Yesterday’s sharp downward move in the VIX gave me a reason to tweet that the volatility crush
as seen in the SPX and VIX was among the top 25 in history. Upward pressure on the VIX toward the end of the session dropped the VIX down to the 30th
largest VIX decline in history, but along the way the Twitterati raised a number of questions about volatility crushes and the VIX as a measurement tool of broad market volatility crushes.
Since I have never seen any data related to the VIX and volatility crushes before, I thought this might be an opportunity to present some of my data and talk about the findings. In the chart below, I have captured the 25 largest one-day declines in the VIX since 1990 and have presented data showing the forward performance of the SPX in periods from one to 100 days and I have also added some brief commentary regarding the causes. In some cases I link the volatility crush to a previous VIX spike and use some explanatory shorthand in the process. For instance, the top crush day, May 10, 2010, followed 2 days after the 21st
largest VIX spike
[source(s): CBOE, VIX and More]
Of course, most of these volatility crush days coincided with huge jumps in the SPX, but there are some interesting exceptions, not the least of which was the 0.04% decline
in the SPX back on April 11, 1990. That just happens to be the only day for which I cannot find any obvious explanatory catalyst – to the extent that a 0.04% in stocks can actually have a catalyst – but given the proximity to the upcoming Gulf War, my guess is that some sort of news related to Iraq played into this event.
Note also how many of these volatility crush instances follow other important market-moving high fear
events one or two days later, in true mean reversion
fashion. Examples on this list range from the flash crash
, Lehman Brothers and Bear Stearns to several VIX all-time highs, Russian
political and economic crises, the Boston marathon bombing, etc.
Things get even more interesting if you compare the top 25 VIX crushes to the top 25 VIX spikes in history (for an identical table recapping the top 25 spikes refer to Last Two Days Are #5 and #6 One-Day VIX Spikes in History
.) For starters, the top 25 VIX spikes all move up at least 31%, while none of the top 25 VIX crushes managed to eclipse the 30% decline level. Also note the differences in the mean reversion predictive value of spikes versus crushes. In general, the performance of the SPX following VIX crushes is modestly lower than that of the SPX in general. On the other hand, the performance of the SPX following VIX spikes is generally better than the SPX in general – much more so if the September 29, 2008 outlier is dropped from the data set.
Another point that I think is worth making speaks to the overall changes in the volatility space. The critical data point is that 11/25 of the top 25 VIX crushes happened since the beginning 2010, while 14/25 of the top 25 VIX spikes have occurred during the same period. This means that we have had as much in the way of big volatility moves in the list seven years as in the previous twenty years of VIX data. In other words, the volatility landscape is changing and the rise of VIX futures
and VIX ETPs
are no doubt an important part of that change.
For those who may be interested, you can always follow me on Twitter at @VIXandMore
Disclosure(s): the CBOE is an advertiser on VIX and More