January 23, 2018
This post is a test of Tony Cooper's Contango-Backwardation Roll Yield Volatility Strategy (which will be referred to as "Roll Yield Strategy" going forward). Tony Cooper published the well known whitepaper, "Easy Volatility Investing", in which he describes five volatility trading strategies (among many other volatility insights). The Roll Yield Strategy is extremely straightforward and is similar to a recent strategy we shared which used the VIX/VIX3M ratio.
In the whitepaper, Tony Cooper used available data from 2004-2013. It should be noted that XIV's origination did not occur until November 30, 2010. It is possible to recreate pricing history for XIV and VXX prior to origination using the index they track; however, price action on the products would most likely have been very different had they been actively traded during that period. The Roll Yield Strategy boasted annualized returns (CAGR) of 87.9% with a maximum drawdown of 68.7%. The CAGR is fantastic, but the maximum drawdown is too large for many investors to handle. The test we ran indicates that the Roll Yield Strategy performed much worse than previous history would have suggested.
The parameters for the Roll Yield Strategy are as follows: Go long XIV if the 10 day moving average of VIX3M/VIX >1, else go long VXX. The strategy is dependent upon the fact that a reading of >1 indicates that the VIX futures term structure is in contango and a reading of <1 indicates that the VIX futures term structure is in backwardation. The 10 day moving average is applied for its "smoothing" effect and its ability to reduce the number of trades the strategy takes.
The VIX measures the market's expectation of 30-day volatility, whereas the VIX3M (previously known as VXV) measures the market's expectation of 93-day volatility (3 months).
The results of this strategy from 11/30/2010 - 12/29/2017 are as follows:
As you can see, the Roll Yield Strategy outpaced a buy and hold XIV strategy up until the second half of 2014. This goes to show that stellar backtests don't necessarily mean a strategy will hold up to the scrutiny of live trading. There are quite a few flaws with this basic strategy, but the primary issue is that XIV price action is a result of much more than levels of contango or backwardation (which is, essentially, what the VIX3M/VIX ratio is testing). In addition, strategies must be able to be traded in all market environments. A threshold of 1 for the strategy to go long XIV worked well from 2004-2013, but failed to work in the years that followed. Live strategies must be able to adjust to "regime changes" like the one volatility experienced post-2013.
Another issue is the binary nature of the strategy. It is binary in the sense that if it isn-t holding XIV, then it's holding VXX. VXX is a different animal and it is much more difficult to predict positive VXX performance. We would suggest implementing a third possible position: cash. Assigning the "middle ground" of VIX3M/VIX readings to a cash position would, in theory, help the strategy avoid being steamrolled by volatility crushes.
While we respect Tony Cooper and his work in the volatility space, we cannot recommend his Roll Yield Strategy to investors. We will continue to review other strategies he has made public (some of which are the foundations for many retail XIV strategies) and document their performance post 2013.
If you have any questions or would like us to test out other strategies, feel free to email us.