GARCH, etc.

interesting debate. the poster never did quite get an answer though! I never knew you mathematical types were such hard asses. Must be a sign of the (very contentious) times.

As for the question of volatility, I'm a relative novice, but may be able to offer some insights.

Obviously the implied is calculated directly from black-scholes. http://www.ivolatility.com/calc/?ticker=MSFT
Historical volatility is defined here http://www.aspenres.com/Documents/help/userguide/help/aspenHistorical_Volatility.html

As for the historical, I've been attemtping to derive my own methods to calculate the most accurate measure for my trading style. (Swing trades 3-5 holding periods)


A few methods I've been toying with:

1. daily & weekly atr as a percentage of the close
2. differences in weekly/daily atr between 5,12,26 periods
3. average move between peaks/troughs of various time periods ...
4. Of course, the standard Historical Volality calucations ...
5. .... to be continued... all sorts of ways if you have the data


I'm fundementally interested in the probability of a 4-5 + point move over the next 3-5 days. I developed a fairly comprehensive database that I can query ...

I've admitted to being a novice, so please refrain from bashing me too badly.
 
Quote from pattersb:

I'm fundementally interested in the probability of a 4-5 + point move over the next 3-5 days. I developed a fairly comprehensive database that I can query
More...

get your returns and use a prob distribution curve
 
Quote from nonprophet:

Allright, but do we really need GARCH for that? Can't we just buy some deep OTM's and be done with it? (not my idea but Nassim Taleb's).
More...

I agree that perhaps you don't need it: the derivative's worth whatever two people agree upon at that instant (with their information set). Derivatives are just another way to transfer risk. End of story?

In a BS world, the writer does not have to agree with the buyer on probabilities for future states, because he can use the underlying to hedge, and all writers have the same "program." Question is, if the world is not BS, can one writer do better (hedge more efficiently) than another using a better mousetrap (GARCH) - take away the market from the BSers and earn extra profit?
 
Quote from aPismoClam:

welp, glad you guys enjoyed this thread. back to the textbooks and Wilmott forum for me.




Hey, moron, what's your issue with Jesus Christ? Why do you spout idiotic things about Him at every opportunity?

You know, I did that for a while when I was twelve years old and used to write short stories.

I daresay, the time may come when you realize how juvenile (and curious) your obsession is.
More...

I think what pisses you off is that I regard Jesus as a human being who was a part of history. I don't go along with the standard religious mythology and you object. Instead of stating your objection right when the post occurred, you waited until you got something that might be of value to you, and then offered your silly little comment. This approach makes you look like a sniveling little weasel. Just for the record, thats how I view it.
 
Quote from MoralHazard:

I agree that perhaps you don't need it: the derivative's worth whatever two people agree upon at that instant (with their information set). Derivatives are just another way to transfer risk. End of story?

In a BS world, the writer does not have to agree with the buyer on probabilities for future states, because he can use the underlying to hedge, and all writers have the same "program." Question is, if the world is not BS, can one writer do better (hedge more efficiently) than another using a better mousetrap (GARCH) - take away the market from the BSers and earn extra profit?
More...

Maybe you're not really understanding the concept of forecasting volatility

In spite of the fact that two parties can agree to price a derivative in the present, its value in the future (when you may want to exit the trade) can and will change. The ability to hedge dynamically is limited and it adds cost (therefore risk) to the trade. If one had a way of forecasting volatility "shocks" to the system that superior knowledge would allow you to know when to buy and sell for your greatest advantage. Please do continue to use B/S though, if it works for you.

Thanks
Steve
 
Quote from momoneythansens:

FWIW, there have been several studies on the benefits of using the various GARCH models for forecasting volatility on broad index products. The consensus was that these models were no better than the predictive qualities provided by ATM implied volatilities.

....
More...

Useful post. I have not searched, but will. ATM IV from a Gauss fit would certainly match ATM IV from a GARCH model if the tails were tossed out (assuming they are in fact fatter); you add the caveat about fear and greed. Do you know of any studies on indiviudual stocks?
 
Quote from steve46:

Maybe you're not really understanding the concept of forecasting volatility

In spite of the fact that two parties can agree to price a derivative in the present, its value in the future (when you may want to exit the trade) can and will change. The ability to hedge dynamically is limited and it adds cost (therefore risk) to the trade. If one had a way of forecasting volatility "shocks" to the system that superior knowledge would allow you to know when to buy and sell for your greatest advantage. Please do continue to use B/S though, if it works for you.

Thanks
Steve
More...

Seems to me that that's roughly what I was saying. Explain to me why I don't understand the concept of "forecasting volatility."
 
Sure,

The distribution isn't gaussian. So you are mis-understanding the whole concept. It is much more likely that extreme events will happen on both sides of the distribution than your outlook predicts (assuming you think in terms of normal distribution).

Since the frequency of these events is significantly greater, the probability that an event can change the value of a instrument is greater than predicted by standard IV (implied).

Finally since volatility "shocks" seem to cluster together, if you had a consistent approach to deal with that, you would be in a position to buy or sell volatility with more accuracy.

Your recent comment about "tossing out" the tails indicates that you're not seeing the whole screen.

By the way, a derivative isn't worth what two people agree upon. If that were so, no one would buy or sell derivative products. We buy and sell based on our prediction of what a derivative will be worth in future. It is precisely because at least one person disagrees as to future worth that the transaction occurs.

Thats about it.

Good luck
Steve
 
Quote from steve46:

Sure,

The distribution isn't gaussian. So you are mis-understanding the whole concept. It is much more likely that extreme events will happen on both sides of the distribution than your outlook predicts (assuming you think in terms of normal distribution).

Since the frequency of these events is significantly greater, the probability that an event can change the value of a instrument is greater than predicted by standard IV (implied).

Finally since volatility "shocks" seem to cluster together, if you had a consistent approach to deal with that, you would be in a position to buy or sell volatility with more accuracy.

Your recent comment about "tossing out" the tails indicates that you're not seeing the whole screen.

By the way, a derivative isn't worth what two people agree upon. If that were so, no one would buy or sell derivative products. We buy and sell based on our prediction of what a derivative will be worth in future. It is precisely because at least one person disagrees as to future worth that the transaction occurs.

Thats about it.

Good luck
Steve
More...

Thanks, a lot of stuff there.

The innovations can be conditionally Gaussian (or not), and nowhere did I say the unconditional distribution is Gaussian. See my other coments in reply a point by nonprophet. The tails in the unconditional distribution and autocorrelation in the innovation^2 (and covariace matrix) is precisely why asked about G/ARCH being irrelevant to a trader.

My comment about tossing the tails came after yours, and relates to the empirical evidence that momoney brought up. Perhaps you missed the word "if."

The last para of yours opens a can of worms, so tread carefully. Derivative products can be valued, in principle, since the underlying is tradable. Speculation, which is what you seem to have in mind, is one thing; hedging and risk transfer is another. I buy car insurance because I am ra and don't like losing a large fraction of my wealth should I accidentally injure someone - so I pay a premium. The insurance co is rn and can lay off the risk. No one *needs* to feel they're getting a better deal (though of course this may be the case).
 
You must log in or register to reply here.
Back
Top