wilmott.com

would anyone know how to model cliquets on spot commodities. Example oilHere you would need to look at more than one factorthank you

I'm going to copy the reply (slightly edited) I made to a similar-ish question (the question then was about natural gas) months ago._________________________________________I assume you want a model for pricing options on energy-related commodities. A good place to start would be the book by Clewlow and Strickland. There are also papers by Clewlow and Strickland, by Miltersen and Schwartz and by Miltersen. These papers all describe diffusion models. If you wanted a high-end sophisticated model, the state of the art model seems to be the Crosby model. A paper about this model is in Risk magazine (May 2006). The Crosby model includes those of C+S and M+S as special cases (ie when pure diffusion) but it also has multiple Poisson jump processes. Also, when there are jumps in the futures/forward prices of energy-related commodities (and by the way, jumps are a very important feature of markets like these), you see that the long end jumps a lot less than the short end. The Crosby model is the only model I know about which models this feature (it has a sort of exponential damping (by tenor) on the jump sizes). The model also gives you a chance to fit vol. skews which are often significant in the energy-related commodities markets (because of the jumps). The Crosby model is sophisticated but would probably take a while to implement. It depends what you want. If you’re a complete beginner, Crosby would be too ambitious/too much work. It might be best to start with a one factor C+S type approach and work your way up as you get more familiar and as greater sophistication is required.All the above papers are available by googling.There is also the book by Helyette Geman (highly recommended)._______________________________________There are another 2 papers from Crosby which are available here:http://mahd-pc.jbs.cam.ac.uk/seminar/2005-6.html(scroll down the page until you get to the seminars dated 7th October 2005).The first paper shows you how to do Monte Carlo within the model.The second paper specifically looks at pricing cliquet structures, using Fourier methods.I would have thought, personally, that multiple factors would be useful as long as you can calibrate everything okay.

Thank you. I shall look at the papers and revert.many thanks for your prompt assistance.

Is the multi-factor Crosby approach analogous to Libor Market Model, but using commodity futures rather than fwd LIBORS?

No because each forward can be considered as a martingale under the EMM provided you are assuming no correlation with interest rates.

I'd say the Crosby approach was more analogous to a HJM approach but equally I wouldn't push the analogy too far.

I recently read a paper about modelling power markets (Audet/Heiskanen) where they simply model all the forwards as correlated factors with a parameterised correlation structure.Its different from Crosby in that there is no dimensionality reduction - you basically have as many factors as there are forwards.Does anyone ever use this type of thing in commodity markets or is it much more common to use Crosby/Clewlow and Strickland type approaches?