The correlation between VIX and SPY is high at the moment
Even though the CBOE Volatility Index (VIX) remained in the teens throughout the recent Sino-Grecian Scare Fest, the interday and intraday moves were on the violent side. And I know we always say "this time was different," but indeed, this time was a little different. This, from Bloomberg.
"Since April, the VIX has jumped an average of 17 percent on days when the S&P 500 fell 1 percent or more, compared with 12 percent on average for similar daily moves since the start of the bull market, according to Bloomberg data.
" ... measures of correlation between the S&P 500 and VIX also point to "near-panic conditions" in the market, Sundial Capital Research Inc.'s Jason Goepfert wrote in a July 1 note. The two gauges are trading the most inverse to each other since December 2011, Bloomberg data show."
That correlation stat intrigued me a bit. I checked and, of course, Jason and Bloomberg are correct. I calculated a rolling 20-day correlation between the day-over-day VIX moves vs. the day-over-day SPDR S&P 500 ETF Trust (SPY) moves going back to July 2009, and here's how it looks:
We're at negative 0.97 right now, which needless to say is about as low (really high) as it gets. In other words, we're in nearly perfect inverse correlation. Over the entire sample, VIX and SPY have a negative 0.80 correlation, so we're indeed detached from the norm.
Incidentally, those blips "up" toward zero are almost all near the end of calendar years, meaning it's most frequently just a time when the market's not moving much and VIX is correlating more to the pain of paying time decay than to actual volatility in the marketplace. But I digress.
The most important question is whether periods of increased correlation are at all predictive of future market action. And best I can tell, the answer is no.
At the risk of going all "second derivative," I compared the backward-looking, rolling 20-day VIX-SPY correlations with 20-day forward-looking SPY returns. And there was virtually no relationship -- they have a correlation of 0.013. And here's a scatter plot of the data:
That's truly a chart about nothing. So, it is kind of interesting to see the recent spike in emotions. Unfortunately, it doesn't yield us a whole lot in terms of clues going forward.
Disclaimer: Mr. Warner's opinions expressed above do not necessarily represent the views of Schaeffer's Investment Research.