August Accident Aside, Overbought VIX Trade Still Works

The CBOE Volatility Index (VIX) has been overbought four times in the past three months

Adam Warner
Oct 5, 2015 at 10:00 AM
facebook twitter linkedin


What if we had a CBOE Volatility Index (VIX) spike and nobody showed up? That's what seemed to happen last week, when VIX actually nudged into overbought territory on Monday, as Mr. Market made a lower low. And then a funny thing happened: We rallied. And then rallied some more. All told, the SPDR S&P 500 ETF (SPY) popped 1.1% on the week, which doesn't sound like a lot ... until you consider we started the week down 2.57%.

VIX closed the week down 11.1%, which in a vacuum sounds like a large move vs. what was ultimately just a 1.1% market pop. It sounds like an especially large VIX drop when you consider how we arrived at that net return; Friday alone had a 2.73% move.

On the other hand, the news is out.*

*Jobs number makes and misses are always misleadingly expressed, IMHO. The estimate was about 200K jobs created, while the actual was 142K, a miss of 58K. If you use a basis of 200K, then it missed by 29%, wow! If you use a basis of the 140 million or so jobs out there, it's more like a tiny rounding error. Throw in that the actual number is really an estimate adjusted for seasonality and many other things, and it really strikes me that it's kind of "whatever." As it almost always is, beat or miss. The trend is, of course, important, as are revisions to prior numbers. But that only serves to remind us that this number will itself get revised a couple times. But I really digress.

Were we really that nervous about this jobs report? I doubt it, though I suppose it's one data point out of the way and the market basically held the August lows so far.

Which brings us back to VIX. I count VIX as overbought when it closes 20% above its 10-day simple moving average. I declare it a separate incident once VIX has had a close back below its 10-day. I score SPY returns one month and three months out, as well as out until VIX closes below its 10-day. On average, that takes about 5.5 trading days. Anyway ...

151005Warner1

We had some clusters of overbought VIX last year: three in a three-month period, achieved twice. That's a lot, considering it usually happens about three to four times a year.

Well guess what, we're even more clustered now. This was the fourth overbought VIX in three months. That sounds ominous -- although so far, so good. It took only three days to "correct," and it came with a 2.26% "win." Of course, that's coming off a disaster of a system trade. If you went long Aug. 20, you had to hold for nine trading days and saw a nasty 4.22% hit. If you're into 2011 analogs, that's easily the worst trade since an 11-day 7.22% scalping initiated July 29, 2011.

By and large, though, this trade tends to work, even including the recent accident. The median return is 0.99%, compared to a median return of 0.08% if you held for any random five-day period. As you can see by the chart, it mostly wins, or loses very small.

Future VIX seems to have settled into a bit of equilibrium:

151005Warner2

December is a holiday month, and it tends to see VIX dip a bit. Thus it's a very flat term structure, thought it's interesting to see an upslope with VIX north of 20.

Not much has changed lately. Here's Friday alongside the term structure from 9/18 and 8/20. In both instances, VIX was in the 20-21 range. The first was right before the market drop.

151005Warner3

Again, not much going on, relatively speaking. Ten-day historical vol is about 20, too, so it's tough to argue with the price of anything right now.

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


A Schaeffer's exclusive!

The Expert's Guide

Access your FREE trading earnings guide for Q3 before it's too late!


  
 

Partnercenter