Stocks quoted in this article:
There was a spirited debate between the "Fast Money" halftime show guys the other day regarding the value of the CBOE Volatility Index (VIX). In one corner, we have Pete Najarian:
And in the other corner, we have (Downtown) Josh Brown:
OK, it's not exactly something out of "First Take." And we can all be thankful for that. But who's right?
Well, it's more gray than black-and-white. But I'm going to lean towards Josh's side here.
I mean, Pete's right that we shouldn't dismiss VIX indications. After all, if we could, well -- what am I doing writing about it all the time? There's certainly meaning wrapped in our numbers here.
But the numbers do suggest VIX does a better job telling you what already happened than what's about to happen next. That is, it's more coincident than predictive.
I ran the numbers, going back a decade and change, to the beginning of August 2004. That covers just about any sort of market, and pretty much the highest and lowest VIXes since the 1987 crash ... when we didn't actually have a VIX, and have to back-calculate it.
Over the past decade, VIX itself has a very strong negative 0.75 correlation to the SPDR S&P 500 ETF Trust (SPY), which does certainly suggest that it performs reasonably close to how an inverse exchange-traded fund (ETF) might act. But VIX is a volatility measure, so it's more important to see how it predicts actual volatility.
So, I took standard deviation measures on SPY from close to close. It's not a perfect proxy for realized volatility, but it's close enough for these purposes, where we're just looking for the general relationship.
How does VIX "predict" going forward? Reasonably well. Over the last decade, VIX had a 0.59 correlation to 10-day forward standard deviation in SPY. But it "predicted" the past even better. It had a 0.71 correlation to 10-day backward SPY standard deviation.
But VIX proxies 30-calendar-day volatility, you say? OK, so let's look at that.
VIX had a 0.42 correlation with 21-trading-day (about 30 calendar days) forward SPY standard deviation. And that pales when compared to the correlation of backwards standard deviation over 21 trading days, which came out to 0.729.
I think we have a verdict here. It strongly suggests VIX is more coincident than predictive.
But again, that doesn't mean Pete's wrong. VIX absolutely has times when it adds value. I post every three minutes or so on overbought VIXes and what they foretell for future moves in SPY, or the iPath S&P 500 VIX Short-Term Futures ETN (VXX), or whatever. I believe that sort of use for VIX is always operational, for example.
It's probably safe to say that the majority of the time VIX tells you something you already know, but there are enough divergences and extremes to give you info that's helpful. The trick is finding those times when VIX adds to the story, as opposed to just rehashing it.
Disclaimer: Mr. Warner's opinions expressed above do not necessarily represent the views of Schaeffer's Investment Research.