How to harness the power of VIX to protect your portfolio? (Part 2)
I hope you enjoyed reading Part 1. I had fun writing it and discovering Medium at the same time. Last week, I spent some time fine-tuning the predictive model I want to describe (it will probably be part 4); I’m quite pleased with the results! More on this in a few weeks, for now, let’s go back to VIX basics.
The Cash VIX term structure
In the previous article, I mentioned that VIX (30 days) has siblings, namely VIX9D (9 days), VIX3M (90 days), VIX6M (180 days) and VIX1Y (252 days). When we plot all these indices together we obtain what is known as the Cash VIX term structure. We will look at this curve for 3 dates: 2-Jan-2020, 24-Feb-2020 and 3-Mar-2020.
Under “normal” circumstances, the Cash VIX term structure is upward sloping (i.e. indices further into the future have a higher value). Indeed, as uncertainty increases with time, option writers demand higher compensation when the option expiry date is further in the future. This situation is called contango (it is a term that originally comes from Commodities Futures where Futures were invented).
The graph of the Cash VIX term structure as of 2 Jan 2020 shows a perfect contango (remember, at that date, very few of us had heard of COVID19 and even fewer could imagine the consequences). We have VIX/VIX3M = 0.82
As of 24 Feb 2020, we can observe 2 major changes. The level of these VIX indices is much higher and the term structure is now downward sloping; this is called Backwardation. At this stage, the ratio VIX/VIX3M has already reached 1.14. Finally, on 3rd Mar 2020, the term structure is still in backwardation, the ratio VIX/VIX3M has reached 1.20 and the overall level is even higher than on 24 Feb.
Multiple indicators can be extracted from this Cash VIX term structure. I like to compare VIX9D with VIX, VIX with VIX3M, VIX3M with VIX6M. Let us observe the distribution of the ratio VIX/VIX3M which is often called the VRatio.
This ratio is lower than 1 in contango and greater than 1 in backwardation. Looking at its distribution, we can see that this ratio is positive most of the time (roughly 90% of the time) this illustrates the fact that most of the time the market is in contango.
The VIX Futures curve
The CBOE has also created futures contracts on the VIX (VIX Futures). These contracts allow investors to position themselves up or down on the VIX index. There are many VIX Futures. The most liquid contracts are the monthly VIX Futures that expire each month (usually the 3rd Wednesday of the month). The price of these Futures gives us visibility on the level of volatility (VIX) that investors expect in the coming months. All of these prices for the coming months are called the VIX Futures term structure.
As we have seen with the cash VIX term structure, under “normal” circumstances, the VIX Futures curve is in contango. The price of the first month (commonly known as VX1) is lower than the price of the second month (known as VX2) and so on.
The graph of the “VIX term structure” as of January 2, 2020 (below) shows a perfect contango.
When markets are turbulent, the term structure is inverted. The price of VX1 is higher than the price of VX2, the price of VX2 is higher than the price of VX3, and so on. The graph of the “VIX term structure” as of March 2, 2020 (only 2 months after the previous graph) shows a perfect backwardation which shows that investors have radically reassessed their market outlook.
But the most interesting thing is that this term structure went into backwardation as early as 24 February 2020. It is certainly not before the start of the crash that we could observe, but let’s acknowledge that many investors would have been happy to cut their position at that time. Remember: the VIX measures the implied volatility of 30-day options. It is this measure of market anticipation that interests us.
A lot of information can be extracted from this term structure. One of the best-known measures is the VIX Future “contango” VX1:VX2 which is calculated as follows: VX1:VX2 = VX2/VX1–1. This percentage is positive in contango and negative in backwardation. Looking at its distribution, we can see at first sight that this ratio is positive most of the time (more than 80% of the time) this illustrates the fact that most of the time the market is in contango.
In the next article, I’m going to describe a toy model that uses some of these features and share some historical simulation.
