A Lesson in Correlation

Correlation between the SPX and its components has been relatively low of late -- but will this last?

by Adam Warner

Published on Oct 28, 2015 at 9:16 AM
Updated on Jun 24, 2020 at 10:16 AM

One important thing to remember about the CBOE Volatility Index (VIX) is that all VIXes are not created equal. By that I mean today's 15 VIX isn't necessarily the same as yesterday's 15 VIX.

VIX proxies implied volatility on the S&P 500 Index (SPX). Any implied volatility measure of an index has two base components. One is the implied volatility of the component stocks. The higher the implied vol of the components, the higher the implied vol of the index -- that's pretty self-explanatory. Two is the degree to which the components are correlating with one another. If the stocks are all moving in different directions and magnitudes, then the moves will offset each other and the index itself will not be that volatile.

I bring this up because, well, VIX has tanked lately. And, per Michael Khouw, it looks like we can blame "Factor 2":

"The one-month implied correlation is below 30%, well below the average of over 40% we've seen over the past couple years. What that is telling us is that the options market believes we are in a stock picker's market. Rather than simply buying an index, one who chooses individual longs (and shorts, if you make bearish bets too) wisely may find the coming quarter to be more rewarding. To me it also suggests that index portfolio hedges, using things like SPY or QQQ put spreads are reasonably priced and individual stocks are likely to provide more attractive opportunities for buy-writes and naked put sales."

Fund guys on TV tend to call everything a stock picker's market, but there's a bit of "talking your book" there. If it's not a stock picker's market, then everyone can just index, and they don't provide much of anything resembling a value-add.

In the options world, it has meaningful implications. Larger players effectively always have positions that bet on dispersion. If you're net long gamma in individual names and short gamma in indexes, you are betting on dispersion. You want low correlation and relatively low index volatility. Best case is every option in your portfolio goes bananas, but in an uncorrelated way to all the others. You then hedge or ride your long gamma in the individual names and give back a little in your index vol short, and party like it's 1999 (pretty much the golden age for this sort of trade).

At times of high correlation, the dispersion trade doesn't work so well. The underlying index itself moves too far directionally.

If you expect correlation to remain low, then what he says above is correct. Options-selling strategies in indexes will look relatively attractive. But will this correlation last? My friend and fellow Mets fan Elliot Turner has a good question.

Correlation is imperfect to compare over time. The CBOE has three rotating correlation indexes. Short answer is, the duration isn't constant. Long answer is, it's going to take a full post to explain. For the purposes of today's topic, it's a good observation -- you'd think earnings season would lead to lower correlation. But that hasn't really been the case; there's no noteworthy correlation dips at peak earnings time. Perhaps that's because these are measures of implied volatilty (anticipation) rather than the direction and magnitude the individual names ultimately move. Thus, I do think the lower anticipated correlation goes beyond earnings season.

Disclaimer: Mr. Warner's opinions expressed above do not necessarily represent the views of Schaeffer's Investment Research.

A Schaeffer's 39th Anniversary Exclusive!

8 Top Stock Picks for 2020

Access your FREE insider report before it's too late!