighted Vega, it is a measure, that weights how sensitive you position is to real IV change. Basically it takes into account that the front month moves a lot more than the backs. It is the most info 99% of option world will ever need. It probably uses all those fancy names that remind me of wheel of fortune anyway. http://www.option911.com
It depends on which type of underlying you trade, and what you do with. If you trade skew volatility on currency options, then there is a robust model to price every options based on vanna and volga. That way, it's useful. Every option trader derives weighted vegas. Sometime vegas are weighted by the square root of time, sometime they are based on local volatility... But for sure I agree, it's not necessary for basic strategies.
Help me understand - are you talking about an empirical result (ie, the tendency of the term slope to change in equity ops markets as a result of sharp moves in the underlying) or is this a pure Greeks calculation?
There is no such a thing as a pure greeks calculation that will tell you how volatility would move. The only way is by forecasting a behavior for short term volatilty wrt long term volatility. Then to weight vegas the same way.