Did it feel like the VIX was up 45% in October? It sure didn’t look like it when reviewing various “Long Vol” funds across the hedge fund space or watching the VIX futures day in and day out. Not to mention the stock market falling just little more than -2%. Which brings us to the question of just what does the VIX tell us about the volatility such programs are trying to capture.
The VIX, it’s relationship to volatility, and its use as a good benchmark for when to expect long volatility trading programs to see gains or losses is incredibly complex. In 8 out of 10 cases, it would be fair to expect the long vol investments to do well when the VIX appreciates more than +10%.
Unfortunately, we could write a small book or PhD thesis on that 9th and 10th case. Consider the October performance of various investable VIX products vs. the VIX itself in October. You had the VIX printing +44%, and the index of mid term (3-6mo) futures prices only up 1.3%, with none of the investable products above 7%.
Why the big discrepancies?
Well, we all use the VIX Index number that you see on CNBC or other financial media as a shorthand for what is happening in the volatility space. Problem is, the VIX is a measurement of the estimate of option prices over the next 30 days, but there are options ranging everywhere from the next day to next 500 days. And indeed, the creator of the VIX, the CBOE, also has indices on option prices for 9, 93, 180, and 1 year prices – but we’ve all sort of decided as a financial society to just quote the VIX as the defacto volatility gauge.
There’s also the need to reflect what is happening across many different option strike prices, which the VIX does by sampling between 175 and 200 option prices (you can see the listing here if you’re interested). But, again, it is using a subset of all available options.
Now, most of the time it does a pretty good job of reflecting what is happening across all these different time frames and market levels. But every now and then the timing (and/or the pricing) matters. Like when there is a potentially world changing event like the US election smack dab in the middle of that 30 day window.
Vol traders analyze all of this data in what they call the vol surface, which looks something like below and models option pricing across price, time, and volatility.
Any time you see the VIX spike and few, if any, VIX traders or products with a similar type of large gain, what can be happening is the volatility increase being localized in that specific time/money portion of the surface the VIX is calculated off of. Using the chart above, that would be the red portions spiking up, but you can also see portions of the surface going down. In such a setup, the VIX doesn’t do the best job at reflecting what is happening across this entire surface. The volatility surface represents an entire landscape of volatility prices (and expectations). The VIX, on the other hand, is representative of just one slice at one point in time.
So why wouldn’t vol traders just trade that portion of the curve which aligns with the VIX? Good question. They analyze these surfaces to pinpoint opportunity, sometimes finding trades where they see a portion of the surface to be underpriced or overpriced. There can be premiums to the options included in the VIX and skew towards Puts and so forth which make them appear more expensive than their probabilities would calculate to. Traders search out those types of discrepancies, but that doesn’t always mean a spike in volatility will show up on that part of the surface they’ve identified.
But more importantly, vol traders don’t do it because the VIX is not a tradable index. You could, in theory, replicate it by buying all the options in the calculation, but it has a constant maturity of 30 days, so it gets complex in a hurry when you consider how exactly you would match that tenor, not to mention the dynamic adding/subtracting of different strikes. To skip that complexity/impossibility, the CBOE launched VIX futures, which became one of the most successful futures contracts of all time.
Problem with those is, they expire, as futures contracts are wont to do. The October VIX futures, for example, expired on October 21st. So, if you wanted to keep exposure to the VIX via futures, you would need to sell the October futures contract before the 21st and buy the November contract. That November contract, however, is essentially a bet on where the VIX index will be on the November contract expiration (Nov. 18), and so forth. If you plot all these VIX futures contracts out in time, you get the VIX futures curve which looks like below from vixcentral.com (as of Oct. 1):
Here again, we’re faced with the truth that there are many sides to the VIX (9 different VIX contracts in this picture at varying prices). What’s more, our sell October and buy November futures example above would have had us selling at 30.6 and buying at 32.4 to maintain exposure, highlighting the VIX futures propensity to lose value at expiration (the closer to the current month), which creates a sort of headwind to buying and rolling futures month after month.
In conclusion, this is hard, multi-dimensional stuff not easily summarized into a tagline that x does y when VIX does z. It’s nuanced and reliant on the factors of time, distance, and volatility.
We highly recommend Vance Harwood’s posts: How does the VIX work? and No, the VIX is Not Broken if you want to go into even greater detail about the VIX and some of the factors that make it a frustrating benchmark for volatility at times – including calendar day annualization, overhead premium, limited predictive power, factoring in mean reversion, sensitivity to Out-of-the-Money Puts, and Monday Effects.
The bottom line of the VIX’s various idiosyncrasies is that the VIX number is a highly imperfect measure of how trading strategies utilizing the VIX or options perform. Whether it reflects their performance depends on where they are on that volatility surface, and whether the entire surface is moving along with the VIX reading.
[note: much of this piece was worked on by Jeff Malec collaboratively with the Mutiny Fund, and may also be include in their materials]