XIV & ZIV historical data


Perusing the internet I found a neat little site: quantstrattrader.wordpress.com 

On one specific post, a commentator posted his link to reconstructed XIV, ZIV, VXX, and VXZ from 2004 using historical VIX futures and the methodology from the prospectus. Good stuff and huge thanks to quantstrat and Helmuth Vollmeier!

VXX: https://dl.dropboxusercontent.com/s/950x55x7jtm9x2q/VXXlong.TXT

XIV: https://dl.dropboxusercontent.com/s/jk6der1s5lxtcfy/XIVlong.TXT

VXZ: https://www.dropbox.com/s/y3cg6d3vwtkwtqx/VXZlong.TXT
ZIV: https://www.dropbox.com/s/jk3ortdyru4sg4n/ZIVlong.TXT

XIV 2004


Edit 11/10/14:

After some thought I realized the simulated XIV/ZIV data may be incorrect. While it follows the underlying methodology from the prospectus, I’m not sure if it includes ‘compounding drag’ inherent in the inverse product. You can see the effects here:

XIV vs VXX 3 month performance

I sent some inquires out, hopefully the author of the data can weigh in soon!

Edit 11/25/14:

Vance Harwood at SixFigureInvesting actually has also recreated XIV back to 2004. I’m not sure which came first but either way in the comment section of his blog, a reader had a similar question as I did regarding ‘drag’ on the inverse ETN; here is the response:

I stand by my calculations. I think the core part of your argument is that you think a 439% gain (my number is 409%) from March 2004 to December 2005 seems unlikely. It turns out that period of the market was extraordinarily quiet with a maximum closing VIX of 19.96. In quiet periods such as this the contango losses on the long side are very heavy. My calculations on the long index for short term volatility (SPVXSP) would be a decrease from 590277 to 100000 over that period, which would amount to a 8.8% monthly drop. This is not unprecedented, in the 13 month period between VXX reverse splits from Oct 2012 to Nov 2013 the long side dropped at a monthly rate of 8.5% even though there were 3 VIX spikes of 20 or higher. A glance at the log chart on my post would show other times where the rate of decline in VXX was higher than the simulated 2004 to Dec 2005 timeframe. A compounded rate of 8.8% would give a gain of 487% for an inverse strategy over the 21 month period you mention. Volatility drag would decrease the realizable amount by some, but the compounding benefits of a trending securityhttp://sixfigureinvesting.com/2012/10/a-hat-trick-for-inverse-leveraged-volatility-funds/ would tend to counteract that–so the 409% value is reasonable.

My simulation tracks the actual XIV IV values within +-0.2% since inception and it tracks the underlying index (SPVXSP) within +- 0.01% to the Dec 20th 2005 index start value. So I think it is highly unlikely my methodology is incorrect. The biggest uncertainty in that 2004 to 2005 timeframe is that in some timeframes the front month futures were not being traded. I used an variance based extrapolation to project the front month values in those cases. The resultant values are within historic term structure ranges.

It turns out that XIV is based on the SPVXSP index not the SPVIXSTR index you mentioned. The differences are not large, but the one you mentioned is a total returns index which includes a treasury bill component that SPVXSP does not have.




Be the first to comment on "XIV & ZIV historical data"

Leave a comment

Your email address will not be published.