Abstract
Bitcoin
is thought of as a medium of exchange, due to its layer on the internet having
an immutable blockchain ledger, cryptographic security protocol, a
predetermined supply free of centralized control, and permanence of
transaction. It also carries weight as a
form of currency, able to transact in very small increments, allowing the
cryptocurrency to be practical and ubiquitous as a global money. One issue with using Bitcoin (BTC) is the
price volatility. While BTC is valued at
$9,065 today, it has swung from a 2018 low of $3,232 to a high of $12,907 in
2019. Within the blockchain structure of
BTC lies a feature, unspent transaction output (UTXO) which acts as digital
signature from sender to receiver. This
paper will attempt to correlate historic unspent transaction outputs of BTC in
order to determine its potential for predicting BTC price volatility.
A. Bitcoin background
Bitcoin (BTC) is governed by rules which
are purely abstract and based on mathematics. These rules are oblivious to
social conventions, irrespective of their nature (Kurda, 2012). Bitcoin can
have an “enormous impact on liberating the users of Bitcoin from social norms
they disagree with” (Kurda, 2012). Some
liberal viewpoints also see BTC as means to escape from the binds of a state,
by avoiding taxation and commit money laundering. The economics of BTC and price potential is
swayed by societal norms and views of money.
As a form of money, Bitcoin can be
thought of as digital gold or gold 2.0.
It possesses several features that are perceived to carry inherent
value: immutable blockchain ledger,
cryptographic security to uphold transactions, supply which is predetermined
and degressive over time, a proof of work algorithm to very transactions and
charge fees, maintenance of ledger by network of computers, and carries
multiple inputs and outputs. It is also
divisible by up to 8 decimal places.
B. Digital signature of BTC
To track the transactions of each
Bitcoin address, Bitcoin is designed with an architecture that avoids a
potential problem in the banking industry known as double spending. This problem is solved using an accounting
structure called unspent transaction output, or UTXO. Each transaction of every block record of
state includes the input, and the output via this structure. Unspent transaction outputs are broken up so
that the correct amount, including fees, are distributed while the remaining
value of the Bitcoin is returned to the sender as change (see figure 1). Across time, it may be used as meaningful intelligence
to understanding Bitcoin pricing.
C. Choice of analytical tool
R software was chosen, due to its
bevy of statistical packages and popularity in evaluating markets within the
finance industry (data mining, technical trading, and performance analysis). R can also directly import real-time data
from stock market indices (yet such data for Bitcoin was not available). R also allows for creating easy and
customizable graphic charts and figures, including time series plots.
D. Bitcoin datasets
Using Blockchain.com (2020) data, I
gathered UTXO and USD price data for the preceding two years (March 13, 2018 to
March 4, 2020). Below is a sample of the
raw data from both Excel sheets (blockchain.com, 2020):
E. Analysis and visualizations
Using Excel, data was prepared by
taking weekly averages of UTXO and prices in US dollars. The data was then combined into a single
Excel sheet, and imported into R. I
performed manipulation of the factor column data into dates. Then I used R basic functions to generate time
series plots, from which R users could forecast price performance (Zhang,
2016).
As one can see, mapping both plots
of unspent transaction output (scale 40-67 million) and dollar prices of BTC
(scale $3000-$13000) was impractical to show visual correlation. Changing the y-axis scale aesthetics did not
yield adequate plots. This required knowledge
of high-level plotting techniques. I
sought to correlate the data using statistical packages built into R.
I evaluated the usefulness of the given
continuous data by testing for correlation assumptions. This is performed by visually scatter
plotting the UTXO/prices to check for linearity between them. Then, using normality plots with the ggpubr
library, I can discover whether the data falls under a normal distribution
(CRAN, 2018):
Because the UTXOs follow a sigmoid
versus a normal distribution, proper statistical methods recommend using a
non-parametric correlation—Spearman or Kendall rank-based correlation tests. Spearman’s correlation test is defined as:
rho= ∑(x′−mx′)(y′i−my′)
√∑(x′−mx′)2∑(y′−my′)2
√∑(x′−mx′)2∑(y′−my′)2
Where x′=rank(x)x′=rank(x) and y′=rank(y)y′=rank(y)
The correlation coefficient between
x and y are 0.6457 and the p-value is < 2.2-16. The test indicates a moderately positive
correlation—signifying prices of BTC increases with unspent
transaction outputs of BTC.
F. Summary
The nature of Bitcoin historically shows
swings of volatility from one year to the next.
UTXOs may become a unique indicator of buy/sell pressure in the market
for Bitcoin exchanges. BTC does show price
increases that are moderately correlated to amount of BTC unspent transaction
outputs (UTXO). Indeed, the analysis
would yield more reliable results if more historical UTXO/price data were
used. With use of real-time transaction
data, the model may undergo forecasting by extrapolating days, weeks, or even
months to show asset managers and industry analysts whether to invest more or
less proportions of Bitcoin for their portfolio. BTC would not only provide an excellent
medium of exchange, but also signal measures that could properly mitigate risk
of investing in the cryptocurrency.
References
Blockchain.com. (2020).
Blockchain charts. Retrieved from
https://www.blockchain.com/charts
Surda, P. (2012 Nov 9).
Economics of bitcoin: is bitcoin an alternative to fiat currencies and
gold. Wirtschafts University. Retrieved from https://nakamotoinstitute.org/static/docs/economics-of-bitcoin.pdf
Grigg, I. (2016).
The message is the medium.
Retrieved from https://steemit.com/eos/@iang/the-message-is-the-medium
Zhang, L-C. (2016 May 13). R in finance: introduction to r and its
applications in finance. Retrieved from https://www.researchgate.net/publication/302956522_R_in_Finance_Introduction_to_R_and_Its_Applications_in_Finance
(2018). Comprehensive R archive network (CRAN). Retrieved from https://cloud.r-project.org/doc/manuals/r-release/R-intro.html#Graphics
