There are two forces acting on VIX ETPs that can be modeled quantitatively and captured by traders.
The first is the impact of the term-structure/volatility risk premium that we spend so much time on this blog talking about. Most of the simple strategies we’ve backtested on this blog are attempting to capture this effect, which we’ve broken down statistically here.
The second, and the subject of this post, is mean-reversion. We haven’t shown many strategies that take advantage of mean-reversion because it’s much harder to capture with VIX ETPs and futures. The VIX Index itself is highly mean-reverting, but as the next three graphs illustrate, much of that MR has been traded out of VIX products.
In the first graph, I’ve shown the results of two simple opposing strategies: go long the VIX index at the close when the VIX is either below its 7-day moving average (orange, i.e. mean-reversion), or above (blue, i.e. momentum).
We can’t actually trade the VIX index (it’s a mathematical calculation, not a tradable asset), but it proves the point nicely. The VIX index is highly mean-reverting as evidenced by the ultra-smooth positive “returns” of the orange line, and negative returns of the blue line.
But when we attempt to apply this same strategy to VIX futures using the same entry and exit points as graph #1, it’s much less effective. Here I’ve applied the exact same trades to the 30-day constant maturity price of VIX futures (also a mathematical calculation, not a tradable asset).
Over the last 4 years, mean-reversion (in the very narrow way I’ve defined it here) has returned to VIX futures, but it hasn’t been nearly as consistent as we saw with the VIX index. Presumably because mean-reversion in the VIX is so predictable, most of it has been traded out of VIX futures.
But that all really misses the key question: does mean-reversion exist in things we can actually trade? Here I’ve again applied the same trades from graph #1 to the VIX ETP VXX.
Both portfolios, mean-reversion (orange) and momentum (blue), have consistently gone down, down, down.
Why? Because the impact of the term-structure/volatility risk premium, or the tendency of VIX futures to overestimate the future VIX index and slide down to meet the index as they approach expiration, is stronger than the impact of mean-reversion, so mean-reversion effectively gets washed out of the results.
My point is not that mean-reversion doesn’t exist in VIX ETPs and futures. Mean-reversion based on very overstretched prices (ex. TM’s RSI(2)) may still be an effective trade, just not one that presents itself very often because, by its nature, it’s uncommon.
Lastly, note that the two effects discussed in this post, (a) term-structure/volatility risk premium and (b) mean-reversion, generally run counter to one another. For example, during significant VIX spikes, mean-reversion favors short VIX positions, but at the same time, those significant VIX spikes tend to be accompanied by backwardation and negative VRP, favoring long VIX positions. Managing these opposing forces is important to trading VIX products.
Click to see Volatility Made Simple’s own elegant solution to the VIX ETP puzzle.
Volatility Made Simple
Wonk note: data prior to the launch of VXX has been simulated. We’re able to do this accurately using a combination of the indices and the futures data on which VXX is based. Read more about simulating data for VIX ETPs.