The Velocity and Dormancy of Bitcoin. It contains important observations and links to a research paper.
See also the comment I added to the bottom of that blog article, which should appear if the owner of that blog accepts my comment, which is quoted below.
You assume that only “change” needs to be removed from the total of
transactions when computing the velocity. Transfers to self in and out
of the system also must be removed, because only the fee in money
transfers is part of the GDP when the transfer is not to another
person for a real good or service. Ditto your section making the
erroneous claim that transfers in bitcoins external to the main
database should be factored into the velocity without regard to
whether they are transfers to pay for a good or service.
I find it unfathomable that every Bitcoin owner is spending his entire
balance 7 times every quarter. This would require he has the means to
receive bitcoins 7 times every quarter from someone else paying for
goods and services. Such a Bitcoin economy does not exist.
If data is analyzed more deeply, perhaps with the aid of data from the
FX services, I am confident that vast majority of transactions would
be transfers buying in and cashing out of Bitcoin, or moving money to
self across borders and through exchange vehicles such as real estate,
etc. There are not that many merchants accepting Bitcoin. It is
fathomable that there might be a significant quantity of illicit or
black market transfers, since Bitcoin aids anonymity. However, even if
so, such markets can not likely scale to velocity of fiat for the
mainstream population. Also FinCEN is preparing to crack down on this
As you noted, the problem with computing dormancy in Bitcoin is there
is no way to know which are transfers to self. Perhaps one could make
the reasonable assumption that large value transactions that paid a
significant transaction fee are not likely transfers to self, since
one would be motivated in that case to likely tradeoff speed of
verification for cost.
Since FinCEN will be regulating transfers (to self or other party),
but not transactions in exchange for goods and services, users do not
have an incentive to reveal the data we need.
So perhaps the only metric that is reliable is your weighted-average
of dormancy relative to itself, not as an indication of relative
velocity to those computed with GDP. I think that may have been your
intended point, but IMO you did not elucidate it.
Your weighted-average of dormancy is not Bitcoin Days Destroyed—
the latter weights the past dormancy of coins that have been
transferred over any chosen period, whereas yours is a snapshot of the
current dormancy of coins. The histogram of the percentage of Days
Destroyed graphs distribution of dormancy over the measured period.
Update: I expounded on why the 7 velocity is not likely at this time for Bitcoin.