Hi Everyone,Any paticular recommendation on how to price an option on a basket on commodities?I am especially concerned on the volatility aspect (let us forget about the correlation...)Let us say all underlyings of the index are mean reverting and have volatility curves in backwardation.How would that change the pricing of an option on the index?I thought of using a simple Vasicek like model but it looks not easy to estimate parameters on commodity underlyings,Should I say consider the index as a whole mean reverting (like all its underlying) and estimate a Vasicek using it?Any hints?Thanks for your help.

ask jezza.. hes the king.

I sent him an email.Wiating for his response.Thanks.

Could you represent the forward prices of each of the underlying assets at expiration using Clewlow & Strickland's one-factor model (http://www.energyforum.net/feature/feat150.shtml) - which permits vols to decline with increasing future's tenor - & then match moments to the assumed mean-reverting process your index follows (ie, similar to Brigo & Mercurio's paper on basket moment matching http://papers.ssrn.com/sol3/papers.cfm? ... erDownload)?

Sorry Wdido, saw that very late, but here we go.I can't be too specific but here is what I can say.1. how to price an option on a basket on commodities? I am especially concerned on the volatility aspect (let us forget about the correlation...)Let us say all underlyings of the index are mean reverting and have volatility curves in backwardation.It doesnt really matter how theses assets behave or how their forward curves are shaped as long as:- You have the plain vanilla vol associated with each forward price in the basket. The vol must be a standard cumul. vol for the maturity of the option. If you are pricing a serial basket (where say the basket option expires in one year, but the forwards in the basket have a maturity of two years) then, it becomes a very different ball game.- you also have a good approximation of cumulative covariance/correlation. Now THIS is the trick. You can be a little wrong on the vols, but you can be VERY wrong on your end price with correlations. 2. How would that change the pricing of an option on the index?The model is the same (black scholes - pre wash, bla bla bla), but the parameters are very different. Now you need a good term structure of ER vols and a good TS or ER correlations.Have a look at Strickland and Clewlow as already mentioned by ahew, but the models in there are about 5/6 years old, and we do much better now. The thing is, this is an area where profits are still subsantial at banks and the models are still very well guarded.3. Should I say consider the index as a whole mean reverting (like all its underlying) and estimate a Vasicek using it?Absolutely not. Have a commodity by commodity approach and reconstruct your basket.PM if necessary.Jezza