When people talk
about the VIX you often hear them refer to mean
reversion, which refers to the tendency of the VIX to be pulled inexorably
in the direction of its long-term mean.
With 27 years of data from the CBOE (including some historically
reconstructed data), it is possible to calculate the lifetime VIX mean, which
happens to be 19.71 at the present.
As an options
trader, however, I am wary of giving too much weight to outliers when it comes
to predicting the most likely outcome in another options expiration cycle or
two. For this reason, I am more
interested in knowing the lifetime VIX median, which is only 17.84. The median is the 50th percentile while
the mean just happens to be in the 60th percentile. The discrepancy is due to the fact that VIX
values are not normally distributed. Instead,
VIX values exhibit a positive
skew (a topic for a future post), due to the fact that there are a handful
of VIX historical extremes in the 50s, 60s, 70s and 80s. Meanwhile, the middle 50% of VIX values (the
25th to 75th percentiles) range from 14.04 to 23.98.
So…if the VIX
median is so important, why is it that we never hear about it or about median
reversion? Good question. I touched on that subject on “Drilling Down on VIX Mean Reversion” in
the January 2013 issue of Expiring
Monthly: The Option Traders Journal. As I see it, anyone who is focusing on means
in a skewed distribution is necessarily assuming a normal distribution and statistics
related to normal distributions when no such distribution or relevant
statistical analysis exists.
I mention all of
this because yesterday was one of those periodic bursts of activity for me on
Twitter. In some Twitter conversations,
we discussed the median VIX vs. the mean VIX and there was a request for a
chart of a five-year moving average of the median VIX. Since I have never seen such a chart – or any
VIX median chart for that matter – I present below a chart of the five-year
moving average of the median VIX, using data going back to 1990.
[source(s):
VIX and More, CBOE]
Note that the
current five-year median VIX is 15.07, while the five-year mean VIX is
16.63. For the full history of the VIX,
going back to 1990, the lifetime median VIX is 17.84, 9.5% below the lifetime
mean VIX of 19.71. What does all this
mean? Mostly that one should be careful
using statistics that are associated with a normal distribution when analyzing
the VIX. Perhaps more importantly, VIX
traders should also think at least as much about median reversion as mean
reversion.
As an aside,
while I have not been active on the VIX and More blog as of late (this is about
to change soon, starting today), I have been active in various other media
incarnations. Last Friday, for instance,
I was a guest on the Volatility
Views weekly podcast hosted by Mark Longo of The Options Insider. Tomorrow at 2:00 ET, I will be a speaker on a
webinar, Trading
VIX to Hedge Market Risks: What You Need to Know, with Tom Lydon of ETF Trends, Greg King of REX Shares and Vinit Srivastava of S&P Dow Jones Indices. On the print side, this Saturday I will also be
a guest columnist at Barron’s, pinch
hitting for Steve Sears. Last but not
least, if you wish to follow me on Twitter, where I have been active for ten (!)
years, you can find me at @VIXandMore.
Related posts:
- The Biggest VIX Spike Ever: A Retrospective
- Low Volatility: How To Profit From a Quiet VIX (Guest Columnist at Barron’s)
- Today’s 34% VIX Spike and What to Expect Going Forward
- Availability Bias and Disaster Imprinting
- The VIX:VXV Ratio, Availability Bias and Disaster Imprinting
- VIX Data to Support Availability Bias and Disaster Imprinting Hypothesis
- VIX Unspikes as Stocks Rebound
- VIX Punches Through Upper Bollinger Band
- Mean Reversion After Big Drops
- When Is Echo Volatility Safely Behind Us?
- On Short-Term VIX Mean Reversion and Echo Volatility
- VIX Sets Some New Records, Suggesting Volatility Near Peak
- A Conceptual Framework for Volatility Events
Disclosure(s): the CBOE is an advertiser on VIX and More