I think the underlying idea is that the future ATM IV is a substitute for future volatility realized. But the ATM IV, spot or future, is not a good proxy for expected volatility, if there is a significant correlation between the underlying and volatility. FVA has nothing to do with Volswaps. This is Forward Volatility Agreement and you enter into a purchase/sale of a vanilla launch option in advance with black scholes settings (except spot price) that were set today. As I understand it, an FVA is a swap on the volatility of under-induced money in the future, which is ensured by a forward startup/straddle option. In a very current (fairly condensed) discussion paper, I saw that Rolloos also deducted a price approximation without a model for forward start-up flights-swaps: in terms of sensitivity, it looks like flight/var swaps starting forward, because you don`t have gamma at the moment and you`re exposed in front of flight. However, it is different that you are exposed to standard vega deformations of the vanilla and MTM options because of the tilt, as the spot moves away from the original trading date. Especially as far as FX is concerned, but I think it`s a general question. any good reference would be appreciated. FVA is not mentioned in Derman`s paper (“More than you ever wanted to know about volatility swaps”) A pre-volatility start swap is really a future volatility swap. In another thread, I wrote that Rolloos -Arslan wrote an interesting document on the approximation of prices without a Spot-Start-Volswap model.
This is used to increase exposure to implied forward volatility and is generally similar to trading with a longer option and cutting your gamma exposure with another option with expiration equal to the start date in advance, constantly balanced, so that you are flat gamma. Mathematics in this last document seems right – but I haven`t yet seen any numerical tests of the result without a model. Who tested the latest Rolloos result, comments/ideas on it?.