PHENOMINAL ARTICLE!! Need to update my R code now! However when they talk about hedging at "lower/higher then implied vol". Lets say I buy a straddle and at the time IV is 10%. Tomorrow the stock jumps 10 points and I am long 100 deltas, the IV is 9% and I decide to short 100 deltas of the underlying. Am i locking in a loss? Since I hedged at a lower IV???
Bear in mind the article includes data only to 2005 (!), so the empircal results (such as Figure 1) are outdated -- though presumably the math is fine.
If the Math is fine I don't see why the data matters. This article and strategies are just as applicable in todays markets as they were when the article was written. If you can predict vol better than the market, you will make money. Do you think otherwise?
If you are hedging a long gamma position at volatility that's lower than realized you will be increasing your total expectation of PnL but also increasing variance of the PnL. Here is a simple example - imagine that you buy an option that's slightly OTM and then decide to hedge it at zero vol. You can imagine that in that case many of the possible paths will lead to PnL of zero, but a big portion will lead to a very high windfall. Alternatively, imagine that you are hedging that option at very high vol - you are not going to be rebalancing your delta frequently enough to make money, but you will be losing money very smoothly. In a non-theoretical setting (especially for a prop trader who is not judged for his MtM), your hedging frequency, AC of the underlying and many other things will matter more than the implied volatility that you are hedging at.
Thank you for simplifying that for me!!!! And there is only 1 other poster i know on ET who uses AC for auto correlation..great to see you back on ET!!!! For a non money manager, the article mentions using actual vol to hedge is optimal even tho as you said pnl would be more volatile. Say I buy 10 day a straddle and the Parkinson model (which I use for HV) says past 10 days was 15 vol but iv says 10. Would it make sense to use that to calculate my Greeks instead throught the options life?
You will probably be short the indexed gamma, since you're selling the low IV and buying the high IV. This would mean you'd be paying theta for a negative gamma. Doesn't mean it has to lose money... you just hope the index stays more steady than AVGO. But a 1:1 move does mean you will lose...
If you're long 100 deltas, that basically means the call now has d100 and the put d0, so... there's no more vega left in it... IV could drop to zero and you would not lose on that shift. The question should be, did the 10 point rise cover the straddle value or not?
What does the article conclude? My intuition says you should hedge with your actual/real vol (15). You think that real vol > implied vol, so the greeks based on a higher IV will show lower gamma, therefore you would underhedge compared to implied. Which is what you want, since you think it moves more anyway, so you get more opportunity to hedge.
If I am shorting the index gamma, am I not shorting the High IV (index vol is usually rich). I see what your saying, but relative to HV im shorting the expensive gamma(SPX) and buying cheap (AVGO). The article concludes the same, if your not a PM who has to Mark to market then hedging at realized vol is optimal. But what gets me is, should I keep my original forecasted vol as the input for the life of the trade, or should I keep adjusting the vol input as my outlook changes.
Yeah I know Index is usually rich, that's why dispersion is usually done this way, short index long singles. With indexed gamma, I meant the nett portfolio gamma... you would be short. So that means when they both move 1:1 you will lose money. Regarding vol input, I would say change it when your outlook changes. There's no point in hedging on a wrong basis. Initially you assume real vol > implied vol, so you'd hedge with real... but when real vol doesn't exceed implied, but are the same the hedging wouldn't matter... since they are the same. And if real drops below implied... I guess you would exit.