** Don't you mean that skew flattens when realized starts to exceed implied? ** Of course vol can take any number of paths, but take the last couple of weeks in the sp500 as an example. The second week of Feb had a few days of relative calm and the front month 25d rr priced out at a 4-6% vol diff at that time. But from late last week realized vol has had a bit of a ride, and the front month 25d rr has nearly doubled to 9-11%.
For some reason when I plot RV-IV against the realized skew, I don't see the relationship you speak of. Instead, I see no relationship. I've seen other people mention the RV/r-skew relationship so I know it must be an error somewhere in my methods, but I'm not sure what I'm doing wrong. Any pointers?
What would be the fundamental reason/logic behind why this happens? Is there any? (Skew steepening when RV exceeds implied)
The answer to the question, is that no, the skew does not predict a current IV based on it's RV after a market crashes. That said, if you're going to predict the volatility of skews when the market changes, a variable is what the skew is currently compared to recent history at this atm volatility and market conditions. The skew right now could be relatively sharp, so when the market decreases, it could even respectively decrease compared to the atm's, even if atm IV increases. And you used the word 'crash', but that is a very subjective term...what does crash mean in terms of movement over time? Massive move overnight? Over the next month? So the study of RV vs IV, and broken down further into relationship to changes in skews, is very complex because no move is exactly the same as another in terms of how far it goes, how far it goes in relation to previous recent volatility, how long it takes to get there, what the current IV is compared to RV, and then that relationship to skew and it's recent history and so on. IV, including skew IV, is both a compilation of current RV and anticipated RV, which is a very subjective (and often emotional) condition based on current conditions and sentiment. Predicting all of those variables, including sentiment, in the future is a tall task. That said, my apologies if I didn't read the question correctly and this is way off track.
I think with enough sophistication, tracking the skew curve over time and assigning values at different points in the skew that indicate it's level of acceleration (or deceleration) from the atm IV, along with how many days until expiration (because skew will change as it gets closer to xdate), it could be done pretty successfully? I'm sure otm puts must sharpen in down moves and vice verse, and those swings, especially sharp swings in the entire skew, must have profit opportunities...but yeah, that's not me lol.