Estimating vol skew from historical underlying price data

[sonoma]

Quote from TskTsk:

Yeah, I actually found some papers on it.

Are you looking to demonstrate "what if" scenarios to a group of novices, or are you looking to place positions based on some mispricing revealed by your skew model? If the latter, you can't do it. Too many uncertainties, especially in the back months and specifically if your position will be functionally path-dependent.

 

[opt789]

Quote from TskTsk:



I dont have the actual historical option prices.

Then pay a small amount of money for them, and stop playing guessing games.
http://www.marketdataexpress.com/

 

[TskTsk]

Quote from sonoma:

Are you looking to demonstrate "what if" scenarios to a group of novices, or are you looking to place positions based on some mispricing revealed by your skew model? If the latter, you can't do it. Too many uncertainties, especially in the back months and specifically if your position will be functionally path-dependent.

Looking to demonstrate what-if scenarios first and foremost, to one particular novice...myself

Quote from opt789:

Then pay a small amount of money for them, and stop playing guessing games.
http://www.marketdataexpress.com/

It's much more practical dealing with one set of widely available, free data, rather than a ton of individual data per symbol and maturity. As long as you're aware of model limitations, I don't see the problem.

 

[Martinghoul]

Quote from sle:
The answer is both "yes" and "no". There is a whole bunch of statistical studies you could do, depending on what you are trying to achieve.

+1

All sorts of things you can do. Whether they're useful in practice is an entirely different question.