Forums  > Pricing & Modelling  > Asian options on WTI Crude Oil  
     
Page 1 of 1
Display using:  

hilss


Total Posts: 55
Joined: Jun 2007
 
Posted: 2017-04-04 20:17
Hello everyone,

I have a customer who is a market-maker on CL options. He would like to start trading Asian options. As I'm sure you know, the averaging period for CL options includes 2 futures. Example, if we want to price the M Asian Option (expiring June 30), we will depend on the N and Q futures. My question is: is there a way to derive the variance during the averaging period using the variance of the N and Q futures that is implied from the standard (american and european) options on CL?

I did look around on this forum and found some posts. Some stuff didn't address my question, and some stuff I don't understand, I'm afraid.

Thanks,
hilss

Patrik
Founding Member

Total Posts: 1333
Joined: Mar 2004
 
Posted: 2017-04-04 22:37
Yes - and no Smiley

What I mean by that is it can help you imply something if you are willing to specify a more general vol model and then calibrate this model using a combination of observable bullet options and asian options.

Depending on tenors etc what is a good model can vary - but if it's for market making wti Asians you'll probably do ok with a 2factor model.

Capital Structure Demolition LLC Radiation

hilss


Total Posts: 55
Joined: Jun 2007
 
Posted: 2017-04-05 01:00
Thanks Patrik.

So for the example I used above (N being the july future and Q being the august future) can I assume the following:

dM = a * dN + (1- a) * dQ

dN/N = sigmaN * dwN
dQ/Q = sigmaQ * dwQ
dwN * dwQ = correlation * dt

Once we have N and Q fit (for the regular options), can the M asian option be fit by modifying only the correlation? Is this only valid for ATM? what about other strikes?

Or did you have a different two-factor model in mind... Sorry if this is not what you meant..
hilss



Patrik
Founding Member

Total Posts: 1333
Joined: Mar 2004
 
Posted: 2017-04-05 10:15
That line of thought would be a very specific type of model and there wouldn't be that much value as you'd add a lot of parameters for all the pairs and so forth. By that I mean you'd have one for M, one for N etc and all would have parameters - so you kind of end up marking a bunch of correlations and you haven't reduced the degrees of freedom much.

What tends to be done is calibrating a more general term structure model that's able to price a much bigger class of derivatives without too many parameters and hence have some consistency of looking at a book of say bullets, asians and swaptions. With regards to smile/skew you tend to see a pragmatic approach of marking some points on the smile and using interpolation - especially if the use case doesn't have strong dependencies on smile dynamics. For more complex derivatives introducing e.g. stochastic vol models that can reflect a smile can make sense.

Capital Structure Demolition LLC Radiation

TakeItAndRun


Total Posts: 89
Joined: Apr 2010
 
Posted: 2017-04-06 18:16
@hilss I think the definition of the payoff of the Asian option and the underlying is needed to correctly answer your question.

If this is a regular cash settlement European average price option on WTI spot, what is needed is a method to interpolate/extrapolate volatilities during the averaging period from futures options.
In this case, there is no need of an underlying consisting of two correlated assets.

hilss


Total Posts: 55
Joined: Jun 2007
 
Posted: 2017-04-07 11:13
@Patrik, so the 2-factor model you're referring to is strictly for the ATM vol term structure?

@TakeItAndRun, this is the "regular" European cash settlement option on WTI, but it's on the front month future. Its symbol is AO on the cme

hilss

Patrik
Founding Member

Total Posts: 1333
Joined: Mar 2004
 
Posted: 2017-04-07 23:23
In practice, when using a simple 2f model that doesn't produce a smile within model so to say, you tend to calibrate the model for discrete slices, e.g. 10,15,25,35,50 delta, while keeping unobservable parameters the same across the smile. Obviously trying to mark those unobservables so that the APOs, swaptions etc you can observe fit as well as possible. Then interpolate for the case at hand.

Capital Structure Demolition LLC Radiation

hilss


Total Posts: 55
Joined: Jun 2007
 
Posted: 2017-04-07 23:27
Thank you very much Patrik...
Previous Thread :: Next Thread 
Page 1 of 1