Skip to content

S&P500 and VIX Consensus at Extremes

February 5, 2010

While catching up on my nightly reading I noticed the following post by Frank of Trading the Odds:  In this post, Frank presents his standard approach to measuring the edge of a given setup. He does a very good job of crunching the numbers for all to see on a daily basis. In this case, the VIX surging over 20% in one day was the setup described. In general the analysis is very thorough, and the new additions to the setup evaluation framework that Frank recently built in are definitely a good improvement. That said, long-time readers know that I prefer not to use fixed percentage trading rules etc since they are not normalized within the context of a particular environment. 

 Another concern, was whether looking at the VIX vs the S&P500 directly was redundant. Generally speaking, I have found little value in using the VIX as an overbought/oversold indicator to trade the S&P500  in lieu of using the S&P500 directly (however the divergences between the two are very valuable).  In this case, my central question was whether there was information that could be derived from a consensus between the S&P500 and VIX at extremes. Could looking at both together produce superior information concerning a short-term bottom than either in isolation? To run this test I used the last 5000 bars of data on both the S&P500 index and the VIX  from Yahoo Finance. I took the one day returns for the VIX and normalized them using a 50-day percentile ranking, I did the same for the S&P500–only I subtracted the final number from 1 to account for the negative correlation between the two. Thus the worst days in the S&P500 were 90th percentile and higher using this method. Similarly, the biggest up days for the VIX were also represented by high percentile values. Lets look at the numbers:

  Next Day    max tail loss
  Return W% (worst next day ret)
Both>99th 0.88% 86.36% -1.48%
Both>95th 0.50% 65.93% -2.70%
Both>90th 0.27% 60.00% -5.74%
VIX>99th 0.25% 69.00% -6.87%
VIX>95th 0.20% 58.66% -6.87%
VIX>90th 0.12% 56.19% -6.87%
S&P500>99th 0.62% 68.06% -3.29%
S&P500>95th 0.17% 58.89% -6.71%
S&P500>90th 0.13% 56.73% -6.80%

As you can clearly see, consensus indeed helps to confirm temporary market bottoms (at least for a 1-day hold). Combining both screens produced higher next day returns, higher winning percentages and lower risk of an adverse loss. However, while not shown, this “consensus” method does not help add any value outside of the extremes. Apparently, bottoms require BOTH a major spike in the fear gauge and a major loss for the S&P500 which in hindsight makes sense.

8 Comments leave one →
  1. WTP permalink
    February 5, 2010 8:44 am

    Once again you have added unique value by seeing what others do not; thank you!
    Did you happen to test the extremes in the other direction for a short term top?

    • david varadi permalink*
      February 5, 2010 10:52 am

      wtp, thanks very much for the kind words. for shorts you can use the 95th percentile indeed and this also works well to adjust your size since you are concerned about the underlying moving to the upside—is this what you were referring too?

    • david varadi permalink*
      February 5, 2010 10:58 am

      oops, responding to the wrong post—-no WTP I did not, but plan to test that–thank you very much for pointing that out.


  2. WTP permalink
    February 5, 2010 10:57 am

    Yes, shorting/reducing long side for S&P up & VIX down. Thanks

  3. david varadi permalink*
    February 5, 2010 10:59 am

    was referring in my previous reply to DVar position sizing. I will test this out later today.

  4. kostas permalink
    February 8, 2010 12:50 am

    Hi DV, something in the summary stats looks to me off… have you posted the spreadsheet anywhere to give it a second look?
    Good work!

  5. kostas permalink
    February 8, 2010 12:51 am

    Ooops, sorry DV and all–I was referring to the spreadsheet of the position sizing. I will repost the comment there. Sorry…

  6. Josh permalink
    June 9, 2010 7:26 pm

    Hi David,

    How many instances of a 99+ VIX without a 99+ S&P occurred or vice versa? I always have doubts about any new VIX based indicator’s ability to add info above and beyond a price-derived indicator. It seems (have yet to quantitatively test it) that even divergences are already revealed by your standard RSI or Rate of Change divergence.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: