SPOUDAI Journal of Economics and Business, Vol. 69 (2019), Issue 3, pp.

55-74

SPOUDAI
Journal of Economics and Business
University Σπουδαί
of Piraeus http://spoudai.unipi.gr

“Chicago Mercantile Exchange Bitcoin Futures: Volatility,

Liquidity and Margin”

Carl Lufta, Jin Man Leeb, Jin W. Choic

a
Department of Finance, DePaul University,
Chicago, Illinois, U.S.A. Email: [email protected]
b
Department of Economics, DePaul University,
Chicago, Illinois, U.S.A. Email: [email protected]
c
Department of Economics, DePaul University,
Chicago, Illinois, U.S.A. Email: [email protected]

Abstract

This paper explores empirically the behavior of the Chicago Mercantile Exchange (CME) bitcoin
futures contract. The analysis focuses on the time period between the launch of the CME bitcoin
futures contract on December 18, 2017, and September 17, 2018. The behavior of the bitcoin spot
market and CME futures market is compared and analyzed along several dimensions: price, volatility
and liquidity. By comparing the Garman-Klass volatilities of bitcoin spot and futures prices with those
of different assets, we find that both the bitcoin spot and futures markets exhibit relatively high
volatility compared to other assets. When the ratio of trading volume over open interest is used to
measure liquidity, the bitcoin futures market shows a mid-level liquidity. We also find while the
exchange margin is set to meet the normal price volatility that can cover the daily price movements
within one standard deviation, the brokerage margin for bitcoin futures is set at beyond two standard
deviations. Some brokerage firms impose non-margin requirements such as high net account balance
and open position limits in addition to regular margins. We conclude that the brokerage firms’
relatively high margin and non-margin requirements impede trading activity such as short-sales and
thus, liquidity and efficiency in the bitcoin futures market has been slow to develop.

JEL Classification: G1
Keywords: bitcoin, futures, Garman-Klass, volatility, liquidity, margin

1. Introduction
There is on-going debate on what a cryptocurrency is. It can be viewed as money, security,
derivative, gambling tool, computer software, etc. While Brito, Shadab, and Castillo (2014)
provide its legal interpretation from various angles and proper regulatory actions for it,
Yermack (2015) and Cheah and Fry (2015), for example, conclude that bitcoin is not a
currency, but a speculative asset, because it has no intrinsic value. However, there are many

55
 C. Luft, J. M. Lee, J. W. Choi, SPOUDAI Journal, Vol. 69 (2019), Issue 3, pp. 55-74

who see it as an important asset because it is something that had never appeared in the human
history so far and has large unproven value. Consequently, those who view it as an excellent
store of value have fueled a rapid increase in its price in 2016 and 2017. Also, those who
believe in the technology underlying cryptocurrencies have spread their popularity without
clearly defining what they are advocating – currency or technology. In fact, Caginalp and
Caginalp (2018) state that a handful of cryptocurrency traders can move the price without the
help of their majority and make a strong prediction that cryptocurrencies will have high price
volatility without the backing of any underlying asset. Consequently, as their popularity and
enthusiasm wax and wane, the public’s view of high price volatility has been solidified. In
order for investors to hedge against such volatile price movements, two futures exchanges 1
began introducing futures contracts on cryptocurrency – namely, bitcoin – in December 2017.
Despite the very volatile nature of the cryptocurrencies in recent periods, many investors and
even, central banks have entertained the idea of incorporating them into their portfolios or a
pool of monetary policy tools. For example, Bordo and Levin (2017) endorses a central bank
digital currency (CBDC) for money by saying, “[W]e have found that CBDC can serve as a
practically costless medium of exchange, secure store of value, and stable unit of account.”
(2017, P19). Under this overall scheme, Kumhof and Noone (2018) develop three model of
CBDC and analyze its implications on balance sheets of central banks and commercial banks.
As such, there are positive aspects of cryptocurrencies that can benefit the investment
community and the monetary system. However, one of the fundamental questions for a
cryptocurrency to be either an investment vehicle or a monetary policy tool or both is its
volatility in value 2 . When the underlying asset to a monetary system or an investment
portfolio exhibits a great deal of volatility in price, its value as a monetary policy tool or an
investment vehicle decreases and can be harmful to the economy. In this spirit, we examine
the volatility of a cryptocurrency – especially, the futures contracts on it. We measure the
magnitude of bitcoin futures price volatility relative to its underlying spot price and other
asset price volatilities along with their levels of liquidity. We also provide a possible answer
to the question that Hale, Krishnamurthy, Kudlyak, and Shultz (2018) posed: “Why, then, did
the price of bitcoin fall somewhat gradually rather than collapse overnight?” when the
introduction of bitcoin futures enabled the short-selling of over-valued bitcoin. We identify
and evaluate the bitcoin futures margin requirement as one of the reasons for the reduced
liquidity and gradual, not immediate, collapse in bitcoin price.
Section 1 provides brief background information on bitcoin spot 3 and bitcoin futures markets
while Section 2 is devoted to the analysis of the bitcoin spot and futures price behavior
including descriptive statistics on volatilities observed in various asset classes. The Garman-
Klass volatility measure is used for its simple and yet robust nature of incorporating the
information included in the open, high, low, and closing prices of the bitcoin spot and futures
markets. Also, the relative liquidity of the bitcoin futures contract is compared with other
futures contracts. Section 3 examines the degree of bitcoin futures volatility relative to other
assets. The study period is divided into the first 3-month period when bitcoin futures was
introduced and the cryptocurrency market was volatile, and the next 6-month period when the
cryptocurrency volatility was somewhat subdued. Section 4 presents the role of bitcoin
1
Chicago Board Options Exchange (CBOE) and Chicago Mercantile Exchange (CME) listed them. We chose
to study the CME bitcoin futures market in this paper.
2
Despite the obvious difference, we treat value the same as price in this paper because how to measure the value
of a cryptocurrency is still in debate. In fact, Caginalp and Caginalp (2018, p.1131) state that “The value of a
cryptocurrency such as Bitcoin is unchartered territory in economics.”
3
Throughout this paper, we use the term, bitcoin spot price or spot market, instead of bitcoin cash price or cash
market, to distinguish bitcoin from the bitcoin cash, a variant form of cryptocurrency that was hard-forked from
bitcoin.
56
 C. Luft, J. M. Lee, J. W. Choi, SPOUDAI Journal, Vol. 69 (2019), Issue 3, pp. 55-74

futures margins on the market liquidity. The level of the margins set by the exchange is
compared with that set by the futures brokerage firms. Summary and conclusions are
presented in Section 5.
1.1 The Bitcoin Spot Market
The definition of cryptocurrency has not been settled and there are many views on what it is.
However, the Bank of International Settlement (BIS) defines it as a currency that is
electronic, functions like cash in peer-to-peer transactions and is not the liability of a
financial institution. It is created by individuals, not governments or central banks. 4 Bitcoin
was the first cryptocurrency. It was created by Satoshi Nakamato in 2009 based on a paper
describing a peer-to-peer electronic cash system. 5
Bitcoin allows peer-to-peer transactions based on a distributed ledger technology known as
Blockchain. 6 Bitcoin uses a decentralized database where peer-to-peer transactions are
broadcast to a network of users. The transactions are validated by network nodes, miners. The
transactions, or blocks, are recorded in the ledger and linked to the previous version of the
ledger to form a chain of transactions. This ledger is based on an infrastructure known as the
protocol which is supposed to align the incentives of all the participants and insure that all
participants adhere to the rules. One of the most important features of the bitcoin protocol is
that no more than 21 million bitcoins can exist. 7 Individuals who use bitcoins to pay for
goods and services in peer-to-peer cash market transactions may obtain bitcoins on an
organized bitcoin exchange. Conversely, individuals who receive bitcoins and want to
convert them into cash may sell their bitcoins on an organized bitcoin exchange. Even though
there are numerous variant forms of bitcoin such as bitcoin cash, ethereum, ethereum classic,
litecoin, ripple, etc., we focus on bitcoin because the Chicago Mercantile Exchange (CME)
lists and trades the futures contract on an aggregate index of bitcoin spot prices. During our
study period between December 18, 2017 and September 17, 2018, the daily spot market
turnover value, bitcoin price times quantity exchanged, averaged $1.034 billion. 8
1.2 Chicago Mercantile Exchange Bitcoin Futures Market
On December 18, 2017, the Chicago Mercantile Exchange (CME) launched its bitcoin futures
contract. Table 1 provides its salient features. The contract size is 5 bitcoins and expires on
the March quarterly cycle with first 2 nearby months serially listed.
The CME bitcoin futures contract does not track the price of bitcoin from a single bitcoin
exchange. In fact, it follows the Bitcoin Reference Rate (BRR) that is a combination of
bitcoin spot prices from a minimum of four bitcoin spot exchanges. The BRR aggregates
bitcoin trading activity across major bitcoin spot exchanges between 3:00 p.m. and 4:00 p.m.
London time. For example, there are four bitcoin constituent spot exchanges: GDAX,
Bitstamp, Kraken and itBit. 9 Calculation rules are geared toward a maximum of transparency

4
https://www.bis.org/publ/qtrpdf/r_qt1709f.pdf. Bech, Morten and Rodney Garratt, “Central Bank
Cryptocurrencies” BIS quarterly Review, September 2017, p 57.
5
https://bitcoin.org/bitcoin.pdf “Bitcoin: A Peer-to-Peer Electronic Cash System” Satoshi Nakamoto
[email protected], www.bitcoin.org.
6
https://www.bis.org/publ/qtrpdf/r_qt1709f.pdf. Bech, Morten and Rodney Garratt, “Central Bank
Cryptocurrencies” BIS quarterly Review, September 2017, p 58.
7
https://www.bis.org/publ/arpdf/ar2018e5.htm “Cryptocurrencies: looking beyond the hype.” BIS Annual
Economic Report, June 17, 2018, p 98.
8
Aalborg, Molnar, and de Vries (2018), for example, document the patterns of price, volatility, and trading
volume of bitcoin spot market in order to identify the factors that influence bitcoin volatility.
9
David Lerman, “CME Bitcoin Futures: The Basics” Bitcoin Webinar, June 5, 2018.
57
 C. Luft, J. M. Lee, J. W. Choi, SPOUDAI Journal, Vol. 69 (2019), Issue 3, pp. 55-74

and real-time replicability in underlying spot markets. The bitcoin futures contract is settled
in cash based on the BRR.

Table 1: Salient Features of the Chicago Mercantile Exchange Bitcoin Futures Contract

Contract Unit 5 bitcoin, as defined by the CME CF Bitcoin Reference Rate

(BRR)

Minimum Price Fluctuation Outright: $5.00 per bitcoin = $25.00 per contract. Calendar
Spread: $1.00 per bitcoin = $5.00 per contract.

Trading Hours CME Globex and CME ClearPort: 5:00 p.m. – 4:00 p.m. CT
Sunday – Friday.

Listing Cycle Nearest 2 months in the March Quarterly cycle (Mar, Jun, Sep,
Dec) plus the nearest 2 "serial" months not in the March
Quarterly cycle.

Termination of Trading Last Day of Trading is the last Friday of contract month.
Trading in expiring futures terminates at 4:00 p.m. London
time on Last Day of Trading.

Position Limits Spot Position Limits are set at 1,000 contracts. A position
accountability level of 5,000 contracts will be applied to
positions in single months outside the spot month and in all
months combined.

Price Limits Price limits for a given Business Day are made by reference to
the most recent Bitcoin Futures settlement price, settled at 3:00
p.m. Central time each Business Day.

Special price fluctuation limits equal to 7% above and below

prior settlement price and 13% above and below prior
settlement price and a price limit of 20% above or below the
previous settlement price. Trading will not be permitted outside
the 20% above and below prior settlement price.

Settlement Cash settled by reference to Final Settlement Price, equal to the

CME CF Bitcoin Reference Rate (BRR) on Last Day of
Trading.

Source: CME, CME Bitcoin Futures Frequently Asked Questions

58
 C. Luft, J. M. Lee, J. W. Choi, SPOUDAI Journal, Vol. 69 (2019), Issue 3, pp. 55-74

2. Bitcoin Futures Price Trend and Relative Price Volatility

2.1 Price Trend
Figure 1 displays the movements of the bitcoin spot and nearby futures prices during the
study period. As well noted in the media, the peak bitcoin price was observed on December
17, 2017, a day before the CME bitcoin futures contract was introduced. Since then, both the
spot and futures prices declined in a close, tight relationship.

Figure 1. Daily Closing Prices of Bitcoin Futures and Bitcoin Spot

(12/18/2018-09/17/2018)

The reason for this steady downward movement in bitcoin price was attributed to the short-
selling opportunities made available via futures trading. Hale, Krishnamurthy, Kudlyak, and
Shultz (2018), however, wonder why this downward price adjustment was gradual, not
immediate, given the nature of short-selling.
Table 2 presents the descriptive statistics for the bitcoin spot and CME bitcoin futures price,
trading volume and open interest for the study period of December 18, 2017 through
September 17, 2018. 10
We observe that the average bitcoin futures and spot price levels are similar, except for the
carrying charge difference, and that their standard deviations also are similar. However, the
trading volumes, expressed as number of bitcoins traded, show a drastic difference between
the markets where the bitcoin spot market dominates the bitcoin futures market. This
observation is not startling, given that the futures contract is at the very infant stage of
development. What is interesting is the fact that the large bitcoin spot market size indicates

10
We focus on the nearby bitcoin futures prices because they are the most liquid bitcoin futures contracts with
the greatest volume and open interest.
59
 C. Luft, J. M. Lee, J. W. Choi, SPOUDAI Journal, Vol. 69 (2019), Issue 3, pp. 55-74

the growth potential for the bitcoin futures market which has a very small trading volume for
now.
Table 2. Descriptive Statistics for Daily Bitcoin Futures and Bitcoin Spot Price, Volume and
Open Interest (12/18/2018-09/17/2018)
Variable Observations Mean Std. Dev. Min Max
Bitcoin Futures
Closing Price 188 8,719.18 2,621.91 5,865.00 19,100.00
Bitcoin Futures
Trading Volume 188 15,626 11,014 2,030 57,765
Bitcoin Futures
Open Interest 188 1,487 658 139 2,460
Bitcoin Spot
Closing Price 188 8,778.93 2,680.01 5,903.44 19,114.20
Bitcoin Spot
Trading Volume 188 795,111 229,895 520,466 1,805,494
Source: CME Bitcoin Futures and Bitcoin spot price data from Yahoo Finance BTC-USD.
While the bitcoin spot and nearby futures closing prices move mostly in tandem, there seems
to be a noticeable downtrend in both the bitcoin spot and futures prices during the study
period. At the start of the bitcoin futures contract in December 2017, the price level was
fluctuating around the level of $15,000 per bitcoin. After that, both the spot and nearby
futures prices declined to a level close to $7,000 during the spring and summer, 2018. In
order to determine if the volatility regime changed during this period, we arbitrarily divided
the study period into two subperiods: the first subperiod is an approximate 3-month period
between December 18, 2017 and March 31, 2018, and the second subperiod is an
approximate 6-month period between April 1, 2018 and September 17, 2018. Figure 1 shows
visually the two distinct volatility regimes. The first period shows a high volatility whereas
the second period, a relatively low volatility.
2.2 Liquidity
We examine the relative market sizes of the bitcoin spot and futures to analyze the relative
liquidity of the futures market to the spot market. In order to measure the size of the spot
market, we convert the spot market turnover, price times number of coins traded, into the
number of bitcoins by dividing the spot market turnover by the daily closing bitcoin spot
price during the study period. For the futures market size, we multiply the daily trading
volume by 5 because the size of the CME bitcoin futures contract is five bitcoins.
Figure 2 displays the respective market sizes of the bitcoin spot and the bitcoin futures
market in number of bitcoins. Paying attention to the difference in scale between the bitcoin
spot market and the nearby bitcoin futures market, we note that during the nine-month study
period, the daily spot market bitcoin volume fluctuated between 500,000 and 1,750,000,
while the CME nearby futures contracts generated daily volumes between 2,000 and 55,000.
The average daily coin volume for the front month, or nearby, bitcoin futures contract was
15,625 coins. Since the average daily coin volume in the bitcoin spot market was 791,685, it
means that the CME bitcoin futures market coin volume was only 2% of the bitcoin spot
market. The pure size difference indicates that the bitcoin futures market is relatively less
liquid than the spot market, holding all else constant.

60
 C. Luft, J. M. Lee, J. W. Choi, SPOUDAI Journal, Vol. 69 (2019), Issue 3, pp. 55-74

Figure 2. Daily Trading Volume of Bitcoin Futures and Bitcoin Spot in Number of Bitcoin
(12/18/2018-09/17/2018)

Figure 2 also shows that while the daily bitcoin spot trading volume in number of bitcoins has
a downtrend through the spring and summer, 2018, the bitcoin nearby futures trading volume
in coins indicates a general uptrend during the same period. This observation is interesting in
light of the price trend observed in Figure 1. Combining the two figures, we see that there is
a strong inverse relationship between coin price and coin trading volume. 11 This may
indicate that investors showed less interest in bitcoin spot market as the price fell while
bitcoin futures traders were trying to build up the market while the price was less volatile.

The coin volume metric provides good insight into the differences in trading activity between
the bitcoin spot and futures markets, but it does not provide information about the liquidity
for the bitcoin futures market relative to other futures markets. Market liquidity can be
measured and evaluated in various ways. For example, one can look at the size of the market
activity and conclude that the larger the trading volume and open interest, the greater the
liquidity. On the other hand, one can look at specific trading characteristics such as the bid-
ask spread, and conclude that the narrower the bid-ask spread, the greater the market
liquidity. While these measures are good in evaluating liquidity in well-developed markets,
they may not be good for measuring the liquidity of new or less-developed markets such as
the bitcoin futures market. In such a market, the ratio of trading volume to open interest is a
more appropriate liquidity measure because it focuses specifically on the trading activity that
creates and supports open interest. A large ratio means that trading volume is large enough to
support the willingness of the traders to carry an open position overnight or longer. On the
other hand, in a well-established market, this ratio can be small because open interest is often
much larger than the trading volume. The large open interest, being associated with a large
hedging demand and use, is often found in a well-developed futures market. To understand
11
This inverse relationship was also observed by Aalborg, Molnar and de Vries (2018, Table 13) in the bitcoin
spot market.
61
 C. Luft, J. M. Lee, J. W. Choi, SPOUDAI Journal, Vol. 69 (2019), Issue 3, pp. 55-74

the liquidity in the bitcoin futures market, we calculate the ratio of trading volume to open
interest, define it as the liquidity ratio, and then plot the ratio’s daily values for the time
period between December 18, 2017 and September 17, 2018 in Figure 3.
Figure 3. Liquidity Ratio of the Ratio of Daily Volume over Daily Open Interest in Bitcoin
Futures (12/18/2018-09/17/2018)

Figure 3 shows that the liquidity ratio increases toward the nearby futures contract expiration
and reaches its peak for that contract on the expiration day, the last Friday of the contract
month. During the first expiration cycle in March 2018 for bitcoin futures, the liquidity ratio
reaches a level greater than 6, indicating that the trading volume is more than 6 times the size
of open interest. The ratio drops significantly after the contract expiration but slowly builds
up over the following months. For the case of June and September contracts, the ratio
reaches a level greater than 2 at expiration. This indicates that the liquidity increases at each
contract expiration, making the orderly expiration and settlement possible. Therefore, there
has been no significant market disturbance in the bitcoin futures market.
This observation of the bitcoin futures market liquidity can be expanded and compared to
other futures contracts as seen in Table 3.
Table 3 shows that the well-developed markets such as E-mini S&P 500 futures and 10-year
T-Note futures have the liquidity ratio below 0.5 whereas the euro futures has 0.527. Also,
the standard deviations of these markets are relatively small, indicating that they are well-
established and stable. However, the gold futures and light sweet crude oil futures show a
liquidity ratio greater than bitcoin futures. The ratios of 0.910 and 0.971 for gold futures and
crude oil futures, respectively, indicate that their volume is almost equal to the size of open
interest. Also, their standard deviations are much larger than E-mini S&P 500 and 10-year T-
Note futures, indicating that they are less-developed and somewhat unstable.

62
 C. Luft, J. M. Lee, J. W. Choi, SPOUDAI Journal, Vol. 69 (2019), Issue 3, pp. 55-74

Table 3: The Liquidity Ratio of Daily Trading Volume over Open Interest for Selected
Futures Contracts (12/18/2018-09/17/2018)
Variable Mean Std. Dev. Min Max

Bitcoin Futures 0.788 0.918 0.013 6.583

E-Mini S&P 500 Futures 0.485 0.191 0.049 1.419

Gold Futures 0.910 1.904 0.000 22.140
10-year T-Note Futures 0.488 0.391 0.094 2.998
Euro Futures 0.527 0.236 0.054 2.040
LS Crude Oil Futures 0.971 0.777 0.166 3.661
Sources: CME Bitcoin Futures, Yahoo Finance BTC-USD, CME E-Mini S&P 500 Futures, COMEX Gold
Futures, CME T-Note Futures, CME Euro Currency Futures, and Light Sweet Crude Oil Futures. All data are
daily trading data.
This relatively large liquidity ratio and large standard deviation may mean that these markets
attract more speculative interest than E-mini S&P 500 and 10-year T-Note futures markets.
As speculative interest rises and falls, the trading volume rises and falls without much change
in open interest. Given this observation, we conclude that the bitcoin futures market is less-
developed, than the E-mini S&P 500 and 10-year T-Note futures markets. Given the short
length of bitcoin futures trading, it is clearly wrong to conclude that the bitcoin futures
market is more mature or better established than gold or crude oil futures markets, simply
based on the bitcoin futures liquidity ratio of 0.788 being lower than both markets. In fact,
given the not-so-insignificant speculative trading present in gold and crude oil, we suspect
that the smaller liquidity ratio found in the bitcoin futures market may indicate that it has less
speculative interest than the gold futures and crude oil futures markets during this period.
2.3 Relative Price Volatility
There are a few studies that examined the volatility of bitcoin spot and futures prices. For
example, Kochling, Muller, and Posch (2018) examine the effect of the bitcoin futures
contract on the efficiency of the bitcoin spot market by employing daily closing price data for
the CME bitcoin futures contracts and bitcoin spot market. They find that after the CME
futures contracts were introduced in December of 2017, the bitcoin spot market became
weak-form efficient and the price predictability disappeared. They attribute this bitcoin spot
market result to the creation of short positions via the CME bitcoin futures contracts. Kapar
and Olmo (2018) also analyze the price discovery function fulfilled by the bitcoin futures
contracts, by using daily price data and a vector cointegrated model. They find that the CME
futures market dominates the bitcoin spot market in the price discovery process. Corbet,
Lucey, Peat, and Vigne (2018), while examining the hedging efficiency of bitcoin futures,
conclude that hedging increases spot price volatility and spot market leads the futures market.
Aalborg, Molnar and de Vries (2018), on the other hand, examine the factors that determine
the bitcoin spot price volatility and conclude that realized volatility can be predicted by the
heterogenous autoregressive model. Given the robustness of the Garman-Klass volatility
measure, Tan, Chan and Ng (2018) use it to improve volatility prediction and measure the
degree of persistence in cryptocurrency prices. They find that the bitcoin spot market
exhibited relatively low volatility that was more predictable when compared to the other
cryptocurrencies.

63
 C. Luft, J. M. Lee, J. W. Choi, SPOUDAI Journal, Vol. 69 (2019), Issue 3, pp. 55-74

While many of these studies dealt with various aspects of bitcoin volatilities, their
concentration was on the bitcoin or cryptocurrency spot market, not on the bitcoin futures
market. Consequently, we examine closely in this paper the volatility of the bitcoin futures
prices in relation to bitcoin spot price and other futures prices. To understand and compare
the true nature of price volatility with other assets, we also choose to employ the Garman-
Klass (1980) extreme value method, as shown below in Equation (1).

HIGH 2 CLOSE 2
𝜎 2 =. 5 �𝑙𝑛 LOW � −.39 �𝑙𝑛 OPEN
� (1)

Where: HIGH denotes the highest price observed during the trading day;
LOW denotes the lowest price observed during the trading day;
OPEN denotes the opening price at the beginning of the trading day;
CLOSE denotes the closing price at the end of the trading day.

The Garman-Klass formula utilizes 4 pieces of price information observed in a given day:
High, low, open and close price. Beckers (1983) has shown that the Garman-Klass volatility
is robust and performs better than volatility estimated from historical closing prices. Since
Equation 1 generates variances, we take the square root of these values to measure the
volatility of the assets included in this study. Table 4 provides descriptive statistics for the
Garman-Klass volatilities for the bitcoin spot and futures markets, as well as assets that enjoy
active futures markets.

Table 4. Descriptive Statistics of the Garman-Klass Annualized Volatility

(12/18/2018-09/17/2018)
Variable Obs Mean Std. Dev. Min Max

Bitcoin Futures 184 0.905 0.551 0.144 3.557

Bitcoin Spot 184 0.733 0.561 0.153 3.669
E-Mini S&P 500 Futures 184 0.141 0.104 0.030 0.840
Gold Futures 184 0.097 0.053 0.000 0.316
10 Year T-Note Futures 184 0.039 0.019 0.011 0.140
Euro Futures 184 0.085 0.029 0.040 0.250
LS Crude Oil Futures 184 0.264 0.093 0.087 0.602

Table 4 shows the descriptive statistics on the Garman-Klass annualized volatilities

calculated for CME bitcoin futures, bitcoin spot and selected futures prices. It is interesting
to note that the mean volatilities of bitcoin futures and bitcoin spot are much higher than
those of the E-mini S&P500, gold, 10-year-T-note, euro and crude oil futures contracts
during the study period. In fact, the bitcoin futures volatility is about 10 times larger than
gold and euro futures, about 6 times larger than the E-mini S&P 500 futures and about 23

64
 C. Luft, J. M. Lee, J. W. Choi, SPOUDAI Journal, Vol. 69 (2019), Issue 3, pp. 55-74

times larger than the 10-year T-Note futures. 12 The annualized price volatility of 90.5% for
the bitcoin futures contract is striking and is not seen often in the futures market. Also, the
higher volatility in bitcoin futures than in bitcoin spot is important and noteworthy because
they should exhibit a very similar magnitude.
Figure 4. The Garman-Klass Annualized Volatility of Bitcoin Futures and Selected Futures
Contracts (12/18/2018-09/17/2018)

We provide a visual comparison of the Garman-Klass volatilities for the futures contracts via
the graphs displayed in Figure 4. This set of graphs shows the Garman-Klass volatility for
selected futures contracts which provide a visual comparison of price volatility across
different types of futures contracts. Noting that the scale of the bitcoin futures graph dwarfs
the other graphs, it shows clearly the extremely large volatility in bitcoin futures as compared
to the other futures contracts. However, we find visually a volatility pattern that is similar to
all futures contracts. The price volatility is clearly higher before April 1, 2018 than after,
except the case of crude oil.

12
Similarly, Baek and Elbeck (2015) find that bitcoin price is 26 times more volatile than S&P 500.
65
 C. Luft, J. M. Lee, J. W. Choi, SPOUDAI Journal, Vol. 69 (2019), Issue 3, pp. 55-74

To quantify and examine the relationship between the bitcoin futures volatility and the other
assets volatilities, we calculate the Pearson product-moment correlation coefficients among
the selected assets and present them in Table 5.
Table 5. Pearson Product-moment Correlation Coefficients of the Garman-Klass Annualized
Volatilities (12/18/2018-09/17/2018)
E-Mini 10 LS
Bitcoin Bitcoin S&P Gold Year Euro Crude
Futures Spot 500 Futures T-Note Futures Oil
Futures Futures Futures
Bitcoin Futures 1.000
Bitcoin Spot 0.928 1.000
E-Mini S&P 500
Futures 0.286 0.253 1.000
Gold Futures 0.045 0.037 0.239 1.000
10 Year T-Note
Futures 0.220 0.169 0.664 0.334 1.000
Euro Futures 0.078 0.025 0.159 0.314 0.380 1.000
LS Crude Oil
Futures -0.122 -0.181 0.290 0.155 0.235 0.055 1.000

The correlation coefficients in Table 5 confirm the visual patterns we observed in Figure 4
but reveal that the bitcoin futures volatility behaves very differently from other asset markets.
The small positive correlation coefficients with E-mini S&P 500 and 10-year T-Note futures
show the small overall market sensitivity linked to bitcoin futures while the near-zero
coefficients with gold and euro futures show virtually no relationship with bitcoin futures at
all. The negative correlation coefficient with the crude oil futures price volatility may be a
reflection of different trading environments where crude oil is closely anchored to the
macroeconomic fluctuation but bitcoin is not. Finally, it is not surprising that the bitcoin
futures price volatility has the strongest positive correlation with bitcoin spot price volatility.

3. Changing Volatility
In Section 2 we used the graphs in Figures 1 and 4 to identify what appears to be a regime
shift in the bitcoin market volatility after April 1, 2018. In this section, we examine the
regime shift by dividing the data into two subperiods: December 18, 2017 through March 31,
2018, and April 1, 2018 through September 17, 2018. Using regression analysis, we first
explore how the volatilities of the bitcoin spot and futures markets behaved. Then, we
examine how the bitcoin futures volatility behaved in relation to the established futures
market volatilities during the two subperiods and the whole period.
3.1 Changing Bitcoin Volatility in Three Separate Periods
The following simple regression model, as expressed in Equation (2), is estimated where the
bitcoin futures volatility is a function of the bitcoin spot volatility and then regressed on the
bitcoin spot volatility. The usual regression estimation assumptions of normally, identically,
and independently distributed error term are assumed.

66
 C. Luft, J. M. Lee, J. W. Choi, SPOUDAI Journal, Vol. 69 (2019), Issue 3, pp. 55-74

𝜎𝑡𝐹 = 𝛼 + 𝛽𝐹 (𝜎𝑡𝐶 ) + 𝑒𝑡 (2)

where 𝜎𝑡𝐹 represents the square root of the Garman-Klass variance for the nearby bitcoin
futures price, and 𝜎𝑡𝐶 means the square root of the Garman-Klass variance for the bitcoin spot
price at time t with i.i.d errors of 𝑒𝑡 .
The statistical significance of the null hypothesis that the slope coefficient, βF, is equal to 1
will indicate if the change in bitcoin futures volatility is the same as that in bitcoin spot
volatility. The magnitude and sign of the intercept term, α, will indicate the presence of a
fixed effect.
The entire study period, December 18, 2017 through September 17, 2018, is divided into two
subperiods: December 18, 2017 through March 31, 2018, and April 1, 2018 through
September 17, 2018. The regression Equation (2) is estimated for the three periods: the first
estimation is for the whole study period, the second for the first subperiod, and the third for
the second subperiod. Table 6 presents the regression results for these periods to show the
changing relationships over time.
Table 6. Simple Regression Output of Bitcoin Futures Price Volatility on Bitcoin Spot Price
Volatility
Whole Period: First Subperiod: Second Subperiod:
12/18/2018- 12/18/2018- 04/01/2018-
09/17/2018 03/31/2018 09/17/2018

𝛽𝐹 0.912*** 0.820*** 1.064***

(0.0271) (0.0438) (0.0761)

Intercept 0.178*** 0.331*** 0.080*

(0.0250) (0.0596) (0.0390)

N 184 68 116
R2 0.862 0.842 0.632
Standard errors in parentheses, * p<0.05, ** p<0.01, *** p<0.001
We note that there is a strong fixed effect present in all three periods when the intercept terms
are all positive and significant at the .05 level of significance or lower. This shows that
bitcoin futures volatility is generally higher than the bitcoin spot volatility during the
December 18, 2017 through September 17, 2018 period. Moreover, all of the slope
coefficients are positive and significant at .001 significance level, indicating that the change
in bitcoin futures price volatility and spot price volatility have a positive and significant
relationship. However, when the slope coefficients were tested against the null hypothesis of
1, the slope coefficients for the whole period and for the first subperiod generated t statistics
of -3.24 and -4.11 respectively which meant the null hypothesis had to be rejected. However,
the t statistic for the second subperiod slope was 0.841 which meant that the null hypothesis
could not be rejected. Thus, these results lead us to conclude that the bitcoin futures volatility
was clearly greater than the bitcoin spot volatility during all three periods of our study.
However, the slope coefficients being less than one in both the whole study period and the
first subperiod indicate that the change in the bitcoin futures volatility was less than the
change in the bitcoin spot volatility after netting out the fixed effect. Even though this
relationship was not observed during the second period when the bitcoin price became lower

67
 C. Luft, J. M. Lee, J. W. Choi, SPOUDAI Journal, Vol. 69 (2019), Issue 3, pp. 55-74

and more stable, the overall absolute level of bitcoin futures volatility is higher than that of
bitcoin spot volatility in all three periods. To confirm this relationship, we estimate the 50-
day rolling window regression coefficients from Equation (2), calculate the corresponding
correlation coefficients as shown in Figures 5.
Prior to April 2018, correlation coefficients between bitcoin spot and futures prices were
around 0.9, indicating close tandem movements. However, this relationship deteriorated until
late August 2018. In the meantime, the regression coefficients peak around late May 2018
and declined thereafter. This indicates that at the initial stage of bitcoin futures trading, the
bitcoin futures market was dominated and led by the relatively high bitcoin spot prices in
2017, showing high correlation coefficients above 0.8 and high regression coefficients above
1. However, from June 2018, both the correlation and regression coefficients declined, which
indicates the gradual strengthening of the bitcoin futures market and less reliance on the
bitcoin spot market.
Figure 5. Rolling Window Coefficient and Correlation Coefficients between Volatilities of
Bitcoin Futures Price and Bitcoin Spot Price (Window Size = 50 Trading Days)

In summary, we note that when the bitcoin price was high during the first subperiod, the
bitcoin futures price volatility was higher than the bitcoin spot price volatility. However,
when the bitcoin price decreased and remained stable during the second subperiod, the
sensitivity of changes in bitcoin spot and futures price volatility converged. Overall, for the
whole period, the price volatility observed in the first subperiod dominates the bitcoin price
volatility observed in the second subperiod. Consequently, its aggregated effect is manifested
as the bitcoin futures price volatility being higher than the bitcoin spot price volatility.
3.2 Bitcoin Volatility Relationships to Other Futures Market Volatilities
Earlier in Table 5 we used correlations to document the relationships between the volatilities
of the bitcoin markets and other futures markets. Now we use regression analysis for the three
periods to examine the robustness of the relationship between the bitcoin futures price
volatility and the price volatility of other futures contracts: E-mini S&P 500, Gold, 10-year T-

68
 C. Luft, J. M. Lee, J. W. Choi, SPOUDAI Journal, Vol. 69 (2019), Issue 3, pp. 55-74

Note, Euros, and Crude Oil. We estimate the simple regression Equation (3) below under the
usual regression assumptions:
𝜎𝑡𝐹 = 𝛼 + 𝛽𝑖 �𝜎𝑡𝑖 � + 𝑒𝑡 (3)
where 𝜎𝑡𝐹 represents the square root of the Garman-Klass variance for the nearby bitcoin
futures contract, and 𝜎𝑡𝑖 means the square root of the Garman-Klass variance for the
established futures contract i at time t with 𝑒𝑡 as errors. i represents one of the following
futures contracts: E-mini S&P 500, Gold, 10 Year T-Note, Euros, and Crude Oil.
Table 7. Cochrane-Orcutt autoregressive estimation output of the daily volatility of bitcoin
futures on alternative financial futures

E-Mini
Gold
S&P 500 10-year T- Euro Crude Oil
Futures Futures Note Futures Futures Futures
A. Whole Period (12/18/2017-10/17/2018)
𝛽𝑖 1.159** 0.329 1.338 0.520 -0.211
(0.399) (0.636) (2.016) (1.037) (0.339)
Intercept 0.681*** 0.813*** 0.792*** 0.801*** 0.901***
(0.0936) (0.101) (0.112) (0.119) (0.120)
N 183 183 183 183 183
R2 0.045 0.001 0.002 0.001 0.002
B. First Subperiod (12/18/2018 - 3/31/2018)
𝛽𝑖 1.105 0.786 1.454 2.251 -0.794
(0.618) (1.141) (3.750) (2.166) (0.873)
Intercept 1.109*** 1.234*** 1.241*** 1.110*** 1.507***
(0.160) (0.154) (0.205) (0.220) (0.245)
N 67 67 67 67 67
R2 0.047 0.007 0.002 0.016 0.013
C. Second Subperiod (4/1/2018-9/17/2018)
𝛽𝑖 0.518 0.499 0.609 -0.737 0.0416
(0.417) (0.582) (1.689) (0.926) (0.274)
Intercept 0.519*** 0.531*** 0.559*** 0.642*** 0.569***
(0.0564) (0.0642) (0.0662) (0.0829) (0.0795)
N 115 115 115 115 115
R2 0.013 0.006 0.001 0.006 0.000
Standard errors in parentheses, * p<0.05, ** p<0.01, *** p< 0.001

Once again, the statistical significance of the null hypothesis that the slope coefficient, βi, is
equal to 1 will indicate if the change in bitcoin futures volatility is the same as that in other
futures volatility. The magnitude and sign of the intercept term, α, will indicate the presence
of a fixed effect. We initially assumed i.i.d. errors but the residuals indicate very strong
positive serial correlations, so we estimate the model using the generalized least-squares
method by Prais and Winsten (1954) proposed by Cochrane and Orcutt (1949). Table 7
presents the regression results.
Table 7 shows that there is a very strong positive fixed effect for the whole period. This
indicates that bitcoin futures volatility has the highest volatility among the selected futures

69
 C. Luft, J. M. Lee, J. W. Choi, SPOUDAI Journal, Vol. 69 (2019), Issue 3, pp. 55-74

contracts. Also, when the slope coefficients were tested against the null hypothesis of 1, only
the crude oil contract generated a t-statistic that justified rejecting the null hypothesis. This is
consistent with the correlations between the bitcoin futures and the crude oil futures reported
in Table 5 and with our earlier conclusion that crude oil is anchored to macroeconomic
fluctuation while bitcoin is not. Failure to reject the null hypothesis of 1 for the E-mini S&P
futures, gold futures, T-Note futures, and euro futures indicates that the change in bitcoin
futures price volatility is the same as volatility change in these more established futures
contracts. However, this is not the case for bitcoin spot and futures volatilities as observed in
Table 6 where it was shown that the bitcoin futures volatility is greater than the bitcoin spot
volatility.
Thus, there rises a puzzling question that deals with volatility. Given that the futures and spot
markets should behave similarly, the bitcoin futures price volatility should not exceed the
bitcoin spot price volatility. But it did, especially during the first subperiod. Can a possible
answer lie in the fact that the volatility difference is caused by a lack of liquidity in the
bitcoin futures market in comparison to that in bitcoin spot market? Given that the bitcoin
spot market is much more liquid than the bitcoin futures market via its pure size difference,
we find a possible answer to this question by examining the margin schemes employed by the
exchange and the brokerage community.

4. The Role of Margin

The reason why bitcoin futures price volatility seems to be higher than the bitcoin spot price
volatility during the study period may be found in the lack of optimal trading activity in the
futures market. This is because the bitcoin futures contract is in the infant stage of
development and requires an establishment of broad public acceptance which may come with
time. That said, we find a possible answer in the margin requirements established by the
exchange and more importantly, that practiced by the brokerage community that is
represented by the futures commission merchants (FCMs). It is well known that the margins
affect the trading volume and open interest. Therefore, we examine the adequacy of
exchange margins and the brokerage margins next.
4.1 Exchange Margins
Bitcoin futures contracts are subject to daily price fluctuation limits of 7%, 13% and 20%.
These limits apply to both upside and downside price changes relative to the prior day’s
bitcoin futures settlement price. The Chicago Mercantile Exchange (CME) margin
requirements 13 for speculators in bitcoin futures contracts as of December 1, 2017 were 43%
of the contract’s notional value for the maintenance margin and initial margin set at 1.1 of
maintenance margin, or 47.3%. For example, assume that a customer purchased one bitcoin
futures contract at $6,230 per bitcoin. Since each contract controls 5 bitcoins, the notional
value of the contract was $31,150. Consequently, the initial margin for the position was
calculated as 1.1 x 0.43 or 0.473 of $31,150 which equals $14,734, while the maintenance
margin, 0.43 of $31,150 which equals $13,394.
Given that the difference between the initial margin and maintenance margin is $1,340, it is
about 4.3% of the futures contract’s notional value, $31,150. This means that the exchange’s
maintenance margin is well below the daily price limit of 7% and thus, can work well in
periods of normal volatility. Also, the initial margin of $13,394 which is equivalent to
$2,678.80 per bitcoin is very similar to the price movement of one standard deviation of

13
See https://www.cmegroup.com/education/bitcoin/cme-bitcoin-futures-frequently-asked-questions.html
70
 C. Luft, J. M. Lee, J. W. Choi, SPOUDAI Journal, Vol. 69 (2019), Issue 3, pp. 55-74

$2,621.91 as observed in Table 2. This means that the initial margin covers about 68% of the
daily price movements. Even though a higher margin that can cover 95% of price
movements may be desirable, it seems that the exchange’s initial and maintenance margins
are set reasonably low enough to not hinder the trading activity. While this finding is
encouraging to the trading activity, a quite different story is told by the role and impact of
actual margins required the customers by brokerage firms.
4.2 Brokerage Margins
It is known that the brokerage community sets their own margins to protect themselves from
possible customer defaults. Even if they wish to have a larger volume of trading to generate
larger revenue, they must put their financial security over profit. In the case of bitcoin futures,
due to the extreme price volatility observed in the bitcoin spot market in 2016 and 2017,
brokerage firms decided to set their own customer margins far above those set by the
exchange. For example, the margin requirements for a speculative account at a prominent
futures brokerage firm based in Chicago were 110% of notional value for initial margin and
94% of notional value for maintenance margin 14. In addition, the brokerage firm required an
account balance of $250,000 and only allowed open positions in 3 bitcoin futures contracts 15.
Therefore, when the notional value of a bitcoin futures contract is $31,150 as in the previous
example, a customer must post an initial margin of 1.1 times $31,150 which equals $34,265.
The maintenance margin is 0.94 of $31,150 which equals $29,281. These levels of a
brokerage firm’s margins are more than double the exchange margins and clearly not in line
with the risk that the exchange saw. Given that one standard deviation of bitcoin futures
price is $2,621.91, which is $13,109.55 per contract, the initial margin of $34,265 per
contract can cover 2.6 standard-deviation movements in bitcoin futures price, which
corresponds to an approximately 99.5% of all price movements. While it is understandable
for the brokerage community to set their own margins higher than those of the exchange in
anticipation of their customer default risk, the impact of high brokerage margins on trading
volume can be very drastically negative. That is what we suspect to be the cause of reduced
liquidity in the bitcoin futures market and the reason behind the slow downward price
adjustment in bitcoin price that Hale, Krishnamurthy, Kudlyak, and Shultz (2018) questioned,
given the short-sale opportunity made available via the futures market. The brokerage
community being afraid of the customers’ potential default possibility in trading bitcoin
futures have often set a margin at 100% or higher of the notional value of the futures contract.
This can clearly hinder the growth of the bitcoin futures trading and in fact, can choke off the
market liquidity that is necessary for the market growth. Consequently, the impact of short-
sale opportunities on bitcoin price was not allowed to fully propagate during the early stage
of bitcoin futures trading, resulting in a gradual downward price adjustment.

5. Summary and Conclusions

This paper explores empirically the behavior of the Chicago Mercantile Exchange (CME)
bitcoin futures contract during the time period between the launch of the CME bitcoin futures
contract on December 18, 2017, and September 17, 2018. Using the Garman-Klass volatility
measure, the price volatility of the CME futures contract is compared with that of the bitcoin
spot market. We find that the bitcoin futures market shows a higher price volatility than the
bitcoin spot market. Also, the Garman-Klass volatility of bitcoin futures prices is compared
14
These levels of margin requirements were not unique to this brokerage firm but widely practiced by many
other brokerage firms in the industry as confirmed by personal interviews.
15
Once again, this type of non-margin requirements was not unique to the firm we interviewed. There were
wide variations among brokerage houses regarding non-margin requirements for bitcoin futures trading.
71
 C. Luft, J. M. Lee, J. W. Choi, SPOUDAI Journal, Vol. 69 (2019), Issue 3, pp. 55-74

with that of different futures contracts such as the E-mini S&P 500, gold, 10-year T-Note,
euro, and crude oil futures to examine the relative volatility of the bitcoin futures contract.
We find that the bitcoin futures market exhibits a relatively higher volatility than all other
futures markets.
When the ratio of trading volume over open interest is used to measure market liquidity, the
bitcoin futures market shows a mid-level liquidity of 0.788, indicating that one open interest
is supported by 0.788 trading volume. On the other hand, the well-developed and highly-
liquid futures markets such as the E-mini S&P 500 and 10-year T-Note show a liquidity ratio
below 0.5, indicating that one open interest is supported by less than 0.5 trading volume.
This means that open interest is larger than trading volume, which is a characteristic of a
well-established futures market. As for the less-developed and less-liquid futures markets
such as the gold and crude oil futures, the liquidity ratio is close to 1, indicating that trading
volume is as large as the open interest.
When the Garman-Klass volatilities of selected futures markets are analyzed via simple
regression models for two subperiods and the whole period of study, we find the bitcoin
futures price having the dominantly higher volatility than all the other markets. In particular,
contrary to common expectation, the bitcoin futures price volatility is higher than the bitcoin
spot price volatility. Also, the short-sale opportunities available via the futures market did
not bring about an immediate downward price adjustment in bitcoin. To explain the reason
behind these observations, we examine the margin requirements set by the exchange and the
brokerage community. We find while the exchange margin is set to meet the normal price
volatility that can cover the daily price movements within one standard deviation, the
brokerage margin for bitcoin futures is set at beyond two standard deviations. In addition,
brokerage firms require customers to meet non-margin requirements such as high net account
balance and open position limits. We conclude that these relatively high margin and non-
margin requirements of brokerage firms discourage trading activity and thus, dampen the
development of liquidity that is needed to build the bitcoin futures market. Consequently, the
brokerage margin has room to be adjusted downward to a level that secures the brokerage
firms’ financial health and at the same time, encourages more active participation of the
customers. While there are many studies on the relationship between margins and price
volatility, Hardouvelis and Kim (1995), for example, look at the metal futures and conclude
that increased exchange margins reduce trading volume but decreased exchange margins
have unclear impact on trading volume. We find the conclusions of the Hardouvelis and Kim
study and others not applicable to our conclusion because they do not examine the role and
impact of brokerage firms’ margin and non-margin requirements on the growth of trading
volume and liquidity.
Overall, however, we understand the limited scope of this study due to the limited data
observed during a very short time period. Despite this shortcoming, we have verified the
high volatility of bitcoin futures price relative to selected futures markets and provide a
valuable insight into the impact of the brokerage firms’ margin and non-margin requirements
on the development of the bitcoin futures market.

72
 C. Luft, J. M. Lee, J. W. Choi, SPOUDAI Journal, Vol. 69 (2019), Issue 3, pp. 55-74

Acknowledgements
We would like to thank anonymous referees for their valuable comments and Rich Mackie and Steve
Ivey for their assistance in collecting margin information. Also, the comments from the participants of
the 2018 Athenian Policy Forum held at Piraeus, Greece, were very helpful and appreciated.

References

Aalborg, H.A., Molnar, P., and de Vries, J.E., 2018. What Can Explain the Price, Volatility and
TradingVolume of Bitcoin? Finance Research Letters,
https://doi.org/10.1016/j.frl.2018.08.010
Baek, C. and Elbeck, M., 2015. Bitcoins as an Investment or Speculative Vehicle? A First Look.
Applied Economics Letters, Vol. 22, No. 1, pp. 30-34.
Bech, M. and Garratt, R., 2017. Central Bank Cryptocurrencies. BIS quarterly Review, September
2017, pp 55-70.
Beckers, S. 1983. Variance of Security Price Returns Based on High, Low, and Closing Prices. The
Journal of Business, Vol. 56 No. 1, pp. 97-112.
BIS, 2018. Cryptocurrencies: looking beyond the hype. BIS Annual Economic Report, June 17, 2018,
pp 91- 114, available at < https://www.bis.org/publ/arpdf/ar2018e5.htm > [accessed: 29 Oct
2018].
Bordo, M.D. and Levin, A.T., 2017. Central bank digital currency and the future of monetary policy.
NBER Working Paper Series 23711, August 2017.
Brito, J., Shadab, H., and Castillo, A., 2014. Bitcoin financial regulation: Securities, derivatives,
prediction markets, and gambling. Columbia Science and Technology Law Review, Volume 16,
pp.144-221.
Caginalp, C., and Caginalp, G., 2018. Valuation, liquidity price, and stability of cryptocurrencies.
PNAS, Vol. 115 No. 6, pp 1131-1134.
Cheah, E.T. and Fry, J., 2015. Speculative Bubbles in bitcoin Markets? An Empirical Investigation
into the Fundamental Value of Bitcoin. Economic Letters, 130, pp 32-36.
CME, CME Bitcoin Futures Frequently Asked Questions, available at
<https://www.cmegroup.com/education/bitcoin/cme-bitcoin-futures-frequently-asked-
questions.html> [accessed: 29 Oct 2018].
CME, Performance Bonds/Margins, available at
<https://www.cmegroup.com/clearing/riskmanagement/performance-bonds-margins.html>
[accessed: 29 Oct 2018].
Cochrane, D., and Orcutt, G.H., 1949. Application of least squares regression to relationships
containing auto-correlated error terms. Journal of the American Statistical Association, 44, pp.
32–61.
Elliott, D.J., DeLims, L., and Singel, R., 2018, Cryptocurrencies and Public Policy Key Questions and
Answers, Oliver Wyman, available at <https://www.oliverwyman.com/content/dam/oliver-
wyman/v2/publications/
2018/February/Cryptocurrencies_and_Public_Policy_Oliver_Wyman.pdf> [accessed: 29 Oct
2018].
Garman, M., and Klass, M., 1980. On the Estimation of Security Price Volatilities from Historical
Data. The Journal of Business, Vol. 53 No 1, pp. 67-78.
Hale, G., Krishnamurthy, A., Kudlyak, M., and Shultz, P., 2018. How Futures Trading Changed
Bitcoin Prices. FRBSF Economic Letter, Federal Reserve Bank of San Francisco, 2018-12, May
7¸2018.
Hardouvelis, G.A. and Kim, D., 1995. Margin Requirements, Price Fluctuations, and Market
Participation in Metal Futures. Journal of Money, Credit, and Banking, Vol. 27, No. 3, pp. 659-
671.
Kapar B. and Olmo, J., 2018. An analysis of Price Discovery Between Bitcoin Futures and Spot
Markets. Economic Letters, https://doi.org/10.1016/j.econlet.2018.10.031

73
 C. Luft, J. M. Lee, J. W. Choi, SPOUDAI Journal, Vol. 69 (2019), Issue 3, pp. 55-74

Kochling, G., Muller, J., and Posch, P.N., 2018. Does the Introduction of Futures Improve the
Efficiency of Bitcoin? Economics Letters, https://doi.org/10.1016/j.econlet.2018.10.031
Kumhof, M. and Noone, C., 2018. Central bank digital currencies-design principles and balance sheet
implications. Staff Working Paper No. 725, Bank of England.
Lerman, D., 2018. “CME Bitcoin Futures: The Basics” Bitcoin Webinar, June 5th, 2018, available at
<https://www.danielstrading.com/video/learn-cme-bitcoin-futures> [accessed: 29 Oct 2018].
Nakamoto, S., 2009. Bitcoin: A Peer-to-Peer Electronic Cash System, available at
<https://bitcoin.org/bitcoin.pdf> [accessed: 29 Oct 2018].
Prais, S.J., and Winsten, C.B., 1954. Trend estimators and serial correlation. Working paper 383,
Cowles Commission. Available at <http://cowles.econ.yale.edu/P/ccdp/st/s-0383.pdf> [accessed:
29 Oct 2018].
Tan, S., So-Kuen, J., and Ng, K., 2018. On the Speculative Nature of Cryptocurrencies: AStudy on
Garman and Klass Volatility Measure. Finance Research Letters,
https://doi.org/10.1016/j.frl.2018.12.203
Wooldridge, J.M., 2013. Econometrics: A Modern Approach, 5th edition, 2013, South-Western,
Mason Ohio.
Yermack, D., 2015. Is Bitcoin a Real Currency? An Economic Appraisal. Handbook of Digital
Currency, Elsevier. pp. 31-40.

74