HiWhen pricing a bespoke CDO, it is probably a market standard pracitce to calibrate base correlations from fhe index tranche quotes and map the base correlation curve to the bespoke portfolio(EL mapping or some other ways).However, what's the usual way of assuming correlation for FTDs?I've seen some bid/offer data of FTDs and correlation levels were around 35%-45%for diversified baskets.Could someone tell what kind of logic is behind this?

As far as I know, the correlation is calcuated based on the correlation of asset values or that of equity prices. Anybody can give any other hints?

If the FTD or NTD correlation were calculated based on the correlation of asset values or equity prices, there would be no tools for hedging the correlation risk.Does anyone think the FTD or NTD correlations are calculated by making use of the base correlation curve?

If you've got an FTD basket of Turkey, Brazil and Gazprom, what do you think is the relevence of an index base correlation?

I agree w/ Wibble.Consider a lot of people are note confortable 1/ with the BC methodology and 2/ with the BC mapping from index to bespoke.If you're thinking of applying the BC (underlying ptf of 125) with a basket (usually max 25 names), I can sell you protection --------------------slym

Apart from having a market for standardized index tranches (CDX or iTraxx) there is also a market for standardized FTDs divided by sectors (financial, industrial, energy, hivol, diversified etc).This market provides an implied correlation number to use either for highly concentrated or diversified bespoke FTDs.Estimating def correlation with equity prices is a bad proxy.

From my experience the cdo and ftd markets are decoupled as already remarked. Base correlation doesn't seem to play any role. We calibrate to sector and inter-sector baskets on the standard CDX and iTraxx baskets. Similarly there are some (more or less liquid) broker quotes for EM ftds as well (asia, latin america or eastern europe). Some inter-region EM quote should be sufficient to back out an implied correlation number to price your basket. It seems that the market indeed uses a Gaussian copula for that since we find that the term structure of (implied Gauss) correlations is rather flat. So calibration to a 5y quote suffices to also price 3y and 7y (or whatever). To get confidence to your mark-to-market you might consider Totem submissions.

Let's say there is a standard FTD basket for the financial sector(sen) as indicated below and sum of spreads of the FTD is 76%.In this case, the implied correlation is about 60% with Gaussian copula model.Entity / CDS spreadAllianz AG / 9 AXA / 9Muenchener Rueck AG / 8Commerzbank AG / 7Capitalia SpA / 7If some entity in the standard basket is replaced by some other entity not included in the basket initially, How should we adjust the correlation?For example if AXA is replaced by ABN Amro, will it have any effect on the correlation? Or if it is relaced by the Bank of China, will the correlation be lowerbecause the bank operates in other region?

this is when trading becomes an art and not a science. no model will ever tell u what to do. and your corr assumptions will depend also on several other technical factors (axes, liquidity of underlying single names, recovery dispersion, maturity).

A pointer to info on the standardized FTDs would be much appreciated.

There are interdealer brokers who send dailiy runs on standardized FTD but apart from that is not an easy data set to find

So not standardized in the CDX, iTraxx sense then?

Hi Stefanone would it be possible for you to hand me a sample broker run like the one you mentioned? The only things I see now are FtD on EM. I PM you with my email.

I dont have acees to them. Sorry.

It is probably true that dealers and traders take into account the position they already haveand other trading tactics when they quote bespoke FTDs to their clients, not merely thepricing model.However, financial institutions should recognize fair values of those OTC derivatives(such as bespoke FTDs)for accounting purposes. Fair values are estimated using present value or other valuation techniques, using inputs based on market conditions existing at the balance sheet dates. So, I still think that there should be some method of adjusting correlation inputs for bespoke FTDs.