can you elaborate on the copula mathod?i dont think a gaussian copula would work very well, as it would probably underestimate the butterlfy and misprice the skew? or maybe not?is there any article on that? can anybody provide references?

looks finebut i actually think the eurusd/usdjpy (the one you would look at in practice) is rather stableif you get the correl wrong by 5%, you should still be inside the bid offer spreadalso, the correl surface certainly varies but probably not ramdomly, i mean there must be some patterns that you can model, eg- if vol of either usdjpy or eurusd goes up, the correl probably goes up- if theres a big move down in usdjpy spot, id expect correl to go down etc...

I´m trying to price some FX options in Emerging market currencies against EUR. FX options against USD are relatively well developed so I´m going to use that correlation rule to price these EM/EUR FX options. EUR FX options aren´t liquid enough as USD. And I´ll price over many strikes and maturities. Which correlation methodology to be applied is a tricky question. a. I prefer not to use multivariate GARCH models in order to forecast correlation T periods ahead (option maturity) because the possibility of having negative-defined matrix.b. Indeed, one could argue that an EWMA model could work (and I prefer this model but...) but correlation estimates are constant no matter how far you´re predicting.c. Calculate the correlation using forward and this measure might have time-to-maturity features embeeded on it. I have these three paths to follow. Anyone has more suggestions or any approach which is the leading in industry?Thank you very muchBob