Skip to main content

The Surreal Nature of Volatility




Even a lot of people who care nothing about art would recognize “The Persistence of Memory” by Salvador Dali. It is the tiny surrealist gem with the melting watches. It is an undoubted masterpiece and shows how great Dali was before he decided being a celebrity was easier, but that isn’t directly relevant here.

Volatility also shows persistence of memory. That is the basis of the GARCH family of models. In the short term volatility persists. And right now volatility is low. On March 30th, 2017 the VIX closed at 11.54. To put that in context, this is in the lowest 6th percentile since the VIX was first calculated at the start of 1990.

Based on this number it might seem like a good time to buy volatility, either through options or VIX futures. There are two problems with this idea.

The first is carry. If you buy VIX futures you pay the basis. As of the 30th of March, the April futures are 1.26 points above the cash index. This means that if the VIX stays where it is and you are long the April futures, you will lose 10.9% in the next 20 days, or a startling 200% per year.

And options don’t help. If you buy options you will pay implied volatility, but get paid out on the basis of realized volatility. And when implied volatility is low, the realized volatility tends to be even lower.

The second problem is persistence. Just because volatility is low doesn’t mean it won’t stay low. In fact the evidence suggests it will.

Here are the implied and realized volatilities of the S&P 500 for each year since 1990.

Year
Implied Volatility (VIX)
Realized Volatility
Volatility Premium (Points)
Volatility Premium (% of VIX)
1990
23.0
15.9
7.1
31%
1991
18.4
14.3
4.1
23%
1992
15.4
9.7
5.7
37%
1993
12.8
8.7
4.1
32%
1994
13.9
9.9
4.0
29%
1995
12.4
7.8
4.6
37%
1996
16.5
11.8
4.7
28%
1997
22.4
18.1
4.3
19%
1998
25.7
20.2
5.5
21%
1999
24.4
18.5
5.9
24%
2000
23.4
22.6
0.8
3%
2001
25.6
20.5
5.1
20%
2002
27.4
26.3
1.1
4%
2003
22.2
17.1
5.1
23%
2004
15.5
11.0
4.5
28%
2005
12.8
10.2
2.6
20%
2006
12.8
9.5
3.3
26%
2007
17.5
15.8
1.7
8%
2008
32.7
35.6
-2.9
-10%
2009
31.5
24.2
7.3
26%
2010
22.5
16.5
6.0
26%
2011
24.2
21.1
3.1
12%
2012
17.8
12.6
5.2
29%
2013
14.2
10.8
3.4
24%
2014
14.2
11.1
3.1
21%
2015
16.7
14.9
1.8
8%
2016
15.9
11.6
4.3
28%

Obviously what follows isn’t a rigorous statistical argument, but if we look at the 5 years with lowest average VIX, we also see an average volatility premium of 30%. This compares to an overall average of 21%. Based on this, low volatility is a sale. When implied volatility is low the subsequent realized volatility is even lower.

Also, look at the period 1993 to 1995. The VIX averaged 13 for 3 years. Then between 2004 and 2006 the VIX averaged 13.7. While volatility was low, it stayed low for a long time. The converse is also true. High volatility also tends to stay high.


Volatility reverts to its long-term mean. This has been well accepted and internalized by most traders. But volatility also persists, and the persistence can dominate for years. I used to sign my books, “buy low, sell high”. I was wrong. When it comes to volatility; sell low, buy high.

Comments