Does This VIX Signal Indicate Trouble Ahead?

The VXV/VIX ratio is over 1.2, but that's no reason to panic

Apr 16, 2015 at 8:44 AM
facebook twitter linkedin

Time for investors to go to the mattresses (again)? This, from CNBC:

"Most investors have never heard of the three-month volatility index, which is known as the VXV. But the relationship between the CBOE three-month volatility index and the options exchange's more familiar 30-day volatility index, the VIX, may signal trouble for stocks.

"Like the VIX, the VXV is a measure of expected volatility in the S&P 500 that is computed from the prices of options on the index. However, while the VIX measures expected volatility over the next 30 days, the VXV measures expected volatility over the next 93 days.

"By comparing the two, an important indicator is formed, according to Bank of America Merrill Lynch technician MacNeil Curry.

"In a recent note, Curry wrote that Thursday's and Friday's closes 'above 1.2 in the VXV/VIX ratio is a significant concern. Historically, the market has struggled to hold its gains when this ratio closes above 1.2' …"

So, should we worry? Well, this is surely a complacency gauge. The CBOE S&P 500 3-Month Volatility Index (VXV) will only become high when CBOE Volatility Index (VIX) itself is low. All volatility indexes anticipate some form of mean reversion. The longer you go out in time -- whether it's a future or a vol index itself -- the less it responds to the here and now, and the more it expects a return to the norm. Thus, a 90-day vol index simply reacts slower to a drop in realized vol than a 30-day index.

So when there's a stretch between the two, it simply suggests that realized vol has caved, or is expected to cave soon (maybe Fed news is out, it's a summer Friday, et. al.). It's also likely that the market itself was recently strong, and it shouldn't be a terrible idea to fade that. On the other hand, complacency itself isn't a great sell signal.

But whatever. That's all my subjective opinion. How about we apply some objective numbers?

There's not a great correlation between the VXV/VIX ratio and future S&P 500 ETF Trust (SPY) returns. VXV data goes back to mid-June 2006. Since then the VXV ratio has a 0.06 correlation to the SPY returns over the ensuing 15 trading days, and a 0.05 correlation to the next 30 trading days. That's pretty close to random, with a very slight tilt to a positive correlation. If this measure was a contra tell, it should be a negative correlation, so nothing from nothing on this basis.

But, alas, the question is whether an "extreme" reading of 1.2 gives us any sort of signal. I looked at this in two ways. First, I compared random 15- and 30-day mean and median SPY returns to mean and median returns if you went long any day VXV/VIX closed >1.2. Then, I did a system trade that eliminated overlaps. That is, once the ratio got > 1.2, I didn't include data again until 15 or 30 days passed. That resulted in 27 trades going out 15 days (about three per year) and 20 trades going out 30 days (a little more than two per year). Anyway, here are the results:

Elevated VXV/VIX Returns

It doesn't suggest running for the hills when the ratio gets cheap, but we do see subnormal returns 15 trading days out. On the other hand, a system trade that simply goes long when the ratio goes over 1.2 actually outperforms the market 30 trading days out.

Remember, again, that we're really just picking up instances of VIX that have drifted ahead of the longer-term vol complex, perhaps thanks to a market pop. So maybe just stepping aside for a couple of weeks and then jumping back in is the best idea of all.

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

JUST RELEASED: The Complete Guide to Earnings Season Profits



Special Offers from Schaeffer's Trading Partners