Performance Implications of VIX and SPX Divergences

I have frequently posted about the issue of SPX and VIX correlations, but some of the comments following yesterday’s SPX Straddle Case Study Update indicate that it is time to dust off some statistics and address some of the issues that have been raised.

Reader Nirvanic lays out his thinking on the subject as follows:

“What I call a VIX/SPX non-confirmation, meaning that both the SPX and VIX were red. One of my main rules is that the VIX and the SPX cannot be the same color. When this happens there is something wrong. If the VIX and the SPX are green, then usually the SPX will go red and the same in reverse.

I have only seen a few days lately where there was a VIX/SPX non-confirmation. They are rare and so I don't have a lot of data on these. In the occurrences I have seen, the VIX has been the 'tell.' If this holds true, then tomorrow should be a green day because the VIX was right, not the SPX.”

Let me start out by saying that one of the my most read posts of 2007 was titled High Positive Correlation Between VIX and SPX Often Signals Market Weakness. The post is probably worth checking out just because it has one of my favorite photos on the blog, but of greater interest is likely one of my conclusions that “in looking at past periods of high positive correlations between the SPX and the VIX, it is notable that the SPX generally performs well below its historical mean for up to three months following the high positive correlation period.”

Nirvanic’s contention is more specific. His “non-confirmation” days are probably a lot more common that he realizes. Going back to 1990, when the VIX ends the day in the red, the SPX also closes down about 21% of the time. When the VIX is green, non-confirmation is even higher, at about 25%.

In terms of performance, it is easy to think of the possibilities as a 2x2 matrix, with the VIX either up or down on a daily basis and the SPX negatively correlated or positively correlated with the VIX.

Let me summarize the considerable research I have done on the subject by saying that if there is one pattern among the four that is most predictive of above average future returns, it is the one in which the SPX finishes the day in the red while the VIX finishes in the green. Conversely, as Nirvanic suggests, the pattern associated with below average future returns is the one in which the SPX is up and the VIX is up.

Each of the two remaining patterns demonstrates very little variation from typical market performance; in fact the SPX down, VIX down pattern slightly outperforms the SPX up, VIX down pattern.

Note that all data cited above reflect single day changes in the SPX and VIX. At some point in the future I will share some of my research into longer-term changes in the SPX and VIX patterns, as well as their trading implications.