In pricing an hybrid swap I derived a formula wich contains a covariance term between two FX fwd rates. I was comparing correlation obtained with historical data with the one implied from FX option prices when I suddenly realize that I was estimating correlations between spot rates and not between forward ones. Any idea on how to get the later? Must I set up a multicurrency model?I guess there is also a difference between the instantaneous correlation and the terminal one (the one on the payoff date). Any flash suggestion?It would be kind from you.

Think of Rhonna..Have a look at USDTRY Spot and Correls between Spot and Correl between USD Rates and TRY Rates.....Think also of Spot Implied vols...you should get what you're looking for..forget the math for the moment....think from an intuitive angle...

Just a flash suggestion...not an expert on this. If the two curves are liquid enough, why dont you just create the fwds from the discount factors in each currency. Then do a correlation estimate based on that.

Thank you all. I think I got it

Hi. It depends on what you need. For a market price you have to calibrate over market data in the risk neutral world. So you can use a model like the libor one to calibrate over the swaptions your instantaneous correlations. For a raw price you can calculate, from historical data, fwd and then correlations for each node. You'll obtain a term structure for correlation. Use last 1y of data to approximate the actual expections in the market. BybyFabio

ciao Fabio,I thank you very much but I was concerned about FX forward rates, not Libor.But My request had no sense. Since the ZC wich enter the Fwd rate calculation have Zero volatility on maturity, they have no effect on the terminal correlation. Furthermore, if I assume a BS-like dynamic for my FX rates as usual, the difference between instantaneous and terminal correlation also vanishes. Someone correct me if I'm wrong..I do seriously need holidays..

Yep sorry Stochastic44 I didn't read carefully. For structured in FX we're implementing a model based on "Consistent pricing and hedging of an FX options book", Mercurio: calibration over the smile with correlated dinamic for the domestic and foreign instantaneous forward rates. Dunno if traders price with a shortcut.Ciao

A delayed reply but few quick thoughts:Since FX options are options on "FX" Forward Rates, Implied Correlation from FX Options prices should be correlations between "FX Fwds Rates" (and not spot) as priced in by the FX Options market. E.g "Implied" Correlation between Forward Rate of EUR/USD and of USD/JPY can be implied from EUR/JPY traded options Implied Vols or Implied Correlation between USD/JPY and USD/TRY 1y-forward rates from 1y TRY/JPY traded options prices. This can give you forward looking term structure of Implied Correlations.You can also obtain backward looking term structure of historical correlations by looking at say 1y USD/TRY Forward Rate (Spot + 1y Fwd Premium) and 1y USD/JPY Forward Rate.All the required variables mentioned above can be obtained easily from any standard market data feed (such Reuters/Bloomberg).Stochastic44, honestly, I did not understand your reference to instantaneous correlation and terminal correlation. A correlation would always have an observation period say 1month. So probably by instantaneous and terminal correlations you mean 1m (say) Historical Correlation at t=0, and same 1m Historical Correlation at t=T (at maturity) and obviously these two would be different as they have different observation windows of length 1 month.Cheers.