I have a question on pricing heat-rate options with daily strike capacility. Say for e.g. the option is active from Jan-10 onwards.From now till Jan-10, I use forward volatilities for power and gas to simulate price paths. Once I enter the month in which the option is active, I use spot volatilities for power and gas and spot correlation in order to simulate the intra-month power and gas prices. If I use this approach, the option price that I get after discounting to the present day is significantly higher than the price quoted in the market.However if I use a weighted average of forward and spot volatilities for power and gas in simulating the intra-month price paths, the resulting option price is much closer to the market quotes.Is anything wrong with the first approach? To me, the first approach seems to be correct theoritically.I would like to know how the market is valuing such daily options based on your perceptions? ThanksKarthik

Hi,Just saw this post, a couple of comments.Margrabe's formula works when there are 0 fixed costs; often times a spark spread or heat rate option will trade with Variable Operating and Maintanence (VOM) as a fixed cost. In this case, Kirk's approximation works fairly well; We have also programmed the Carmona/Durrleman method and it seems to work pretty well.For more information on pricing spark spread options, Energy and Power Risk Management by Eydeland and Wolyniec is an excellent resource.As far as pricing heat rate options with daily calls, pay attention to the Eydeland and Wolyniec sections on correlation.

Well, how much trust do you have in the spot vols? Generally, either blended or forward vols are good estimates, since spot quotes are going to be very nosiy.Also, pay attention to the impact of correlation, which might lead to a break down in your valuation model. Aside from a Real Option Approach, I think what you have seems reasonable (for what it is worth).

Where do you even get Spot Vols? Are these historical estimates? " if I use a weighted average of forward and spot volatilities for power and gas in simulating the intra-month price paths, the resulting option price is much closer to the market quotes" What kind of correlation is getting your Vols to price near market quotes.

QuoteOriginally posted by: rich7804Where do you even get Spot Vols? Are these historical estimates? " if I use a weighted average of forward and spot volatilities for power and gas in simulating the intra-month price paths, the resulting option price is much closer to the market quotes" What kind of correlation is getting your Vols to price near market quotes.ICAP has some quotes, then do some mapping if needed.

AFAIK ICAP has daily vol quotes not spot

Hi Beachcomber,Thanks for your suggestion. I shall look into 'Eydeland and Wolyniec' to get a better understanding. Hi diogenes,I do not have a great deal of trust in 'spot vol' for power. It is extremely noisy. But I am comfortable with spot vol: for gas except for winter months.ICAP has daily vol quotes for selected months.Thanks for your response.Hi rich7804,Yes the spot vols are based on historical estimates. However I am not inclined to use spot vols calculated this way bacause the price levels of natural gas traded in 2007 and 2008 are different from what we are witnessing today. The prices have gone down.As far as correlation is concerned, a blend of forward-spot corr: is getting me the price closer to market quotes.I appreciate your points and thank you for the response.

QuoteOriginally posted by: rich7804AFAIK ICAP has daily vol quotes not spotYup, good call.

volatilities are nothing. what is problem- correlation. there is severe correlation smile. You can estimate volatilities from past data. Correlation- is different matter. The only way to estimate it is to run historical analysis of hedging given option, and then you will be wrong

In some markets (i.e. PJM, NEPOOL) you can get some HR option quotes from brokers and you can backout the market implied correlation. "severe correlation smile" Can you even see the power vol smile? How do you estimate the smile from hist data?"You can estimate volatilities from past data" if you are comfortable with your Vol forecasting you might just forecast the HR Vol that way you don't even need worry about the correlationI personally think all these historical analysis need to be normalized for a particular Gen Stack...It's hard to price these options using a BS framework, given incomplete nature of these markets you just have to leave with a lot of Vol and Corr exposure.

great thread, anyone on here live in Houston? I'm building an options model and would love some feedback, or even just to test some of my values with someones model