The graph below shows the % of days VIX futures were contangoed at various VIX readings since early 2004. Not sure why this is important? Read more after the graph.
I’ve shown two interpretations of VIX contango: (a) the VIX index is below the front month VIX futures contract (blue), and (b) the front month VIX futures contract is below the second month contract (grey) (1).
At lower VIX readings (below ~25), futures are contangoed the vast majority of the time.
Why is contango important?
A chain of “volatility risk premiums” tends to exist when futures are contangoed (see the numbers). VIX futures will tend to overestimate the actual realized VIX, and the spot VIX will tend to overestimate actual realized volatility. Click the cheat sheet at right to zoom.
Because VIX ETPs are priced based on VIX futures, this chain of VRPs puts the wind at our back when we’re short the VIX (ex. long XIV or short VXX) as VIX futures tend to slide down towards the spot VIX as they approach expiration.
This isn’t a guarantee that we won’t lose money mind you, it’s only a tailwind. So when VIX futures are contangoed, it doesn’t necessarily tell us which direction we should trade, but it does tell us that a short VIX position (ex. long XIV or short VXX) is the path of least resistance, in terms of the futures term-structure.
What the first graph showed us is that the percentage of time that futures are contangoed is much higher at lower VIX readings, and vice-versa. As the VIX increases beyond ~25, VIX futures are considerably more likely to flip and become backwardated, losing the tailwind of contango (2).
Click to see Volatility Made Simple’s own elegant solution to the VIX ETP puzzle.
Volatility Made Simple
1. I think a more sound way to approach this analysis would have been to compare the VIX index to the 30-day constant maturity price of VIX futures. Had we done this, the results would have been very similar to the front vs second month VIX futures numbers. I didn’t do this because I didn’t want to make an already confusing topic even more so for readers who are new to the subject.
2. …but increasing the potential for mean-reversion. A topic for another day.