Use greeks: Predict extrinsic value change after an X% move in underlying

[Kevin Schmit]

2 questions:
1) Doesn't this assume that implied volatility is only a function of strike price?

Yes. Not a very realistic assumption 25 days out. The problem is that the evolution of vol is path dependent -- a 10% move over 25 days will affect vol differntly if the change happened all in one day with only small moves the other days vs a steady rise of < half a percent a day. Back-of-the-envelope approximations for vol changes are ok for one day (or hour) out, but pretty useless fpr a 25 day out forecast. So, on second thought, you should probably ignore my previous posts on this thread.

2) Do you know how to consider directionality when predicting the change in extrinsic value?

Since extrinsic is a linear function of implied vol, you need to create a model for the evolution of iv, and then simulate a few million iterations over the coming 25 days, keep all iterations that showed a 25% or greater move in the underlying and the expectation (mean) of the distribution of those kept terminal [iv converted to] extrinsic values is your answer. Try a GARCH-style model with an additional term of sign(-ln(close[t-1]/close[t-2]) * abs(ln(close[t-1]/open[t-2]))[/QUOTE]

 

[tommcginnis]

Need4Greed

• Don't sweat the small stuff.
• All you have to know is that ITM options will change in value almost dollar for dollar to the underlying.
• All other variables will have a negligible impact on the R:R.

Unconscionably wrong again.
(But I'm betting you know that, and just get some sort of perverted pleasure in the thought that you might trick someone. Sad. Sadddddd, you.)
SO:
1) Not sweating the small stuff will blow your account. Always and without exception.
2) Options BY CONSTRUCTION will change as the underlying changes, according to delta, which ranges from |0.000| to |1.000|. It is NOT a "dollar for dollar" relationship by any means.
3) Reward to Risk relationships are situationally specific to time and volatility, and deeming them of "negligible impact" is, like not sweating the small stuff, suicide.

Okayyyyyy! Y'all know that. I know. I know.

 

[Need4Greed]

Need4Greed

• Don't sweat the small stuff.
• All you have to know is that ITM options will change in value almost dollar for dollar to the underlying.
• All other variables will have a negligible impact on the R:R.

I don't think your assumption holds if I buy a call when IV is at its peak. The volatility drop would cause a big drop in premium.

 

[OptionsOptionsOptions]

I don't think your assumption holds if I buy a call when IV is at its peak. The volatility drop would cause a big drop in premium.

• It all comes down to what the underlying is doing.
• High IV or Low IV it doesn't matter.

 

[Need4Greed]

• It all comes down to what the underlying is doing.
• High IV or Low IV it doesn't matter.

You may be forgetting the equation:

Change in premium = change in intrinsic val + change in extrinsic val
Change in intrinsic val = movement of underlying stock
Change in extrinsic val = decay in time value (affected by change in IV and change in days till exp)

Clearly the movement in the underlying is only part of the picture.

 

[TheBigShort]

Yes. Not a very realistic assumption 25 days out. The problem is that the evolution of vol is path dependent -- a 10% move over 25 days will affect vol differntly if the change happened all in one day with only small moves the other days vs a steady rise of < half a percent a day. Back-of-the-envelope approximations for vol changes are ok for one day (or hour) out, but pretty useless fpr a 25 day out forecast. So, on second thought, you should probably ignore my previous posts on this thread.


Since extrinsic is a linear function of implied vol, you need to create a model for the evolution of iv, and then simulate a few million iterations over the coming 25 days, keep all iterations that showed a 25% or greater move in the underlying and the expectation (mean) of the distribution of those kept terminal [iv converted to] extrinsic values is your answer. Try a GARCH-style model with an additional term of sign(-ln(close[t-1]/close[t-2]) * abs(ln(close[t-1]/open[t-2]))

[/QUOTE]

Yes. Not a very realistic assumption 25 days out. The problem is that the evolution of vol is path dependent -- a 10% move over 25 days will affect vol differntly if the change happened all in one day with only small moves the other days vs a steady rise of < half a percent a day. Back-of-the-envelope approximations for vol changes are ok for one day (or hour) out, but pretty useless fpr a 25 day out forecast. So, on second thought, you should probably ignore my previous posts on this thread.

[/QUOTE]

Kev if I plot the dow jones components 1 month IV vs 1 month lag close to close vol, the spread is pretty close to 0. Its usually slightly above 0 because the shorts are short convexity but I mean I think it does a pretty good job at predicting forward realized vol. The Garch family indicators seem to slightly over estimate vol. Here is a website that uses GARCH family to predict future vols on stocks. https://vlab.stern.nyu.edu/en/