Barclays’ XVZ volatility ETN is intended to allow investors to profit from volatility jumps without the contango losses that drag down approaches like Barclays’ VXX short term and VXZ medium term long only products. To accomplish this XVZ switches between 11 different mixes of short term and medium term positions in volatility futures. In volatile times there can be a different setting every day.
The switching itself is driven by the VIX/VIX3M ratio (Barcalys calls this ratio “Implied Volatility Term Structure” or IVTS) —the higher the ratio, the higher the fear that is being experienced in the market. With settings 1 through 5 shown in the table below XVZ is holding short positions in short term volatility futures—which attempt to be profitable when volatility futures are in contango. The rest of the settings are long both short and medium term futures in various mixes.
|IVTS = VIX/VIX3M||Set||Actual Short Term Alloc.||Actual Medium Term Alloc.||Backtest Days||Description|
|IVTS < 0.9||1-E||-0.3||0.7||307||Contango mode, very low fear|
|0.9 <= IVTS < 1.0||2-E||-0.2||0.8||402||Contango mode, moderate fear|
|1.0 < IVTS <= 1.05||6-E||0||1||80||Transition|
|1.05 <= IVTS <= 1.15||9-E||0.25||0.75||43|
|IVTS > 1.15||11=E||0.5||0.5||26||Backwardation,very high fear|
In XVZ’s prospectus only 5 of these settings are explicitly called out (the sets with the”-E suffix), but the fine print specifies that allocations won’t change day to day by more than 12.5%. In practice this creates intermediate points to smooth out the changes. Since some of these setting are closer than 12.5% from each other, it is possible to skip over settings in a day-to-day shift. The other thing to remember in tracking XVZ’s movements is that today’s performance is driven by the allocations that were put in place the previous business day—and those allocations were driven by the VIX/VIX3M ratio at the close of the market the business day before that.
In the table notice that of the 947 trading days of the backtest 709 of them (75%) would have been in the two lowest, contango mode settings.
I’ve updated my XVZ backtest chart below, showing VIX and VIX/VIX3M multiplied by 10. XVZ is making big decisions based on -10% / + 15% moves in the VIX/VIX3M ratio.
The graph below shows my simulated XVZ (red) vs the actual XVZ (green) for August and September. Things are tracking nicely.
If you are interested in purchasing the XVZ backtest data, or the data plus the underlying indexes and formulas see this post.