FX correlation swap

newbanker · February 4th, 2010, 8:08 am

Most banks compute correlation between two currency pairs in the mathematical sense. Goldman, however,have a version they call "drift-free", whereby one computes the correlation as if the logarithmic ratesof change of both currency pairs have average = 0. Naturally there are differences between the standardcorrelation and the "drift-free" one.Now why would Goldman do that?

islington · February 4th, 2010, 9:17 am

I have never been a big fan of centered estimators for vols or correls tradingwise. It is not linked to any hedging strategy wheareas the uncentered one are. If a stock is up 1% everyday on a given period, the centered vol estimator for daily vol will be 0, but I can tell you you felt the pain if it went through a strike you're short.Plus it's so much simpler to write the termsheet.

pb273 · February 4th, 2010, 2:29 pm

QuoteOriginally posted by: newbankerMost banks compute correlation between two currency pairs in the mathematical sense. Goldman, however,have a version they call "drift-free", whereby one computes the correlation as if the logarithmic ratesof change of both currency pairs have average = 0. Naturally there are differences between the standardcorrelation and the "drift-free" one.Now why would Goldman do that?often either because the hedgies will misprice or it is easier to replicate

rmeenaks · February 4th, 2010, 5:24 pm

Can you post the termsheet here or the actual formula for the computation? I want to take a look. Thanks,Ram