Exotic options portfolio

[spiff]

hi there..



i want to know if traders carry out the greeks risk management of the options portfolio on a class basis or u jus run a book including of all different options and then just manage the greeks all together?



thanks

[FDAXHunter]

You definitely want to combine them else you won't be able to hedge efficiently.



I should add that while you want to run a book with everything in it, this isn't necessarily what you end up doing. There are various reasons why you might to remove single deals or an entire class of deals from the book for hedging purposes: You're speculating (i.e. "punting" or "back booking"), your system cannot capture the nuances of some instrument and you'd rather deal with it seperately, etc.



Having said that, you will nevertheless always have to be aware how a single position or a class of positions affects the risks of the entire book (i.e. "This skew risk here comes from these barrier here and here, mostly").



Granchio is the man here as far as exotics go. If you're lucky, he'll drop in later.

[spiff]

thanks a lot fdax hunter..



i wud love it if granchio also drops in.. my doubt was mainly abt pin risks which u hav rightly pointed out.. but i also want to know hedging on lets say u hv an asian and a basket option (so there is no pin risk, rite?) and then i also have an option with a pin risk (barrier/digital/ one touch whtever) but the tenor are all different (as it wud happen in a portfolio) how do i get a picture on where my vega lies?? more so how do i hedge it now?



where i am also coming from is the math behind this computation...



thirdly, this was for an options portfolio, does a similar exercise also work for an exotic swaps portfolio?



neways, thanks a again fdax :-)

[granchio]

here I am, but not too much to add to what FDAX already said, just bits and pieces in  stream-of-consciousness mode. 



- many traders would want to have their book divided in subsets, which they will probably hedge separately. For example: a subset which you want to dynamically rehedge as little as possible (i.e. the trades which you have statically hedged to your satisfaction). Here you will have just an occasional cursory glance to make sure that things are still OK. Another subset with the punts. Other subsets divided according to the kind of risks and/or the hedging methods appropriate: for example one with no gamma nor convexity, but just delta tracking error, another one where you aggregate trades with stock specific risks, etc. You will probably also have a few special books with special exotics, maybe because they are so big to need special attention, maybe because you don't understand them yet, maybe because they require specific numerical attention, you name it. Finally the big main book with the rest of the undecomposable risks. You will probably cross trades between this book and the simpler ones. Main advantage of this approach is to simplify the understanding of your daily PL moves. Against it: the more you divide, the less you realize the "portfolio" effects of exotics hedging one another. Hence you will look for a compromise somewhere.



- greeks according to option classes: if you have a proper risk management system, it will allow you to sort/ filter you positions using various criteria, including type of payoff. This is certainly useful when studying your book



-your example: one asian, one basket, one barrier (by the way I don't like using "pin" risk to describe gap/digital risk... I prefer to say "pin" to refer to vanillas expiring close to their strike). How do you analyze it, for example to find your vega exposure in (k,T) space? it all depends on how good your risk management sys is. If it is no good, you will do it in a spreadsheet. You have to perturb the volsurfaces somehow: the easier way of course is to perturb the parameters used to parametrize the surfaces. Then you have your exposure to the parameters (which probably will be either Heston-like parameters +..., or some kind of euristic params like termstructures of atmF vols, skew, curvature, etc), and you can convert that approximately into a set of vanilla options with similar exposure. If your volsurfaces use parametrizations which are too complex to perturb usefully (e.g. unrelated set of params at each expiry), or different parametrizations for different underlyings, then a substitute approach is to fit them with a simple model (e.g. Heston), recompute the value of the portfolio, and perturb. This of course has many probs, including the initial fitting, which we will leave unmentioned.  Finally there are ways of perturbing directly the numerical surface, let's just say you have to make sure that you do it in ways which make sense (e.g. do not perturb to arbitrageable surfaces- though that's a bit of a laugh, given that often many of the starting surfaces are arbable, in theory at least).  Finally: a decent trader will know the book as a whole, with a rough mental picture of where the various risks are. Very very rough of course, for the details you need the numerical tools.



hope it helps... either too many words, or too few.

[DerFund]

great explanation granchio.



DF

[spiff]

hi granchio..



thanks so much for ur reply, it was amazing one.. it sure does solve my query, but ur points have raised fresh ones Expressionless  



i had some understanding of options, however when it comes to IR im almost blank.. in your experience wht happens in case of a portfolio of callables/ range accruals and CMS spreads.. here u hedge the whole portfolio together (wit restrictions), or its product-wise portfolio and rather have different traders/books itself for IR exotics..??

[granchio]

i am an eqderiv guy, i also blank at fixed income risks :)



IMHO it is just another risk: same comments apply. basically depending on how big it is you will dedicate more or less attention to its various forms. the most interesting is the exposure to the correlation between fixed income and equity. I do not know much more than euristics and stresses to study that, I'd welcome comments by others

[spiff]

cool.. ok..thanks again mate..



while we are on it.. i wanted to ask u, wht sort of reading wud u recommend for exotics trading as well as understanding of the risks.. note, im not from a math bacground but wud want to build my faculties there .. hv traded vanilla options for a brief period.. also worked on options risk management for sumtime.. (all fx) aim to trade exotics eventually.. any help wud be great



thanks again..

[granchio]

the best is a degree in physics, with special care to fluid dynamics

[Strange]

** in case of a portfolio of callables/ range accruals and CMS spreads **



That falls into my little world. Personally I like to separate the products into two wide groups, ones that have a lot of vega exposure (i.e. zero-cpn callbles, any compound options such as callble IFRN or range accruals) and less-sensitive i.e. tarns and snowballs etc. This is mainly because models for the second ones tend to smear vega all over the swaption surface and make it difficult to understand the vega bucketing. Also, every once in a while it's worth empirically analysing the performance of each trade type, this way you can see if the model greeks are actually similar to real greeks.   For instance, I have an Euribor CRA position that tends to trade very short, while the greeks keep showing up as long.  But after all, after all adjustments are done, it's better to have global hedges for a "book effect".



ps. Last thing worth remembering is that "A spanked bottom is the most sensetive model of all". If your model shows you long but you keep getting fisted in a rally, you better buy some.