implied volatility forecasting

[lovepopov]

Hello guys,

I am working on forecasting volatilities. Right now I have implied volatilities backed out from options expiring in promt month, 1 month later, ... 6 month later, and i want to use these data as well as historical prices to forecast volatilites for the next year. Any ideas on that? Becasue i am really new on this, and i want to use garch, but how? I am so confused.

btw, if i also want to incoporate the difference between IV and realised volatility in my forecasting (because i back out IV since Jan, so i have all the IV, prices for the past 6 months), then how?

Thanks, and any comment is welcome.

 

[mmone]

hey..if you ever work it out (correctly) , let me know and in turn I will show you how to make some serious money

historical vols and past implied vols will not really help you in forecasting vols for the future (neither implied or actual) only gives you guides as to where it has been..same as looking at historical price movement on a stock and saying it will go up in the future..prob not the answer you are looking for but my two bobs worth anyway

 

[lovepopov]

haha, thanks mmone, i'll let you know once i work it out.

But the problem is that i have to build a volatility term structure, extending to Dec 09. So i guess the historical data are all i have..

 

[jamescummins]

Are you trying to build a path-conditional volatility structure? There are going to be a lot of assumptions involved that make your results pretty propietary which ever way you go about this. If you want to use garch (I dont really see why you are, but then that does not mean anything!) you have to ensure the data is appropriate - not all markets are, or rather, not all data sets have the statistic properties to warrant using Garch.

There are a few different ways to go about modelling what a volatility surface will look like, or rather what you assess its fair value to be. Without knowing what the data is you are building from, and what specifically you are trying to achieve I wouldnt really be of much help.

Which, I just realised, means I have been no help here either! So sorry about that

I would be interested to know which route you went in the end though. And what you are doing this for - just to build a prop model for yourself? Are you trading the vols of it? Loads of questions!

James

 

[lovepopov]

actually i am building this for traders. And i decided to use options on 6-month forward contracts(because i am modeling power products, which are likely to be very unreliable after 6 months) to build my model. I just dont know how to incorporate the IV in the garch model, which usually deals with historical data.

What assumptions that i have to make? thx

 

[A Dashing Blade]

I am working on forecasting volatilities. . . .

Exactly which products ?

 

[jamescummins]

actually i am building this for traders. And i decided to use options on 6-month forward contracts(because i am modeling power products, which are likely to be very unreliable after 6 months) to build my model. I just dont know how to incorporate the IV in the garch model, which usually deals with historical data.

Ahh power products! Interesting stuff
Well you can use either historical volatility, or implied volatility - they are both essentially 'historical' for the purpose of building a forward estimate. If you can let me know exactly what data it is you are using, and exactly what outcome you are hopign to get, I can have a look if you want. I make no promises as to me being of any use though!!

And how complex do you want the model to be? I assume since you are looking at options, and power, and thinking of GARCH, that you are not assuming constant volatility, and hence you want the model! Are you looking to derive when volatility is cheap along the forward curve? If so, I think there are a few flaws in that, but that is a whole new big discussion!!

Basically, the more info you can share, the more we can all try to help. Having said that, if you are getting a monster bonus I am not helping to build a prop model for your employer, traders, and yourself to make a bucket load of cash without sharing lol

 

[tompernon]

Hi lovepopov
I am doing an internship right now and well I am trying to build the same thing
Except I am focusing on soft commodities. I wanted to use Garch(1,1) also but didnt get any really probing results...
I got some old implied volatilities and tried it out and the results were pretty good but with all that has been happening in the last 3 or 4 years I need some newer data to verify if it actually works now...
I would be really interested in seeing if you get any good results, maybe we could share.
I ordered the rest of the data and should be receiving it today so I will try it out tonight.
Keep in touch

 

[AnEdge]

While I understand that you are trying to forcast Implied Volatility...how can you actually do that ? You can look at historical IV and then try to predict it forwards, but my guess is that it would be very hit/miss...for example just look at the last month in the markets. Even more regular patterns like IV changes for Stock Options due to Earnings which have traditionally been very predictable are now evolving changing as more people trade them...

So how do you propose to actually do this, can you give any of the details ?? A lot of readers here would probably be interested...