gamma01 www.wilmott.com volatility process is exogenously speci
http://www.wilmott.com/messageview.cfm?catid=3&threadid=44947
Thu Jan 11, 07 05:02 PM
|
|
Realized and implied volatility was low in FX markets for most of 2006. The Jan 2007 CitiFX Currency Advisor says
"Two problem arise when considering a long volatility position in the current market.
The first problem is that due to depressed short-term realized volatility in spot, many computer models have consistently produced sell signals. Therefore dealers have consistently been given vol. This combined with a natural propensity of dealers to be long gamma and created an environment of bank desks consistently taking profit on spot moves in order to capitalized on their long gamma positions generated by model accounts selling vol. The take profit hedging activity contributed to further depress short-term realized vol levels, thus creating a vicious cycle."
Do you think dealer hedging of long vol positions is depressing volatility? |
|
|
|
dopeman Member
Posts: 40 Joined: May 2006
|
Thu Jan 11, 07 11:58 PM |
|
Yes, dealer hedging long gamma positions can depress realized volatility. In the standard option theory, the volatility process is exogenously specified, whether constant or stochastic, ie, there are no feedback effects. However, feedback effects do exist in real life, when the (hedging volume)/(other volume) ratio is not small. In the extreme case, the underlying asset can be "pinned" at the strike price, which is a well-known fact in equity markets on option expiration days. |
|
|
|
Alan Senior Member
Posts: 4643 Joined: Dec 2001
|
Fri Jan 12, 07 04:31 PM |
|
I elaborate on my previous question.
Looked at a couple of papers briefly, esp. a nice empirical study by Ni, Pearson, and Poteshman. (Stock Price Clustering on Option Expiration Dates). I don't quibble with the data -- definitely something going on.
Here is the main theoretical explanation: long vol + delta hedging -> pinning at the strike. The argument is starting to bug me a little.
Here is my summary of the theory argument: Consider hedging a long call under Black-Scholes. It's easy to show or visualize that d Delta/dt flips sign as the stock price crosses the strike (and as time to expiration shrinks to zero.). After all, Delta is tending to 1 or 0 on either side. If you imagine S near K and Delta(t) near 1/2, then Delta(t) will be increasing on one side and decreasing on the other as time passes. The implications: If you're long a call and delta-hedging your position, then you're going to be selling stock to adjust your position when S(t) > K and buying if S(t) < K. If you're doing this in size and your counter-parties aren't, then this will tend to pin the stock price at S(T) = K. The argument also works if you're long a put.
Does this argument hold water? How model independent is it?
-If- the stock price, say, is going to be pinned at the strike, then the starting formula (Black-Scholes) is badly flawed since the process is not GBM at all. Of course, we know the GBM idea is flawed for other reasons. But there's something a little fishy here -- a lack of self-consistency or at least closure in the argument.
Suppose you're a market maker who is net long vol. in a group of market makers who tend to delta hedge. Expiration is coming up and the stock price is close to a strike. You anticipate that it may be pinned. What do you do? Since everybody is thinking the same way, what is the net result? (Honest question - if anybody can close the argument, I'd appreciate hearing it.)
Edited: Fri Jan 12, 07 at 07:27 PM by Alan
|
|
|
|
您的位置:
文学城 »
论坛 »
音乐快递 »
gamma01 www.wilmott.com volatility process is exogenously speci