Page 1 of 1

A Correlation swap

Posted: Sun Jun 13, 2004 8:32 pm
by Nonius
Ok, there have been many threads on this elsewhere, but, I want to get to the bottom of it.  Let's fix equities.  I have my own ideas, but I wanna hear your thoughts.  In your collective experiences, how do people price/risk manage this?  I mean, it seems like a punt, no?  Is this just like insurance?  figure out a tail, and demand a return on that tail?



the one thing correlation as a structure has going for it is that it can't go to the moon, nor can it go to hell.  As a stochastic process, it actually lives in a compact set (then again, same is true for discount bonds).  from the other Nonian ideas, this reduces to modeling brownian motion on the cartesian product of a well known curved space and euclidean space.  at least, this is the way I would think about it.

A Correlation swap

Posted: Sun Jun 20, 2004 1:17 pm
by granchio
but is it a brownian motion? Smiley



anyway I am also interested in this correlation swap thing. I assume you are talking about the one on implied correlation, not just the option on basket vs basket of options (some brokers call that a corrswap).



I do not know how other people price and model it, and I have not really made up my mind yet.  I am tempted not to model it at all, and just mark it to market, like a forward contract.



Let me blurb out a few things on corr that might be relevant to the discussion. If we look at the correlation market in equities, at least as traded in the vanilla baskets, at the moment it seems that pairwise correlation is trading basically without any skew nor termstructure. this of course would point to a very simple process.  but is this borne out by the realized correlation?



this is relevant esp. if you  have a large exposure to long term correlation, marked flat, so the realized PL will depend on the difference.



the other issue which IMHO is worth discussing is the relation between volatilty and correlation (as well as spot and correlation). various people, including myself, used to make the conservative assumption that these 2 were positively correlated (vol up, corr up), using basically the same argument used for the negative skew: in equity, when we go down, we go down together, and fast... Note also that the skew on index volsurface is larger than on the singlestocks ones, again pointing to a corr up-vol-up scenario.



some players stopped even thinking about pairwise correlation and went for other stats, which incorporated the corr up-vol up intrinsically. other stuck to the pairwise correlation framework, but run long vega against short corr to compensate (or try to).



this scenario worked quite well, e.g. over the 97 98 crisis... but has broken down, certainly over the last year and a half. implied correlation was basically steady over last year (including the market drop of first quarter), and is creeping up, with volatilities going steadily down.



there are structural reasons for this, mainly the of crossderivatives of the exotic books: how their correlation exposure change with spot and vol moves (have we got names for this greeeks? Smiley )



to go back to the beginning: IMHO the equity correlation swap was introduced as a hedging instrument, to be marketd to punters (you're short zillions of corr, you can pay above historical, so you hope somebody will sell it). I do not know how to model it, but it is a very topical question in this market. looking to hearing more on this.

A Correlation swap

Posted: Sun Jun 20, 2004 1:56 pm
by NeroTulip
to go back to the beginning: IMHO the equity correlation swap was introduced as a hedging instrument, to be marketd to punters (you're short zillions of corr, you can pay above historical, so you hope somebody will sell it). I do not know how to model it, but it is a very topical question in this market. looking to hearing more on this.



This is my understanding as well. But does it mean it's a good deal for the punters?

A Correlation swap

Posted: Tue Jun 22, 2004 9:33 am
by granchio
good for the punters? maybe on a case by case basis.



question: I am thinking of forward starting correlation swaps.



in the system here right now they would show zero correlation exposure... I do not like that. comments?

A Correlation swap

Posted: Mon Jun 28, 2004 11:48 pm
by granchio
nothing more on correlation? swaps or otherwise? come on!

A Correlation swap

Posted: Tue Jun 29, 2004 7:10 am
by Nonius
I was busy fighting fires, trolls, and in-house political enemies.  Let's price one up?  Give me an example.  and then, let's see how we would manage it.  I am curious.