The current issue and full text archive of this journal is available on Emerald Insight at:

www.emeraldinsight.com/1368-5201.htm

JMLC
22,1 The risk analysis of Bitcoin and
major currencies: value at
risk approach
38 Umut Uyar and Ibrahim Korkmaz Kahraman
Pamukkale Universitesi, Denizli, Turkey

Abstract
Purpose – This study aims to compare investors of major conventional currencies and Bitcoin (BTC)
investors by using the value at risk (VaR) method common risk measure.
Design/methodology/approach – The paper used a risk analysis named as VaR. The analysis has
various computations that Historical Simulation and Monte Carlo Simulation methods were used for
this paper.
Findings – Findings of the analysis are assessed in two different aspects of singular currency risk and
portfolios built. First, BTC is found to be significantly risky with respect to the major currencies; and it is six
times riskier than the singular most risky currency. Second, in terms of inclusion of BTC into a portfolio,
which equally weights all currencies, it elevates overall portfolio risk by 98 per cent.
Practical implications – In spite of the remarkable risk level, it could be considered that investors are
desirous of making an investment on BTC could mitigate their overall exposed risk relatively by building
a portfolio.
Originality/value – The paper questions the risk level of Bitcoin, which is a digital currency. BTC, a
matter of debate in the contemporary period, is seen as a digital currency free from control or supervision of a
regulatory board. With the comparison of major currencies and BTC shows that how could be risky of
a financial instrument without regulations. However, there is some advice for investors who would like to
invest digital currencies despite the risk level in this study.

Keywords Value at risk, Bitcoin, Major currencies, Money market

Paper type Research paper

Journal of Money Laundering
Control
Vol. 22 No. 1, 2019
pp. 38-52
© Emerald Publishing Limited
1368-5201
DOI 10.1108/JMLC-01-2018-0005
Introduction
As a result of the unprecedented pace of technological advancements accompanied by the
further globalization of the world, money transfers and trades are executed heavily on the
internet in our contemporary world. Fast-paced penetration of the internet usage emerged
digital money transfer system. In such unreliable system free from supervision of any
authority, various digital currencies are used e.g. Bitcoin (BTC). Development of digital
currencies as a new exchange method has long been foreseen. Nobel Prize economist Milton
Friedman addressed in 1999 that expansion of internet usage and advancements in
technology would eventuate in the invention of digital money allowing trade and out of
reach of government control (Wegdell and Andersson, 2014, p. 8).
The BTC system, the contemporary argument, relies on an article of “Peer to Peer
Electronic Cash System” published by Satoshi Nakamoto in 2008 (Nakamoto, 2008). Satoshi
Nakamoto, deemed to be the Founder of the BTC system, does not stand for a real person.
Furthermore, it is claimed that this name is associated with large-sized technology
companies in a way that while “Sa” part of this name refers Samsung, “Toshi” refers
Toshiba, “Naka” refers Nakamushi and “Moto” refers Motorola (Sonmez, 2014). Nakamoto, The risk
afterward of the foundation of the system, quit from the project without further clarifying analysis of
his identity by the end of 2010. Since then, number of developers besides Nakamoto have
contributed to the BTC. It is known that the first BTC operation was executed by system
Bitcoin and
founders in January 2009, which the purchase order was given by Laszlo Hanyecz on May major currencies
22, 2010. In this first BTC trade, Laszlo purchased two pizzas worth of US$25 in the
exchange of 10,000 BTC. In the later years, this exchange is recognized as the most
expensive trade of the world (Kocoglu et al., 2016). The general perception that BTC 39
operations are executed anonymously and its cycling throughout users have intensified
concerns about it. It is estimated that majority of drug traffic using BTC in 2012 was
executed on an online trade portal called Silk Road. This portal made 600,000 BTC
commission earning out of gross sales amount about 9.5m BTC (Ron and Shamir, 2013).
When BTC’s development, properties and operation are taken into account, a notable
increase in its value could be seen, which is argued that it could be an economic bubble.
According to Rodrigue (2011), economic bubbles characteristically have four distinguished
phases as they were illustrated below (Figure 1): the first, “stealth phase” in which
speculative operations with prices are not noticed by investors yet; the second, “awareness
phase” in which corporate investors participate in investment for-profit; the third, “mania
phase” in which almost everyone start investing owing to attention of media and rumors;
and the fourth, “blow off phase” in which the tool loses its popularity slowly and prices
return to equilibrium point . Based on the latest price movements, it is seen that BTC has
already matured its “mania phase” and transited into its “blow off phase” (Sonmez, 2014).
value at risk (VaR), one of the methods developed for investors to visualize all relevant risks
of companies as a whole, was developed by JP Morgan’s analysts. JP Morgan’s VaR algorithm
and the data set needed for calculations were made publicly available for free in November 1994.
This resulted in the adoption of the method further in the market. VaR also gained attention in
the literature and found a way for itself. Not only by professionals from the equity markets but
also by banks, pension funds, other financial institutions and by non-financial companies
adopted this method. From the academic point of view, VaR is described as a method, which
could estimate the highest loss that could arise along a certain period and in certain confidence

Figure 1.
Phases of
economic bubbles
JMLC interval in financial markets by adopting forward-looking perspective and that could be
22,1 comprehended by everyone conveniently (as a monetary value) (Bolgun and Akcay, 2009).
In other words, VaR provides a scale based on a probabilistic estimation of value loss
from a portfolio or a single asset within a certain time interval. The method considers
statistical calculations and denominates the risk in monetary aspect. The analysis has been
used in various fields such as risk reporting determination of risk limits, capital adequacy
40 regulations, internal capital allocation and performance measurement. From this point of
view, it is widely used in risk measurement in foreign exchange operations. This method
allows estimation of maximum monetary loss subject to the certain foreign exchange
investment of an investor within the certain time interval.
The objective of the present study compares conventional foreign exchange investor and
BTC investor based on the VaR method. To that end, daily market closure price of eight
major global currencies (Swiss Franc, euro, UK Pound, Japanese Yen, Australian Dollar,
Canadian Dollar and New Zealand Dollar) and BTC were used in the present analysis.
Obtained results will be assessed in terms of the risk level that they are exposed to when
they build their portfolio with the major currencies or the digital money of BTC.
In the later section of the study, the literature review on VaR and BTC was presented. In
the third section of the study, methods used in the VaR calculations were presented while
findings were assessed and compared with the current literature and the last section.

Literature review
The current literature investigated in the light of the present study could be categorized into
three fundamental sections. In the first group, the studies oriented on the VaR from various
aspects were investigated. The second group included studies concentrated on the BTC
subject. In the third group, the studies investigating the BTC and VaR method together were
included.
In the relevant current literature, studies on VaR are mostly concentrated on a
comparison of calculation methods and testing potential daily losses at various confidence
levels. The first group VaR studies were presented in chronological order.
Wang (2002) strived to calculate the impact of liquidity risk through VaR estimations. In
this study, it is determined by the VaR method that liquidity show losses at low amounts
and conclude that Monte Carlo simulation method (MCSM) provides more accurate
approaches to VaR estimations. Bozkus (2005) compares VaR method and expected shortfall
(ES) methods in this study. In this line, the author used daily USD/euro parity and Borsa
Istanbul (BIST)-100 index. The author found the ES method more practical because it does
not have a tail risk and it is more consistent with respect to VaR method. Harmantzis et al.
(2006) analyzed data sets comprised of daily return rates of S&P 500, DAX, CAC, Nikkei,
TSE and FTSE indices against daily return rates of USD, euro, JPY, GBP and CAD.
Obtained results suggest that ES models in VaR analyses yield more accurate risk
estimations.
Al Janabi (2006), by using VaR method, evaluates the risk management in the currency
market on developing Moroccan economy. He addressed that risk measurement in the
Moroccan market could be implemented objectively and thus VaR could be used in
developing markets for accurate analyses. Gursakal (2007) concludes in his VaR analysis
study using variance-covariance method (VCM) and historical simulation method (HSM) on
daily euro sales rates and return rates of BIST-30 index that stock market is more risky
investment choice with respect to foreign exchange and that the VaR value estimated for
currency is lower against the one estimated for the stock market.
 Aktas (2008) studies whether the security capital deposits of banks are close to the The risk
required amount by using VaR model on daily data from the 2004 to 2005 period. The VaR analysis of
value of the portfolio comprised of government internal bond and stocks was estimated by
using VCM. Study results show significant deviations from the assumptions of the VCM
Bitcoin and
because of high volatility. Accordingly, the utility of VCM in the determination of the bank major currencies
security deposit for balancing market risk in Turkey is found to be risky. In the study of Tas
and Iltuzer (2008) conducted in the same year of Aktas’s study, government internal bonds
41
and BIST-30 index were used as data set. Authors estimated VaR values of their portfolio by
using MCSM. Obtained results show that VaR values of 27,000 TL equally-weighted
investment on BIST-30 index were 2,435.28 TL and 3,462.48 TL at 95 and 99 per cent
confidence levels, respectively. The VaR values of equally-weighted portfolio investment
size of 50,000.00 TL on Government Internal Bonds were 1,003.84 TL and 1,439.72 TL at 95
and 99 per cent confidence levels, respectively. Authors concluded that risk associated with
the Government Internal Bonds is lower with respect to the stock market. Uckun and
Kandemir (2008) used VCM to calculate and compare the VaR values of two separate
portfolios comprised of stocks from two different indexes for a one-year period (January 1,
December 31). Five stocks with minimum standard deviation from the computer industry
and bank industry indexes were selected and the relevant portfolios were built for quarterly
periods (January-March, April-June, July-September and October-December). The study
findings address that risk characteristics of mentioned indexes are in different
characteristics.
Demireli and Taner (2009) conducted VaR estimations by using Parametric VaR (PVar),
HSM and MCSM on the portfolio comprised of equally weighted euro, Gold and USD for the
315-day period between January 2, 2008 and April 1, 2009. According to the obtained
findings on the assumption that data were normally distributed, VaR values were estimated
as 965.38 TL, 938.23 TL and 547 TL by means of VCM, HSM and MCSM, respectively. The
authors conclude in line with the existing literature that MCSM is the most valid method for
Turkish market with respect to others analyzed. Kayahan and Topal (2009) calculate the
amount of daily loss that could be faced by a company from the manufacturing industry
with its foreign exchange portfolio through PVaR and HSM according to 95 and 99 per cent
confidence levels. Authors conclude that PVaR and HSM are more significant and
convenient methods for manufacturing companies. Study of Zikovic and Aktan (2009)
distinguishes the analysis before- and after-crisis periods. Estimations made based on the
daily return rates of BIST (XU100) and Croatian Stock Market (CROBEX) indexes at
confidence levels of 95, 99 and 99.5 per cent are taken into consideration in VaR analysis. As
a result of the study, BIST (XU100) and Croatian (CROBEX) indexes are found to be similar.
Celik and Kaya (2010) used daily closure prices of five different stocks from the BIST for
the period between January 2, 2004 and November 11, 2008 for VaR estimations based on
logarithmic return rates. As a result of these estimations using VCM, HSM and
exponentially weighted moving averages (EWMA) methods, the VaR value estimated by
means of statistics of extremes was compared with the values estimated with other methods.
Obtained results from comparisons of four methods, statistics of extremes method yielded
values greater than the VCM, but lower than the EWMA and HSM. It yields more accurate
results, which could guide investors in terms of risk assessment. Andjelić et al. (2010)
evaluated the performances of Slovenia (SB120), Croatian (CROBEX), Serbia (Belexline) and
Hungary (BUX) markets for the period between January 10, 2006 and April 1, 2009 by using
HSM and Delta Normal VaR methods on the relevant their return rate data sets. Authors
aimed to estimate the maximum loss at 95 and 99 per cent confidence intervals. In their
JMLC study, authors concluded that the successful methods used for VaR estimations in developed
22,1 markets are required to be used for developing markets as well.
By using VCM, HSM, MCSM and Bootstrapping Methods at 95, 97.5 and 99 per cent
confidence levels, Rejeb et al. (2012) estimated VaR value for Tunis Dinar, USD, euro and
JPY for the period between January 1, 1999 and December 31, 2007; and compared their
findings based on aforesaid methods. Obtained findings suggest that VCM yields the most
42 accurate estimation results. Koldere Akin and Akdugan (2012) built three different
portfolios from the Turkish pension funds for the period covering 2008, 2009 and 2010. The
VaR estimation was made by using VCM based on a data set for one year at 99 per cent
confidence level. Estimated values indicated that risk levels of investment funds were
notably low and they could be classified as low-risk fund group. Akhtekhane and
Mohammadi (2012) used PVaR and HSM to determine the risk level of fluctuations between
Riyal and euro currencies for the period between May 2006 and 2011. The authors concluded
that the lowest VaR value was yielded by the parametric and historical simulation methods
at 99 and 95 per cent confidence levels, respectively. Cakir and Uyar (2013) built ten different
portfolios out of the 91 stocks listed in BIST-100 index in 2001; and estimated VaR value for
each portfolio at 95 per cent confidence level by using MCSM. The authors addressed that as
risk level of portfolio increases, the probability of loss increases as well.
Yildirim and Colakyan (2014) estimated and compared different VaR values of the
portfolio comprised of BIST 100, BIST 30, USD, euro and GBP by using different methods of
VCM, HSM, MCSM and EWMA for the period between January 2, 2008 and December 31,
2012 to determine the most effective method. Authors assume to invest 1,000 TL in each
portfolio constituent; projected totally 5,000 TL investment in a portfolio. They estimated
the probability of loss by the HSM as 116 TL, by the VCM as 100 TL, by the MCSM as 95 TL
and by the EWMA Method as 619 TL. Chen (2014) focused on measuring the market risk
with VaR, stressed VaR, and ES under the Basel accords. The aim of the study is evaluating
how well the reforms embraced by Basel 2.5 and III. As a conclusion, the author indicates
that international banking regulation under the Basel accords achieves precisely this much
and nothing more – on the strength of quantitative risk measures specifying 99 per cent
confidence level over holding periods of 10 trading days. In another study conducted by
Avsarligil et al. (2015), two different portfolios in the worth of 100,000 TL from the football
club’s stocks traded in the BIST and VaR analysis was conducted by using the Variance-
Covariance, Historical Simulation and EWMA methods for the period of 2005-2013. Under
the assumption that the data is normally distributed, it is concluded that using VCM for VaR
estimations is more preferable for investors. Terinte (2015) used daily values to estimate the
VaR values of five different stocks from various industries, traded in Romanian Stock
Market (Bucharest Stock Exchange), for the period of 2011-2015 at 95 and 99 per cent
confidence levels, whereas the total investment is 100,000 Ron, the maximum risk level is
estimated by three different methods (HSM, MCSM and VC). Authors report that the
minimum loss probability of the portfolio is yielded by the HSM, the maximum loss
probability is yielded by the MCSM. Obtained results are as follows HSM (at 95 per cent:
1,381.60; at 99 per cent: 2,666.60), MCSM (at 95 per cent: 4,253.87; at 99 per cent: 3,713.13), VC
(at 95 per cent: 2,527.67; at 99 per cent: 3,574.93).
Aziz and Ansari (2017) estimated VaR values for return rates of the portfolio built by the
stocks traded in the Indian Stock Market for the period of 1999-2014. Findings of authors
suggest that the portfolio built by instruments with high VaR variables yielded higher
returns in comparison with the portfolios built by instruments with low VaR values. Gunay
(2017) used PVaR, historical VaR, HSM and MCSM in his VaR analysis on a portfolio
comprised of equally-weighted BIST-100 index, USD/TL, euro/TL, Brent Crude Oil and Gold
from the data set covering the period of January 3, 2005-June 15, 2015. In the study of Cekici The risk
(2017), the risk level of making an investment on insurance company stocks traded in the analysis of
stock market is estimated through VaR method based on the data set from the period of
April 1, 2016-March 31, 2017. To that end, the portfolio in the worth of 5,000 TL build by five
Bitcoin and
different stocks invested in an equal amount of 1,000 TL is tested by using VCM. Study major currencies
findings address that the maximum loss will be 1,694.47 TL at 99 per cent confidence level
and that it would be risky to make an investment on stocks of insurance companies because
the estimated loss is rather high. Ugur and Bingol (2017) conducted VaR analysis with VCM 43
to estimate the highest possible loss of portfolios comprised of stocks from the BIST
manufacturing industry for daily, 10-day and 21-day holding periods based on the data
collected from the period of 2003-2016. From the subsectors of the manufacturing industry,
six different portfolios were built and the highest loss determined with the portfolio built
with the stocks of the companies in sub-sector of chemical, oil, rubber and plastic
manufacturing whereas the lowest loss was determined with a portfolio based on the sub-
sector in the manufacturing business of earth and rock.
In the second section of the literature search, the studies concentrated on BTC and other
digital currencies were included. In this regard, current studies in the literature are found
mostly related to operative details of the digital currency of BTC, relevant pros and cons and
system details.
Brière et al. (2015) studied the weekly data from the period of 2010-2013 through an analysis of
conventional and alternative investment tools in a portfolio mixture including BTC investment.
The authors reveal that BTC investment would be a beneficent tool to be applied in portfolio
diversification because of high return and volatility and less correlation with other investment
tools. However, authors emphasized that the risk level might increase in the long-term.
Christopher (2014) analyzes the function of BTC in the USA in terms of anti-money
laundering laws; and addressed the felonies that may arise from the usage of BTC as a
money laundering tool and difficulties in the enforcement of the relevant laws. Gun (2014)
indicated that BTC, non-central currency, emerged as a result of financial crises; and studied
whether BTC has solved the issues or caused new problems. The author underlined that
although it has benefits on the liberation of international operations and reducing operation
costs, it would cause financial bubbles because it is away from the supervision of
government bodies. Wegdell and Andersson (2014) investigated the potentially significant
impact of BTC on financial industry so as to investigate its characteristics. In the conclusion,
authors conclude that BTC needs to be considered as a major innovation and it will be
developed as a complementary tool in the finance industry. Sonmez (2014) studied emerge
and characteristics of BTC, its development across the world and in Turkey and its function
and significance in the economy besides the analysis of the relevant current status. They
report that the system allows illegal transactions in large amounts as BTC investors are
anonymous accordingly BTC contains both strengths and vulnerabilities together.
Cheah and Fry (2015) investigated in BTC prices. Study results, like other assets, suggest
that there is a significant tendency toward speculative bubbles in BTC prices and its base
value is essentially zero. Chu et al. (2015) conducted statistical analysis of the first electronic
payment system of BTC/USD and eight different currencies. The authors conclude that BTC/
USD currency has increased more than 50 times in the past 24 months; and that although it
yields high return rate, its volatility is rather high. In the study of Kristoufek (2015), the factors
effective on BTC prices were investigated and claimed on the contrary to the common belief
that BTC is not a speculative instrument. The factors effective on BTC price are given as
market penetration as an exchange tool, available amount and price level. By considering
experienced significant fluctuations, the author indicates that BTC is still away from being a
JMLC safe harbor in financial term. Atik et al. (2015) aimed to investigate the impact of intensifying
22,1 usage of BTC and its adoption as an investment tool on functional principles of the system
and on conventional currency markets. In the period of 2009-2015, the interaction between
daily BTC prices and the most common cross exchange in the world was analyzed by the
Granger causality analysis. As a result, it is reported that BTC and JPY are effective on each
other with delay; and that a one-way causality relationship from JPY to BTC exists.
44 Kocoglu et al. (2016) analyzed the functional operation of BTC markets, determination of
BTC prices and BTC market. Pricing process of BTC was investigated; and efficiency,
liquidity and volatility of BTC markets were analyzed. When considered as an investment
tool, it is seen that BTC is not a reliable instrument despite its high return rate; it has high
volatility and prone to speculative misuse. Bilir and Cay (2016) discussed the significance of
digital money and digital currency for financial markets and stability. As a result, it is
reported that control and prohibition of digital currencies like BTC without central structure
seems rather difficult; and spending effort to include them into jurisdictional regulations
bears risks due to the difficulty of its control. While governments have not developed a clear
policy concerning these types of currencies, this situation evidences complexity of the
situation. Ates Aslantas (2016) studied digital currencies and their positions in bookkeeping
process in accounting. In their study, authors conclude that companies using BTC as asset
need to consider BTC as a foreign currency and BTC assets need to be appreciated like other
foreign currencies at the end of the fiscal period. As BTC is not quoted by the Republic of
Turkey Central Bank system, its daily quotation data set is retrieved from the BTC market.
Indeed, numbers of institutions recognizing BTC currency or accepting payment on BTC
currency or making payment with BTC or exchanging BTC currency with other major
currencies have increased recently.
The relevant literature has also mentioned the necessity for regulations and codes on the
BTC system in terms of bookkeeping, accounting, taxation and BTC exchange markets.
Khalilov et al. (2016) introduced BTC system as usage and circulation of digital money have
expanded; explained its functions and operation and studies and developments concerning
digital money usage are presented in both Turkey and the world. The authors conclude that
efforts on the proliferation of digital currencies across financial markets have persisted in
spite of numbers of institutions across the world have addressed about associated risks.
In the final section of the literature research, studies using data set comprised of digital
currencies in their VaR analysis are included. Stavroyiannis (2017) sampled significant
digital currencies of BTC, Enthereum, Litecoin, Ripple and S&P 500 index within the scope
of the risk management; compared 10-day VaR values based on the ES methods. Obtained
results suggest that digital currencies pose a high risk.

Data and methodology

Data set used in the present study includes major currencies included in the studies of
Brunnermeier et al. (2008), Baillie (2011) and Neuby and Barrett (2017). Data set consists
of daily closure prices of the concerned foreign exchanges for 1,489 days along the period of
February 2, 2012-November 7, 2017. In the scope of the study, the digital currency of BTC
and major currencies were compared based on results of the VaR estimation methods. To
this end, logarithmic values of daily closure returns, standard deviation and correlation
matrixes were used for VaR estimations.
Table I exhibits descriptive statistics of BTC and major currencies included in the study.
Moreover, Figure 2 illustrates time series of the variations in the data set.
According to Figure 2, it could be seen that all major currencies display volatile behavior
pattern against the USD along with the studies time interval. However, whereas a steady
Statistics BTC/USD CHF/USD EUR/USD GBP/USD
The risk
analysis of
Mean 0.00474 0.00006 0.00008 0.00012 Bitcoin and
Median 0.00370 0.00018 0.00007 0.00006
Maximum 0.38042 0.17136 0.03025 0.02985 major currencies
Minimum 0.35071 0.02553 0.02417 0.08402
SD 0.05271 0.00695 0.00534 0.00562
Skewness 0.37165 10.12384 0.11914 2.24696 45
Kurtosis 13.11094 251.35010 5.27752 37.90394
Jarque-bera 6,376 3,852,028 325 76,837
Obs 1,489 1,489 1,489 1,489
JPY/USD AUD/USD CAD/USD NZD/USD
Mean 0.00027 0.00023 0.00017 0.00013
Median 0.00000 0.00000 0.00023 0.00036
Maximum 0.03279 0.02146 0.01962 0.02434
Minimum 0.03774 0.02403 0.01919 0.03130
SD 0.00739 0.00617 0.00472 0.00679
Skewness 0.03350 0.05528 0.09643 0.03211
Kurtosis 4.98733 3.79846 4.12756 3.79148
Jarque-bera 245 40 81 39
Obs 1,489 1,489 1,489 1,489
Table I.
Notes: BTC: Bitcoin; CHF: Swiss Franc; EUR: euro; GBP: UK Pound; JPY: Japanese Yen; AUD: Australian Descriptive statistics
Dollar; CAD: Canadian Dollar; NZD: New Zealand Dollar of currencies

increase was observed with the BTC/USD rate in the beginning of the five-year analysis
period, this was followed by an increase in the significant rate. Table II exhibits correlation
matrix of BTC and major currencies.
When the correlation matrix in Table II is considered, significant correlation is notable
between AUD/USD and NZD/USD currencies. Furthermore, the negative correlation
between currencies of BTC/USD and GBP/USD, CAD/USD and NZD/USD is remarkable.
In the literature review, it could be seen that VCM, HSM and MCSM are frequently
applied in VaR analyses. Although it could be analyzed by different methods, basic VaR
estimation is given in equation (1) below:

pffiffiffiffi
VaR ¼ M  a  s  T (1)
where:
M = Market value of the portfolio;
a = Confidence level (a = 0.01, 0.05, 0.10);
s = Portfolio standard deviation; and
T = Duration.
In line with the objective of the study, HSM and MCSM were preferred in VaR estimations.

Historical simulation
In the HSM process, historical changes observed with the market prices and historical return
values of the portfolio built are used to create time series of loss or profits, basis of VaR
estimation (Linsmeier and Pearson, 1996). Even though HSM provides statistical
assumption regarding portfolio returns, it is assumed that portfolio mixture remains the
 46

22,1
JMLC
BTC graphs
major currencies and
The exchange rates of
Figure 2. BTC/USD CHF/USD EUR/USD
8,000 1.20 1.4

1.15
6,000 1.3

1.10
4,000 1.2
1.05

2,000 1.1
1.00

0 0.95 1.0
I II III IV I II III IV I II III IV I II III IV I II III I II III IV I II III IV I II III IV I II III IV I II III IV I II III I II III IV I II III IV I II III IV I II III IV I II III IV I II III I II III IV
2012 2013 2014 2015 2016 2017 2012 2013 2014 2015 2016 2017 2012 2013 2014 2015 2016 2017

GBP/USD JPY/USD AUD/USD

1.8 0.014 1.1

1.7 0.013
1.0
0.012
1.6
0.011 0.9
1.5
0.010 0.8
1.4
0.009
0.7
1.3 0.008

1.2 0.007 0.6

I II III IV I II III IV I II III IV I II III IV I II III I II III IV I II III IV I II III IV I II III IV I II III IV I II III I II III IV I II III IV I II III IV I II III IV I II III IV I II III I II III IV
2012 2013 2014 2015 2016 2017 2012 2013 2014 2015 2016 2017 2012 2013 2014 2015 2016 2017

CAD/USD NZD/USD
1.1 0.90

0.85
1.0

0.80
0.9
0.75
0.8
0.70

0.7 0.65

0.6 0.60
I II III IV I II III IV I II III IV I II III IV I II III I II III IV I II III IV I II III IV I II III IV I II III IV I II III I II III IV
2012 2013 2014 2015 2016 2017 2012 2013 2014 2015 2016 2017
same along the analysis period (Manganelli and Engle, 2001). Steps followed in the HSM are The risk
given below (Rodoplu and Ayan, 2008): analysis of
 Determination of basic risk factors in the portfolio.
Bitcoin and
 Acquisition of historical data concerning the risk factors from the relevant risk major currencies
period under evaluation.
 Estimation of VaR values based on historical data for the concerned period. and
comparison of obtained VaR values with present value of a portfolio to find 47
differences (loss/profit).
 Ranking estimated daily (loss/profit) values from the worst to the best.
 Determination of total loss associated with the confidence interval.

Monte Carlo simulation

MCSM is known as the most comprehensive and strong VaR method. The VaR value
includes the impacts of non-linear relationships in the portfolio and potential changes in the
future. Additionally, this method does not assume any distribution requisite for returns (Tas
and Iltuzer, 2008). Steps followed by the MCSM for VaR estimations are given below
(Bolgun and Akcay, 2009):
 Determination of portfolio on which the VaR will be estimated.
 Estimation of fluctuations in portfolio returns caused by risk factors.
 Determination of the distribution of data set of return variation.
 Estimation of correlation and covariance matrixes of risk factors.
 Generating random number relevant with the determined distribution type.
 Generating Cholesky and singular value decomposition matrix in the covariance matrix.
 Reflection of the relationship between multiplication of price series transposed and
randomly-generated according to the distribution determined by Cholesky and singular
value decomposition matrix and past risk factors to the newly generated price series.
 Application of these price sets to the portfolio.
 Finally, estimation of VaR values corresponding to the relevant confidence level.

Findings
The analysis was conducted by using HSM and MCSM according to the model given in
equation (1) at 95 and 99 per cent confidence levels. At the analysis process, by using daily

Currencies BTC/USD CHF/USD EUR/USD GBP/USD JPY/USD AUD/USD CAD/USD NZD/USD

BTC/USD 1
CHF/USD 0.03854 1
EUR/USD 0.00344 0.59549 1
GBP/USD 0.01417 0.34744 0.52356 1
JPY/USD 0.02347 0.28610 0.26421 0.07199 1
AUD/USD 0.00122 0.29418 0.41873 0.42198 0.14049 1 Table II.
CAD/USD 0.02143 0.25173 0.38087 0.40404 0.09179 0.62042 1 Correlation matrix of
NZD/USD 0.00232 0.33686 0.44182 0.42252 0.17432 0.74466 0.56478 1 currencies
JMLC data set of each currency collected for the period of February 2, 2012-November 7, 2017,
22,1 individual VaR values were estimated. Additionally, VaR values of the equally weighted
portfolio comprised of all currencies and the one additionally including BTC currency were
estimated. Table III exhibits VaR values estimated based on two different methods.
To investigate analysis findings more clearly, VaR values yielded by the MCSM and
HSM were illustrated in the Figures 3 and 4, respectively.
48 According to Table III, for the period of February 2, 2012-November 7, 2017, findings from
the analysis of equally-weighted portfolios through MCSM and HSM exhibits parallel results.
In both methods, whereas the lowest probability of loss among all currencies for all
confidence levels was determined with the CAD/USD currency, the highest loss probability
was determined with BTC/USD currency. According to the results of MCSM at 95 and 99
per cent confidence levels, an investor who invests one USD on CAD/USD parity would lose
maximum $0.0076 and 0.0110, respectively. On the other hand, an investor who prefers
BTC/USD parity for his/her one-dollar investment at 95 and 99 per cent confidence levels
would lose maximum $0.0914 and 0.1361 on the next day, respectively. According to the
HSM at 95 and 99 per cent confidence levels, CAD/USD parity would cause maximum
$0.0077 and 0.0117 on the next day, respectively. On the other hand, BTC/USD parity would
cause an investor to lose maximum $0.0722 and 0.1665 at 95 and 99 per cent confidence
levels, respectively. Figures 2 and 3 display the notable difference between major currency
and BTC/USD currency risks clearly. Findings deliberately reveal that BTC currency
exposes investors rather high risk such that VaR value estimated for BTC/USD parity is
about 7 seven times higher than the major currency with the highest VaR value (JPY/USD).

MCSM HSM
VaR values VaR values
Confidence levels Confidence levels
Major currencies VaR 95% VaR 99% Major currencies VaR 95% VaR 99%

Portfolio (BTC) $0.0122 $0.0173 Portfolio (BTC) $0.0109 $0.0213

Portfolio (non-BTC) $0.0071 $0.0094 Portfolio (non-BTC) $0.0059 $0.0085
BTC/USD $0.0914 $0.1361 BTC/USD $0.0722 $0.1665
CHF/USD $0.0108 $0.0145 CHF/USD $0.0091 $0.0135
EUR/USD $0.0089 $0.0121 EUR/USD $0.0082 $0.0140
GBP/USD $0.0089 $0.0131 GBP/USD $0.0083 $0.0137
Table III. JPY/USD $0.0124 $0.0172 JPY/USD $0.0118 $0.0194
The VaR values of AUD/USD $0.0102 $0.0131 AUD/USD $0.0107 $0.0158
major currencies and CAD/USD $0.0076 $0.0110 CAD/USD $0.0077 $0.0117
currency portfolio NZD/USD $0.0115 $0.0153 NZD/USD $0.0113 $0.0172

Monte-Carlo Simulation
Portfolio
Portföy (non-BTC) BTC/USD CHF/USD EUR/USD GBP/USD JPY/USD AUD/USD CAD/USD NZD/USD
$-
$-0.0200
$-0.0400
$-0.0600
$-0.0800
$-0.1000
$-0.1200
$-0.1400
Figure 3. $-0.1600
The graphic of $-0.1800

MCSM VaR values VaR %95 VaR %99

 Historical Simulation The risk
Portfolio Portfolio analysis of
(BTC) (non-BTC) BTC/USD CHF/USD EUR/USD GBP/USD JPY/USD AUD/USD CAD/USD NZD/USD
$- Bitcoin and
$-0.0200
$-0.0400
major currencies
$-0.0600
$-0.0800
$-0.1000 49
$-0.1200
$-0.1400
$-0.1600 Figure 4.
$-0.1800 The graphic of HSM
VaR %95 VaR %99 VaR values

Within the scope of this study, the VaR value was estimated for each major currency, as well
as for two portfolios one of which includes BTC/USD parity beside major currencies,
whereas the other not. Two portfolios whose VaR values displayed in Table III and Figures
3 and 4, currency investments are equally-weighted. According to estimated VaR values
based on the MCSM and HSM, it is observed that the portfolio, which does not include BTC/
USD parity is found to pose less risk with respect to the ones which do not contain BTC/USD
parity. The risk difference between the two portfolios varies between 70 and 150 per cent. In
line with the study analyzing major currencies individually, it is concluded that inclusion of
BTC into the portfolio built has increased the overall risk exposed by investors. On the other
hand, in case of investors to make BTC investment in a distributed portfolio instead of
overweighting BTC, this would decrease their potential losses on their investments.

Conclusion
BTC, a matter of debate in our contemporary period, is seen as a digital currency free from
control or supervision of a regulatory board. The risk management process for financial assets
is vital for investors in terms of the determination of the duration of investment, weights and
values of assets appropriately. When BTC is taken into consideration in the sense of risk
management, measurement of risk is quite a different task as there is no an institution in
charge of digital money, uncertainties about its future and difficulty in appreciation of the real
worth of the underlying asset. Risk measurement of an investment tool is viewed as a
significant issue in our time. With regard to the risk preferences of investors, their investments
are required to be managed. The VaR method in the risk measurement literature is used by
researchers commonly but with different techniques such as basically VCM, HSM and MCSM.
The objective of this study is to compare investors with conventional foreign currencies
and digital currency of BTC by using VaR methods called HSM and MCSM at 95 and 99 per
cent confidence levels. To that end, major currencies suggested by Brunnermeier et al.
(2008), Baillie (2011), Neuby and Barrett (2017) e.g. Swiss Franc, euro, GBP, JPY, AUD, CAD
and NZD and BTC currency were included in the study. In the analyses, the risk level
exposed by investors was oriented and assessed. Daily data of each currency was collected
for the period of February 2, 2012-November 7, 2017.
Findings of the present study are assessed in two different ways; individual risk of each
constituent currency and overall portfolio built. According to the results of the MCSM and
VaR analysis conducted for each currency at 95 and 99 per cent confidence levels, it is seen
that whereas the lowest probability of loss is determined with CAD/USD parity, the highest
probability of loss is determined with BTC/USD parity. Based on the HSM analysis at 95 and
JMLC 99 per cent confidence levels, the lowest probability of loss is determined with CAD/USD
22,1 parity while the highest probability of loss is determined with BTC/USD parity. Obtained
results clearly reveal that BTC currency poses high risk individually. Furthermore, for the
study period, the risk level determined with the BTC/USD parity is six-seven times greater
than the currency (JPY/USD) displaying the highest risk after the BTC.
At the second section of the study, two portfolios are built; the first one including BTC/USD
50 parity and the second not including. Findings of MCSM and HSM indicate that whereas the
later portfolio has lower risk in comparison with the former one. Moreover, it is seen that
inclusion of BTC/USD parity into the portfolio elevated overall portfolio risk by 98 per cent on
average. However, it is revealed that investors could mitigate their general investment risk by
including BTC and major currencies into their portfolio.
In general, when studies on BTC are taken into consideration, it could be seen that BTC
investments are usually risky and not financially safe. In harmony with the existing
literature, the VaR method used in the application of this study reveals that BTC is risky
investment tool. The main factors increasing risk are mostly indicated as the high volatility
of digital currencies and lack of any legislative ground controlling the system. In spite of
aforesaid uncertainties and high risk, the investors who want to make an investment in BTC
currency are suggested to build portfolios comprised of BTC, as well as conventional foreign
currencies.

References
Akhtekhane, S.S. and Mohammadi, P. (2012), “Measuring exchange rate fluctuations risk using the
value-at-risk”, Journal of Applied Finance and Banking, Vol. 2 No. 3, pp. 65-79.
Aktas, M. (2008), “Turkiye piyasalarinda parametrik riske maruz deger modelinin tasidigi riskler”,
Afyon Kocatepe Universitesi IIBF Dergisi, Vol. 10 No. 1, pp. 243-256.
Al Janabi, M.A. (2006), “Foreign-exchange trading risk management with value at risk: case analysis of
the moroccan market”, The Journal of Risk Finance, Vol. 7 No. 3, pp. 273-291.
Andjelić, G., Djaković, V. and Radišić, S. (2010), “Application of VaR in emerging markets: a case of
selected Central and Eastern European countries”, African Journal of Business Management,
Vol. 4 No. 17, pp. 3666-3680.
Ates Aslantas, B. (2016), “Kripto para birimleri, Bitcoin ve muhasebesi”, Cankiri Karatekin Universitesi
Sosyal Bilimler Enstitusu Dergisi, Vol. 7 No. 1, pp. 346-366.
Atik, M., Kose, Y., Yilmaz, B. and Saglam, F. (2015), “Kripto para: Bitcoin ve doviz kurlari uzerine
etkileri”, Bartin Universitesi IIBF Dergisi, Vol. 6 No. 11, pp. 247-262.
Avsarligil, N., Demir, Y. and Dogru, E. (2015), “Riske maruz deger olcum yontemleri araciligiyla
BIST’te islem goren spor kulupleri uzerine bir uygulama”, Eskisehir Osmangazi Universitesi
Sosyal Bilimler Dergisi, Vol. 16 No. 1, pp. 81-107.
Aziz, T. and Ansari, V.A. (2017), “Value-at-risk and stock returns: evidence from India”, International
Journal of Emerging Markets, Vol. 12 No. 2, pp. 384-399.
Baillie, R.T. (2011), “Possible solutions to the forward bias paradox”, Journal of International Financial
Markets, Institutions and Money, Vol. 21 No. 4, pp. 617-622.
Bilir, H. and Cay, S. (2016), “Elektronik para ve finansal piyasalar arasindaki iliski”, Omer Halisdemir
Universitesi Iktisadi ve Idari Bilimler Fakultesi Dergisi, Vol. 9 No. 2, pp. 21-31.
Bolgun, K.E. and Akcay, M.B. (2009), “Risk yonetimi: gelismekte olan turk finans piyasasinda entegre
risk olcum ve yonetim uygulamalari”, Scala Yayincilik.
Bozkus, S. (2005), “Risk olcumunde alternatif yaklasimlar: riske maruz deger (VaR) ve beklenen kayip
(ES) uygulamalari”, D.E.U.II.B.F. Dergisi, Vol. 20 No. 2, pp. 27-45.
Brière, M., Oosterlinck, K. and Szafarz, A. (2015), “Virtual currency, tangible return: portfolio The risk
diversification with bitcoin”, Journal of Asset Management, Vol. 16 No. 6, pp. 365-373.
analysis of
Brunnermeier, M.K., Nagel, S. and Pedersen, L.H. (2008), “Carry trades and currency crashes”, NBER
Macroeconomics Annual, Vol. 23 No. 1, pp. 313-348.
Bitcoin and
Cakir, H.M. and Uyar, U. (2013), “Portfolio risk management with value at risk: a Monte-Carlo simulation
major currencies
on ISE-100”, International Research Journal of Finance and Economics, Vol. 109, pp. 118-126.
Cekici, E.M. (2017), “Parametrik RMD (VaR) Incelemesi: BIST’te islem goren sigorta sirketleri uzerine 51
bir arastirma”, Marmara Universitesi Oneri Dergisi, Vol. 12 No. 48, pp. 217-225.
Celik, N. and Kaya, M.F. (2010), “Uc degerler yontemi ile riske maruz deger’in tahmini ve Istanbul
menkul kiymetler borsasi uzerine bir uygulama”, Bankacilik ve Sigortacilik Arastirmalari
Dergisi, Vol. 1 No. 1, pp. 19-32.
Cheah, E.T. and Fry, J. (2015), “Speculative bubbles in bitcoin markets? An empirical investigation into
the fundamental value of bitcoin”, Economics Letters, Vol. 130, pp. 32-36.
Chen, J.M. (2014), “Measuring market risk under the basel accords: VaR, Stressed VaR, and expected
shortfall”, Aestimatio, the IEB International Journal of Finance, Vol. 8, pp. 184-201.
Christopher, C.M. (2014), “Whack-a-mole: why prosecuting digital currency exchanges won’t stop
online laundering”, Lewis and Clark Law Review, Vol. 18 No. 1, pp. 1-36.
Chu, J., Nadarajah, S. and Chan, S. (2015), “Statistical analysis of the exchange rate of Bitcoin”, PloS
One, Vol. 10 No. 7.
Demireli, E. and Taner, B. (2009), “Risk yonetiminde riske maruz deger yontemleri ve Bir Uygulama”,
Suleyman Demirel Universitesi Iktisadi ve Idari Bilimler Fakultesi Dergisi, Vol. 14 No. 3, pp. 127-148.
Gun, L. (2014), “A new form of currency: description and economic principle”, Journal of Scientific
Research and Reports, Vol. 3 No. 13, pp. 1793-1797.
Gunay, S. (2017), “Value at risk (VaR) analysis for fat tails and long memory in returns”, Eurasian
Economic Review, Vol. 7 No. 2, pp. 215-230.
Gursakal, S. (2007), “Hisse senedi ve doviz piyasasi risklerinin riske maruz deger yontemi ile karsilastirilmasi”,
Uludag Universitesi Iktisadi ve Idari Bilimler Fakultesi Dergisi, Vol. 16 No. 2, pp. 61-76.
Harmantzis, F.C., Miao, L. and Chien, Y. (2006), “Empirical study of value-at-risk and expected shortfall
models with heavy tails”, The Journal of Risk Finance, Vol. 7 No. 2, pp. 117-135.
Kayahan, C. and Topal, Y. (2009), “Tarihsel riske maruz deger (RMD) Finansal riskleri aciklamada
yeterli midir?”, Suleyman Demirel Universitesi Iktisadi ve Idari Bilimler Fakultesi Dergisi, Vol. 14
No. 1, pp. 179-198.
Khalilov, M.C.K., Gundebahar, M. and Kurtulmuslar, I. (2016), “Bitcoin ile dunya ve turkiye’deki dijital
Para calismalari uzerine bir inceleme”, 19. Akademik Bilisim Konferansi, Aksaray, available at:
http://ab.org.tr/ab17/bildiri/100.pdf
Kocoglu, S., Cevik, Y.E. and Tanrioven, C. (2016), “Bitcoin piyasalarinin etkinligi, Likiditesi ve
oynakligi”, Isletme Arastirmalari Dergisi, Vol. 8 No. 2, pp. 77-97.
Koldere Akin, Y. and Akdugan, U. (2012), “Finansal piyasalarda risklerin belirlenmesinde riske maruz
deger yontemine iliskin Bir Uygulama”, Trakya University Journal of Social Science, Vol. 14
No. 1, pp. 225-236.
Kristoufek, L. (2015), “What are the main drivers of the bitcoin price? Evidence from wavelet coherence
analysis”, PloS One, Vol. 10 No. 4.
Linsmeier, T.J. and Pearson, N.D. (1996), “Risk measurement: an introduction to value at risk”, available
at: www.exinfm.com/training/pdfiles/valueatrisk.pdf
Manganelli, S. and Engle, R.F. (2001), “Value at risk models in finance”, WP[75], European Central Bank.
Nakamoto, S. (2008), “Bitcoin: a peer-to-peer electronic cash system”, available at: https://bitcoin.org/
bitcoin.pdf
JMLC Neuby, B.L. and Barrett, S.M. (2017), “The case for a single currency”, Journal of Advances in Economics
and Finance, Vol. 2 No. 2, pp. 83-96.
22,1
Rejeb, A.B., Salha, O.B. and Rejeb, J.B. (2012), “Value-at-risk analysis for the Tunisian currency market: a
comparative study”, International Journal of Economics and Financial Issues, Vol. 2 No. 2, pp. 110-125.
Rodoplu, G. and Ayan, E. (2008), “Basel-II uzlasisinda piyasa riski yonetimi ve turkiye acisindan faiz
riskine iliskin bir uygulama”, Suleyman Demirel Universitesi Iktisadi ve Idari Bilimler Fakultesi
Dergisi, Vol. 13 No. 2, pp. 1-28.
52
Rodrigue, J.P. (2011), “Stages in a bubble”, available at: http://people.hofstra.edu/geotrans/eng/ch7en/
conc7en/stages_in_a_bubble.html (accessed 10 November 2017).
Ron, D. and Shamir, A. (2013), “Quantitative analysis of the full bitcoin transaction graph”, Proceedings
of 17th International Conference on Financial Cryptography and Data Security, Okinawa,
pp. 6-24, available at: https://link.springer.com/content/pdf/10.1007%2F978-3-642-39884-1_2.pdf
Sonmez, A. (2014), “Sanal Para bitcoin”, The Turkish Online Journal of Design, Art and Communication,
Vol. 4 No. 5, pp. 1-14.
Stavroyiannis, S. (2017), “Value-at-risk and expected shortfall for the major digital currencies”, SSRN
Electronic Journal, available at: https://ssrn.com/abstract=3028625 or http://dx.doi.org/10.2139/
ssrn.3028625
Tas, O. and Iltuzer, Z. (2008), “Monte carlo simulasyon yontemiyle riske maruz degerin IMKB30
endeksi ve DIBS portfoyu uzerinde bir uygulamasi”, Dokuz Eylul Universitesi Iktisadi ve Idari
Bilimler Fakultesi Dergisi, Vol. 23 No. 1, pp. 67-87.
Terinte, P.A. (2015), “Applicability of value at risk on romaninan capital market”, Journal of Public
Administration, Finance and Law, Vol. 2, pp. 104-110.
Uckun, N. and Kandemir, S. (2008), “Risk olcumunde riske maruz deger metodolojisi ve IMKB’de bir
uygulama”, Muhasebe ve Finansman Dergisi, Vol. 38.
Ugur, A. and Bingol, N. (2017), “BIST imalat sanayinde riskin olculmesi: riske maruz deger yontemiyle
bir uygulama”, Omer Halisdemir Universitesi Iktisadi ve Idari Bilimler Fakultesi Dergisi, Vol. 10
No. 4, pp. 271-284.
Wang, Z. (2002), “The properties of incremental VaR in Monte Carlo simulations”, The Journal of Risk
Finance, Vol. 3 No. 3, pp. 14-23, available at: https://doi.org/10.1108/eb043491
Wegdell, A. and Andersson, G. (2014), “Prospects of Bitcoin-an evalutation of its future”, available at:
http://lup.lub.lu.se/student-papers/record/4462711
Yildirim, H. and Colakyan, A. (2014), “Finansal yatirim araclarinda riske maruz deger uygulamasi”,
Dokuz Eylul Universitesi Iktisadi ve Idari Bilimler Fakultesi Dergisi, Vol. 29 No. 1, pp. 1-24.
Zikovic, S. and Aktan, B. (2009), “Global financial crisis and VaR performance in emerging markets: a
case of EU candidate states – Turkey and Croatia”, Journal of Economics and Business, Vol. 27
No. 1, pp. 149-170.

Further reading
Delcoure, N., Barkoulas, J., Baum, C.F. and Chakraborty, A. (2003), “The forward rate unbiasedness
hypothesis reexamined: evidence from a new test”, Global Finance Journal, Vol. 14 No. 1, pp. 83-93.

Corresponding author
Umut Uyar can be contacted at: uuyar@pau.edu.tr

For instructions on how to order reprints of this article, please visit our website:
www.emeraldgrouppublishing.com/licensing/reprints.htm
Or contact us for further details: permissions@emeraldinsight.com