On Short-Term VIX Mean Reversion and Echo Volatility
A reader asked a question about the relationship between the magnitude of short-term VIX mean reversion and the implications for long-term mean reversion and echo volatility.
After doing some research, it turns out that this is a very important question. Specifically, there is a strong inverse relationship between the magnitude of short-term VIX reversion and the continued strength of the longer-term mean reversion trend. Said another way: if the VIX snaps back dramatically in a couple of days, it is more likely that echo volatility (VIX aftershocks, if you prefer) will be an important factor in the next few weeks.
On the one hand, this should not come as a surprise: if most of the mean reversion happens in a couple of days, then less of that same mean-reverting move remains for later. On the other hand, my research suggests that you can use the magnitude of the short-term mean reversion move to measure the magnitude of the long-term mean reversion with surprising accuracy – at least with the benefit of 20-20 hindsight. Whether the future will mirror the past remains to be seen.
Of course, the larger question involves expanding the implications of this finding from the VIX world to the world of the SPX and the broader markets. In this arena, the data look to be encouraging as well.
Arrow up on the predictive value of the VIX for the broader markets.
7 comments:
Bill, the VIX is a key element of the global liquidity puzzle. Because it is the best market-based proxy to the stability of the business cycle, it provides important information about the value of collateral -- a key factor in terms of the supply of credit.
Bill,
Excellent post, as always.
Sorry to refer to Echo Volatility as "aftershocks", it just seemed fitting at the time of the comment.
From your analysis, what time frame was echo volatility the most prevalent? Was there a strong possibility of echo volatility immediately following a VIX spike (1-2 weeks) and a strong mean reverting move? Or did your research simply suggest that it was more likely in the more distant weeks that follow the spike?
I don't know if you have any opinion or information in this regard, just curious as to your take.
Also, it sure seems the VIX is doing it's best to make those VIX callers come up empty and make those spreads you described profitable. Not that the amount of calls on the VIX has anything to do with the likelihood of a volatility "event", still, it is interesting.
Bill,
I have noticed that the last several (recent) market declines have been preceded by an unusual level of what appears to be VIX call buys.
The last "correction" before this most recent event, I noticed that approximately 100,000 VIX calls traded the day prior.
Have you noticed this as well? Do you track this in any way? I recognize that it can be difficult to tell who is buying or selling on the day (unless you can watch big blocks go off at the ask, which can be difficult to monitor), but I have confirmed that these were buys by tracking the open interest in the past. Today there seemed to be at least one big block of Jun 20's and many smaller blocks. However, I don't know if that means much given the relatively low cost of those calls. Also, at 15K, that is hardly significant.
I simply point this out to determine if this could be another potential VIX "tell" as to the short term prospects of the broader market.
Robert,
Regarding echo volatility ("aftershock" is just as good a label, FWIW), while it is not statistically significant, I have found that echo volatility tends to cluster in the neighborhood of a VIX spike + 10 trading days. The graph I used for a post-2/27 analysis shows that aftershocks are most common/severe in the 9-15 day range.
Echo volatility didn't show up much in the 4-8 day time frame, but was present in the 1-3 day time frame, which I attribute to event-specific skittishness rather than a second event -- much the way my dog would.
Regarding the pre-spike options action, I have also noticed that someone out there has uncanny timing -- or perhaps it's a different entity that gets lucky each time around.
I track VIX futures volume (to no avail, so far) more than VIX options volume, but I recognize the need to spend more time with the options. In the interim, a good source of VIX options block trades is Options Doggy -- and you can search his site for past reports of VIX blocks with Technorati here.
Good questions; I hope this helps.
Cheers,
-Bill
the open interest on the June and July 20s is somewhat eyeopening
Bill,
Not sure if you've looked at GARCH models but they have a long term, medium term, and short term component. I believe the short term component supports the concept of an echo as well. Just posted a summary on VIX futures at BetaData as well.
Regards,
JC
Thanks for the suggestion JC. GARCH is something that I was not familiar with until I started this blog; for now it is a placeholder for future research. I think I may get MATLAB and play around with it, but I'm wary of opening too many Pandora's boxes in a short time frame.
Cheers,
-Bill
Post a Comment