Modelling Skew

[Profitaker]

Where the volatility smile for any particular option month is known, how, or using what formula, is the option price change <i>solely</i> attributable to Vol smile (skew risk) calculated ?

I know you could manually enter the changing IV into the BS model, but is there a less tedious way of incorporating this phenomena into the model ?

 

[gbos]

For moderate skew and kurtosis values you can try the Corrado-Su model. See attached excel file.

Regards

 

[riskarb]

It's easy to implement in excel. Simply run any model assuming a flat vol-surface and run a column with an +/- fairval using atm implied as model vol. It's made more difficult when attempting to model vol-convexity since sticky stike and delta models are worthless.

 

[Profitaker]

Quote from gbos:

For moderate skew and kurtosis values you can try the Corrado-Su model.

That looks really useful, thanks. Welding it into the BS model though might prove quite challenging - I can see lots of circular references cropping up. Unless of course, you've prepared one earlier ?

Quote from riskarb:

It's made more difficult when attempting to model vol-convexity since sticky stike and delta models are worthless.

Yes, it's precisely the convexity I'm having trouble with. Think that model from gbos just might work.

Thanks.

 

[riskarb]

It's not ideal as it's a sticky-skew model, but I can't reference anything better.

 

[Profitaker]

Quote from riskarb:

It's not ideal as it's a sticky-skew model.

Sure, but I think I could change the "strike" input of the Corrado-Su model to that of the underlying (fed from a dde link), so that the vol smile would actually move in line with underlying movement, rather than stick.

Don't know, early days yet. How do you account for it ?

 

[riskarb]

Quote from Profitaker:

Sure, but I think I could change the "strike" input of the Corrado-Su model to that of the underlying (fed from a dde link), so that the vol smile would actually move in line with underlying movement, rather than stick.

Don't know, early days yet. How do you account for it ?

I use a flat-surface and model the variance at 25d. It's more important to me to know the over/under fairval figure.

 

[Profitaker]

Quote from riskarb:

I use a flat-surface and model the skew variance at 25d.

25d ?

 

[riskarb]

Quote from Profitaker:

25d ?

25deltas -- the wire services use the 25delta fig in their risk-reversal calculations.

 

[Profitaker]

Quote from riskarb:

25deltas

Thanks. Do you use some sort of formula to re-calculate option value using the new IV, or do you input it manually into the model ?