Using the same futures data used to calculate the VXX a fellow yahoo VXX message board poster calculated the historic XIV values.
The idea is good but the model accuracy can be improved by using more data available. For that reason I developed a XIV pricing model that uses the same basic idea, with more learning data (more market data comes available each day). Due to that it has a much better accuracy. Additionally it also considers its tracking error and the management fees.
Anyways to understand how this simplified model originally was made I copy his e-mail that explains in his own words what he did:
Attached is the chart of the daily inverse moves of XIV, along with your modified spreadsheet reversing the VXX daily moves. The chart appears to be accurate, it is almost right on with the XIV prices going back to its launch.
check it out, maybe post in on your blog.
I reviewed how he calculated the XIV and it is based on the XIV definition, whose goal is to replicate the inverse of the daily percentual VXX moves. So if the VXX goes down 1% the XIV goes up 1%. So with the daily percentual VXX changes he could calculate the XIV through all the history since VIX futures were available, note that his calculation is right because it corresponds very closely to the XIV market values.
It is interesting to note that the VXX went down from over 2000 to 40 while the XIV went up from 1.36 to 7.26 in the same period.
Thanks for the great contribution zfunk !
Here below is the graph, and the VXX processed data he used to calculate the XIV can be downloaded here: