the issue with skew is that at the end of the day you are trading fixed strike vol (unless you are trading variance). So the 90-110 today becomes the 100-120 and with that your Vega and gamma profile changes. So you have to make spot vol correlation bets that are tricky. Sometimes the vol resets up. Sometimes it resets down. In the 5percent rally, vol fixed strike vol (at least front month) didn’t come in that much. But it easily could have.
Of course. But nonetheless I'm still having issues calculating a relationship that should exist. https://www.elitetrader.com/et/thre...e-volatility-plays.301129/page-2#post-4303755 Is this relationship consistent with what you have observed as well?
my experience has been that if you are short skew you tend to lose ALOT of money when vol goes up. I don’t know how much is from skew steepening vs being short Dgamma/dvega as the whole curve goes up or getting shorter gamma because you’ve moved into your short strike. Or because you are short vol of vol and that goes up. I don’t think skew should necessarily steepen if vol goes up but common sense would say that the putside should get bid, but my experience is that the call side also gets bid.
My definition of realized skew is different from sle’s. I would call realized skew as Vega pnl of your two strikes + gamma pnl related to the marking of those strikes. I don’t know what he’s defining there.
All of these complexities and position management issues are why I don't trade options anymore. Once I started trading stocks, never looked back.
I'd have to know more detail about the regressions you're assessing in order to comment. But to your last sentence, if you elaborate a bit more on your thoughts, maybe you'll get some pointers from the forum. You've already gotten several good comments from others that are food for thought. For instance, maybe you're exploring whether there's a trade you can arb depending on skew? Or perhaps you're exploring whether a change in skew can be used as a forecast of some other variable? There are certainly trades that can be constructed based on degree of skew (percentiles, etc), but whether they result in profit is another matter. And it's relevant to remember that changes in realized vol need not only be to the downside.