wilmott.com

Hmmm..."Effective pricing of wind power" looks good, but no paper link?

QuoteOriginally posted by: outrunQuoteOriginally posted by: Hendersonthe article is at:http://www.maycroft.com/resources/Artic ... 2003.pdfIt would be a interesting to send Maycroft and email (dejong @ ...) He will be publishing a new paper on gas storage in the journal of derivates later this year, he also has some conference papers, and he's always happy to help.Great!

Hi,I have gone through this interesting thread and have struggled to grasp one particular point made by a few people: the fact that the "rolling intrinsic"-based valuation cannot be hedged.What does that mean precisely? Are we talking about a model-independent hedge? In that case, I can see there are a priori no instruments out there you can use to fully lock in your storage "price" (= the expected value of the rolling intrinsic strategy).But assuming we are happy with the specifications of the underlying (forward curve dynamics), can't we generate deltas, vegas...etc and hedge the storage using forwards and vanilla options? There will still be un-hedgeable risks (e.g. the correlation between the forwards, or more generally any model parameter that is not captured in vanilla prices) and one will have to come up with a number for those, but that is what traders do.Am I missing something?Thanks in advance.

Rolling Intrinsic consists in hedging the optimal profile at t_0 and adjusting afterwards the hedging portfolio according to the new optimals.Thus your hedging portfolio at time t should only be reflecting your optimal profile at time t; otherwise it would mean that your profile at time t is not perfectly hedged and you would not be applying the RI strategy.

Spread options is the worst one- there are no implied correlations and you have all the problems with correlation smiles- i.e. summer never goes above winter (Europe).Spot based is ok, but still you have to be careful to calibrate to spot volatilities/mean reversion- i.e. gas curve does not behave as 1-factor or 2-factor models.For calibration in Europe you need to have a thourough understanding of pricing not only NBP, but also Russian gas flows.... and LNG supplies- i.e. HH curves.Rolling intrinsic- the easiest one to follow and a lot of "traditional" traders use it as there are many other things to worry about.You can not hedge even remotely vega of storage with vega of vanilla options- they only share the name, but thats it. Only people from academia fall in that trap.The only way to proxy hedge is with other storage or swing or daily or LNG terminal or weather derivative. But noone trade weather derivatives.

does the govt subside insurance?

Pasargad - the problem is precisely with the specification of a gas model that implies fwd curve dynamics one can trust. I am not aware of any that have been back-tested and work well.I tend to agree with jd113 in that the CSO-based approach is the only one that is remotely hedgeable. The problem is with a specification of the correlations (even if one uses a calibrated factor model) and especially the difficulty of replicating the value of the CSOs using delta + vega hedging strategy to moneytize the extrinsic value (since one really can't sell the CSOs).

Hallo,in my phd theses I work with Gas storage and now I would like to implement the Modell of Boogert and De Jong in Matlab - is there anywhere a matlab code available? Its verry hard to develop a completely new code.ThanksThomas

Dear jd1123,I am actually taking the spread option approach and have some specific questions. The hardest part for me, so far, is the estimate of the volatility (for the spot) and correlation (matrix when considering different months). My problem is specific to a option to park. So my approach is to capture the spread between the month of park and month of withdrawal. Do you have any techniques so to have an arbitrage free matrix. Also, what techniques do you recommend for the estimation of the spot volatility for the park month?Any expertise will be much appreciated!

For the volatility of park month and withdraw month, i assume there is no forward hedges in place. In that case, I would look at the history of the two months to come up with historical volatilities. I assume your exposure is month to month, versus daily.There is a new paper about spread options approach and estimation of correlations. I am not sure if it answers your specific question about parking, but I think it is worth looking at. http://papers.ssrn.com/sol3/papers.cfm? ... id=1687313