Monday, January 16, 2012

S&P 500 / VIX Matrix

As a follow up to last week's VIX as a Predictor of Equity Returns and Model Building / Data Mining posts, I put together the following two 'VIX Matrix' tables. These tables show the:
  • One month forward return of the S&P 500
  • One month change in the VIX index
against a number of scenarios involving the one month month change in the VIX and the absolute value of the VIX since its 1990 inception.
  • S&P 500: Less than -1.0% = Red, -1.0% to 1.0% = Yellow, Greater than 1.0% = Green
  • VIX: Less than -2.5% = Red, -2.5% to 2.5% = Yellow, Greater than 2.5% = Green
Lots of interesting information in these tables that I won't bother summarizing (look at it yourself), except to say that even in these turbulent times (bull market 90's, roller coaster 00's), markets were (on average) very mean reverting.

Note that there are plenty of limitations to these tables, most of which involve the limited data points for a number of the cells.


One-Month Forward S&P 500 Performance




One-Month Forward VIX Change



Update: much more on the above from my friend Bill over at Vix and More blog.

5 comments:

A different Tom said...

You may have a typo. Both charts have "One month change in VIX" as the top (left to right) axis, but the title of the lower one seems to imply that the data on that axis is really S&P 500 change? (and the colors represent the VIX change?)

Jake said...

The numbers in the actual tables (not the header) are the change in the S&P 500 (table 1) and the VIX (table 2). The headers (VIX level and change in VIX) are the same in both.

Bill Luby said...

A worthy successor to the earlier effort, Jake.

FYI, I elaborate a little on your post in EconomPic Data’s VIX and S&P 500 Performance Matrices, which I just posted.

http://vixandmore.blogspot.com/2012/01/econompic-datas-vix-and-s-500.html


Cheers,

-Bill

Namazu said...

In service of the perpetual quest to find cheap and reliable hedging vehicles, I'd be curious to see a third table: (change in SPX/change in VIX) against VIX--curious enough to run the numbers myself, given access to the underlying data set. Of course, depending on number of observations, it may not tell us much more than these two. Welcome back!

Jake said...

Namazu- if you check this, here you go: http://i.imgur.com/txwHz.png I'll post more broadly soon.

Share via Twitter

Facebook Share