Content uploaded by Jiarun Hu
Author content
All content in this area was uploaded by Jiarun Hu on Jun 04, 2020
Content may be subject to copyright.
International Journal of Applied Research in Management and Economics
ISSN 2538-8053
______________________________
* Corresponding Author E-Mail Address: one@jiarunhu.com
2538-8053/ © 2020 IJARME. All rights reserved.
The Fluctuations of Bitcoin Price during the Hacks
Jiarun Hu1*, Qian Luo2 and Jiaen Zhang3
1 Fudan University, China
2 Zhongnan University of Economics and Law, China
3 DePaul University, USA
ARTICLE INFO
ABSTRACT
Keywords:
Bitcoin price
Price fluctuation
Bitcoin hack
Cryptocurrency thefts
Security breaches of the cryptocurrency exchanges usually cause the
price fluctuation in the market. Approximately one hundred
cryptocurrency thefts, including hacks and scams, has occurred since
2012 to 2018, half of which are hacks of Bitcoin. Based on the thirty
Bitcoin hacks, this study portrays the general price pattern during the
hack. And it illustrates the link between the size of the hack and the
subsequent price change of Bitcoin. The tests reveal that the larger
the volume of the hack, the stronger the price drop. However, a
similar obvious relationship does not exist for the recovery of the
price. The study might be the first piece of research focus on the
hacks and the price pattern in a short time period.
1. Introduction
Cryptocurrencies have caught the global attention for its unique way of the transaction and the
vibrant price movements in recent years, especially at the end of 2017 when the prices of major
cryptocurrencies hit historical maxima. Among all cryptocurrencies, Bitcoin is the most
prominent open source peer-to-peer cryptocurrency which is operating without central
authority (Nakamoto, 2008). Cryptocurrencies bring a lower-cost way of transferring assets,
especially in international commerce in which the comparatively intricate procedure is required
through fiat currency exchange. On the other hand, cryptocurrency serves not only as an
electronic medium of exchange but also as a speculative investment asset with trading available
24-hours a day, seven days a week.
Digital currencies achieve decentralization and convenience but so far only at the cost of lower
security and higher volatility. From June 13th in 2011 to February 5th in 2019, there are more
than one hundred occurrences of cryptocurrency thefts (hacks and scams), of which the cases
of Bitcoin thefts account for more than 90% of the total occurrences. Specifically, there are
mainly two types of theft: hacks and exit scams. The exit scam cases are a basic human fraud
and as such market participants do not blame the cryptocurrencies directly for them.
Conversely, hacks usually bring into question the suitability and sustainability of
cryptocurrency projects and thus cause the price to change. With more transactions through
Bitcoin and larger investment volume from the capital market, the potential risk toward Bitcoin
exchange security becomes higher than all the other time period. On February 7th of 2014, one
of the largest Bitcoin exchanges in the world, Mt. Gox (Magic the Gathering Online Exchange),
reported a major hack of approximately 460 million USD involving 700,000 bitcoins which at
the time was 7% of the total volume of the bitcoin (DeVries, 2016). Another characteristic of
bitcoin is the high volatility of its price. Although Bitcoin started at nearly zero value in 2009,
the price ended around $1100 at the December of 2013. A year later, the price dropped below
$300 and then started a powerful revert, resulting in a boost to $19,000 (Ciaian, Rajcaniova, &
Kancs, 2016). Another dramatic example to show its price volatility is on May 22nd of 2010,
INTL. J. APPL. Res. MANAGE. & ECON., 3 (1):10-20, 2020
12
the IT engineer Laszlo Hanyecz ordered a pizza with 10,000 BTC when the price of per bitcoin
used to be 0.33 USD while in December of 2017, and this amount of bitcoin could make him
a ten billion fortune.
According to the prior researches on the topic of Bitcoin price, the traditional market
determinants and digital currency-specific factors are identified as two significant factors to
the price formation of Bitcoin (Ciaian et al., 2016). However, when a haphazard cryptocurrency
hack happens, not only will the exchange lose a certain amount of currencies, but the value of
the Bitcoins for all users will usually drop. Therefore, when the shock of a cryptocurrency hack
disseminates to the market and public, the holders of Bitcoin will probably foresee the risk of
a new price drop and sell the digital asset, thereby reducing its price further. Apart from the
former condition of market shock, the hackers usually will transfer one cryptocurrency being
hacked into another one, which usually will change the supply and demand of certain
cryptocurrencies instantly.
In the present investigation, the historical price of 30 hacks (Appendix 6.1) of Bitcoin hacks
from 2012 to 2019 has been collected. For the reason that, of all kinds of cryptocurrency,
Bitcoin has the most recorded hacks, the highest value and the most extensive market cap. As
no research has been done to our knowledge on the impact of hacks on the Bitcoin price, we
first collect the historical price data for each hack. This study investigates the hypothesis that
there is an immediate price drop of Bitcoin following the news about a hack. In most cases, we
find that the effect of hacks on price is negative on average, it causes the price drop in different
levels but the degree to which price drops varies from large to insignificant, and even no impact
in some cases. Further, we examine the impact that the scale of the hack has on the price drop
by regressing the price change on the variable “adv” (the amount of bitcoin hacked divided by
volume). The result shows statistical significance in the price change between 2 days before
and the exact day hack happens, which proves that a hack with a larger volume will probably
lead to a stronger price drop.
Another aspect of the research that we consider is the recovery of the price change after the
hacks. First, however, it has to be noted that not all 30 hacks have a noticeable price decline
for some hacks may have a limited scale against the market trading volume. Besides, the hacks
might interfere with other material economic events. Nonetheless, by using the linear
regression method, we do not find statistical significance with the scale of the hack (adv) and
the price changes, which shows the hack itself is not a causal determinant for the post-hack
period. However, the absolute amount of the stolen cryptocurrency has correlation with the
price of the first and second day after the hack happens.
The paper is structured as follows: (2) Related works in the price and hacks of cryptocurrency
area; (3) The dataset, variables the methods of the research; (4) The analysis of the research
and cases study of Mt. Gox and Bitfinex Breach Hacks; (5) The conclusion and the discussion
of the research.
2. Related Works
The topic of Bitcoin and the other cryptocurrencies has attracted a growing interest in the
Bitcoin price analysis and the security risk study. However, the most of the prior researches
investigate these two subjects separately. The analogous researches can be borrowed from the
analysis of marketing reaction in stock market.
Among the researches of price analysis, Ciaian et al. (2016) analyze Bitcoin price formation
from both the traditional determinants of currency price and digital currencies specific factors
which illustrate the long-term price formation of Bitcoin price. Urquhart (2017) highlights
prices ending with 00 decimals compared to other variations through his examination to Bitcoin
prices for clustering, the potential trading benefit from such clustering and the determinants of
the clustering. In the security risk part, Feder and et al (2017) investigate how one such risk,
INTL. J. APPL. Res. MANAGE. & ECON., 3 (1):10-20, 2020
13
distributed denial-of-service (DDoS) attack, affects the Bitcoin ecosystem and the potential to
be financially lucrative from DDoS attacks on currency exchanges. Due to the extraordinary
scale and influence, the Mt. Gox hack make itself a good sample for the researchers to analyze
the relationship between the hacks and potential manipulation of the Bitcoin price.
Apart from the study that mainly focusing on one particular side of Bitcoin, Feng, Wang and
Zhang (2018) investigate informed trading in the Bitcoin market by examining the Bitcoin
price sensitivity with the material events including events of market, government and hacking.
By classifying the events into negative and positive ones, the research detects the quantiles of
the order sizes of buyer-initiated and seller-initiated orders are abnormal and further test the
timing of informed trading.
Inspired from a parallel research from the stock market, our research is similar to the market’s
reaction research of Carter and Simkins (2003). Unlike the earthquake and hurricanes for the
airlines, the September 11th event had an influence on the all the US airline companies with no
exception. Due to the panic emotion resulted from the catastrophic event, the airline company
stocks plumped in the first trading date on September 17th, 2001. Further, they analyze the
reaction of various airline stocks to estimate whether the price reaction was consistent with
rational pricing. However, in our study toward the price of Bitcoin, more concerns are paid into
the general pattern of fluctuation rather than the influence of the hack to the exchanges from
the globe.
Besides, another two studies of stock price reaction to unexpected events are conducted by
Salas (2010), Datta and Dhillon (1993). For instance, Salas estimated the stock market reaction
to sudden executive deaths to illustrate how executive death as an unexpected event will
influence the company’s stock price in various conditions. Differently, Datta and Dhillon tested
the reaction of stock market to unexpected earnings and found the response of the bond is
symmetric with the stock market.
3. Data
The Bitcoin price data used in this paper are from Coinmarketcap.com from April 27th, 2013
to January 5th, 2019 and Investing.com for the previous data in 2012 and early 2013. We collect
the historical daily price of Bitcoin during the hacks and the trading volume of the day when
hacks happened. In order to observe the fluctuation during the hack, we create a new set of
variables named as "pbX” and “paX" (price X days before and after hacks). For the variable
"p0" we chose the low price of the date the hack happens, which might indicate the more
substantial impact compared to the close price of the day of the hack. For the other dates, close
prices are used to represent the final price level after one day’s fluctuation. The scale of the
hacks can be judged by two determinants: the amount measured in USD and the proportion of
the market volume that the hack constituted. However, due to the different impact that a hack
of a certain size can have depending on the overall size of the market, the proportion will
illustrate the scale of the hack more precisely and consistently for our purposes.
The 30 hacks cover the time from March 2nd, 2012 to December 27th, 2018. In order to obtain
a more representative result, as mentioned before firstly we exclude the exit scam. These cases
of financial fraud involve a long and vague timeline while the hacks are unexpectable incidents
with traceable sources of information to locate the precise dates. The same approach has been
taken to the two extortion cases
1
. Besides, for those hacks, the study excludes the Mt. Gox
Hack as its vast amount makes it singularly unique (40 times larger than the second-largest
Bitfinex Security Breach in August 2016).
1
Sailesh Bhatt alleged extortion (April 10th, 2018) and William Kopko Ransom extortion (October 15th,2018)
INTL. J. APPL. Res. MANAGE. & ECON., 3 (1):10-20, 2020
14
4. Analysis and Case Study
We observe the price drop from the previous days. 26 of the 30 hacks had seen the price drop
between the day before and the day of the hack. Additionally, the only abnormal values for
ch1_0 and ch2_0 during which the price increased happened in December of 2017, when the
value of Bitcoin skyrocketed to the historical price peak while the other three hacks had an
increase that beneath 0.45 USD which is about 0.03% of the price. By using the box plot
(attached below), we observe the price change from a number of days before the hack to the
day the hack happens. For two days before, there are 17 hacks for which the change in price
was negative, and the mean level of the change was -0.0252. From the box plot for 3 days
before is visible that the first hypothesis for the negative impact of the hacks seems clear.
Figure1. The box plot of price change
Further, we test if the size of the hack has an impact on the price fluctuation. Through our
regression with “chX_0” (X=1,2,3,7,15,30) on the volume variable “adv”, we find the sizes
and price changes in 2 and 3 days have statistical significance (detailed results are available in
the Appendix). Thus, if the proportion of the trading volume is larger, the drop in the value will
be stronger in 2 and 3 days, which proves the hypothesis that not only the hacks impact the
Bitcoin price negatively, but also that there exists a relationship between the size and price
drop.
For the post hack period, the same regression with the price after the hack shows no statistical
significance. However, the test between the actual amount of the lost value and the change in
price after the hack has decent statistical significance (in regressions of the price change
variables ch0_1 and ch0_2 on the stolen amount variable StolenAmountK). One possible
explanation for the divergent behavior of Bitcoin price before and after the hack might be that
the shock of the actual number of the stolen amount may be more effective than the trading
volume in impacting the confidence for the recovery of Bitcoin price in the short term.
Although the linear regressions demonstrate the correlation between the scale of the hack and
the price change pattern, the extraordinary case of 2014 Mt Gox hack deserves a closer
examination of its background and the profound influence that it had on the market. Similarly,
later in 2016, a large Bitfinex hack can be qualified as a representative to show the desired
price pattern.
INTL. J. APPL. Res. MANAGE. & ECON., 3 (1):10-20, 2020
15
4.1. Mt. Gox Major Hack on February 7th, 2014
In our prior research, we regard the Mt. Gox as an extraordinary event due to its tremendous
size ($300,000,000). As the historical price data graph displays, the price started to decline
from the open price of Bitcoin at $783.2 to the bottom of the day at $654.4. The drop of the
day reached 16% of the value, which ranks as the second largest on the hack list.
2
However, the background of the end of 2013 is worth our additional attention, because of
another type of events which can significantly impact Bitcoin price, the government
restrictions. Two months before the Mt. Gox case in the December of 2103, Chinese
government banned financial institutions from using Bitcoins which already made the Bitcoin
market suffer from a considerable decline (50% from the peak to the bottom) since December
of 2013. Although Mt. Gox hack has an enormous size comparing to any other cryptocurrency
hacks, the drop-recovery pattern can also be applied to this significant hack while the only
difference is the longer time needed to recover. Only after 20 days did the price recover to 60%
of the initial value on the day when the hack happened, through a giant crash which nearly
wiped off 90% of the bitcoin value. Research by Willy Report (2014) demonstrates that the
trading bots of the exchange added to the problem as they magnified the trading volume before
the Mt. Gox hack.
Figure 2. Historical Price of Bitcoin during Mt. Gox Hack in 2014 9 (Data source: Coinmarketcap.com)
4.2. Bitfinex Security Breach on August 2nd,2016
The Bitfinex hack can also be qualified as a massive hack for nearly 120,000 bitcoins worth up
to $72 million were stolen. Different from the Mt. Gox hack, however, during Bitfinex security
breach the price change displayed a “standard” behavior without other material incidents. On
the first day of the hack, the price temporarily declined by 10%. It has soon recovered, however,
and returned to 93.4% of pb1 on the next day and 95.4% for the price of two days after hacking.
The result also matches with our result that if a hack causes the price to decline, it leads to an
insufficient recovery, in that the price does not return to the level observed before the hack.
Furthermore, the second halving day of Bitcoin fell on July 9th, 2016, shortly before the hack.
The halving two kinds of effects: for the Bitcoin miners, it implies the halving of the reward;
for the market, the fall in the supply of Bitcoin can lead to an increase in price, at least
temporarily. Thus, the price of Bitcoin was subjected to considerable fluctuation but a moderate
increase after the halving.
2
The largest one happened at October 24th, 2013, the Input.io Wallet Hack, when the price increased for more
than three months to the day before the hack at $213.62 but dropped to the low price at $168.52.
INTL. J. APPL. Res. MANAGE. & ECON., 3 (1):10-20, 2020
16
Another difference between the Bitfinex hack and the Mt. Gox hack was that the exchange
instantly reported the information of the loss and the counter-measures taken through Reddit
and the status page. This could have relieved the panic of the market and made the recovery of
the price closer to the average level.
Figure 3. Historical Price of Bitcoin during Bitfinex Hack in 2016 (Data source: Coinmarketcap.com)
5. Conclusion
With increasing attention from the market and the high price fluctuations, cryptocurrency
exchanges are targeted by the destructive hacks since 2012. Looking at the frequency and value
of stolen cryptocurrencies, we analyze 30 Bitcoin hacks from 2012 to 2018. In the samples, 26
hacks out of 30 have seen a price drop from the first day before the hack.
Further, the study the impact of the size of the hack in two different ways, the proportion of the
amount against the trading volume and the actual amount. Our study shows a positive
correlation between the volume of cryptocurrencies stolen as a proportion of the trading volume
and the price 2 and 3 days before the hack. At the same time the absolute amount of
cryptocurrencies being stolen can affect the recovery of the bitcoin price in the first and second
day after the hack happens.
Due to the lack of previous similar research, my paper does not explore the pattern and
mechanism of the hacks. Although the research includes 30 Bitcoin hacks (the total number of
the recorded hacks in the database is 54) from a time scale of six years, there are still 24 hacks
excluded from our study for the ambiguity of the time or the meagre amount. Another concern
for the research into the price patterns may be the multitude of the background events which
happened around the same time as the hacks. Hacks are only one of the factors that disturb the
price at a specified period. The essential negative or positive incidents will all contribute to the
price fluctuation and may make the price change more complicated to analyze. Our research
could be further improved by more endeavor to gather the data and to build a more nuanced
linear regression to involve those additional factors for a better understanding of the Bitcoin
price pattern in the short term, especially when the market faces unexpected incidents. Given
their rising importance, we expect to see more research about the Bitcoin and other
cryptocurrencies’ price pattern in the near future.
INTL. J. APPL. Res. MANAGE. & ECON., 3 (1):10-20, 2020
17
References
Carter, D. A., & Simkins, B. J. (2004). The market’s reaction to unexpected, catastrophic
events: the case of airline stock returns and the September 11th attacks. The Quarterly
Review of Economics and Finance, 44(4), 539-558. doi: 10.1016/j.qref.2003.10.001
Datta, S., & Dhillon, U. (1993). Bond and Stock Market Response to Unexpected Earnings
Announcements. The Journal of Financial and Quantitative Analysis, 28(4), 565-577.
doi:10.2307/2331166
Feder, A., Gandal, N., Hamrick, J. T., & Moore, T. (2017). The impact of DDoS and other
security shocks on Bitcoin currency exchanges: Evidence from Mt. Gox. Journal of
Cybersecurity, 3(2), 137-144. doi: 10.1093/cybsec/tyx012
Feng, W., Wang, Y., & Zhang, Z. (2018). Informed trading in the Bitcoin market. Finance
Research Letters, 26(November 2017), 63–70.
Nakamoto, S. (2008). Bitcoin: A peer-to-peer electronic cash system. Available at
https://Bitcoinorg/Bitcoin.pdf.
Salas, J. M. (2010). Entrenchment , governance , and the stock price reaction to sudden
executive deaths. Journal of Banking and Finance, 34(3), 656–666.
Urquhart, A. (2017). Price clustering in Bitcoin. Economics letters, 159, 145–147. doi:
10.1016/j.econlet.2017.07.035
INTL. J. APPL. Res. MANAGE. & ECON., 3 (1):10-20, 2020
18
6. Appendix
6.1. The list of 30 Bitcoin hacks
Date
Name
Company /
Victim
Amount Stolen
($K)
Amount
Stolen
03/02/12
Linode Webhost Cloud Server
Hack
Bitcoinica
228
46,703
05/06/12
Bitcoinica Hot Wallet Hack
Bitcoinica
90
18,548
09/04/12
Bitfloor Exchange Hack
Bitfloor
240
24,000
11/16/12
2012 Trojan Wallet Hack
Individual
40
3,457
01/11/13
Vircurex Exchange Hack
Vircurex
23
1,666
03/10/13
BTCGuild Mining Pool Hack
BTCGuild
58
1,254
07/15/13
Just Dice Incident
Just Dice
121
1,300
10/24/13
Input.io Wallet Hack
Tradefortress
820
4,100
11/19/13
BIPS Payment Services Hack
BIPS
1,000
1,295
11/30/13
Picostocks Cold Wallet Hack
Picostocks
6,000
5,875
03/04/14
Flexcoin Hot Wallet Hack
Flexcoin
600
896
03/11/14
CryptoRush Hack
CryptoRush
600
950
01/04/15
Bitstamp Hot Wallet Hack
Bitstamp
5,000
19,000
01/28/15
796 Exchange Hack
796 Bitcoin
230
1,000
02/15/15
BTER Cold Wallet Hack
BTER
1,750
7,170
02/19/15
Kipcoin Exchange Hack
Kipcoin
690
3,000
05/22/15
Bitfinex Hot Wallet Hack
Bitfinex
400
1,400
04/07/16
Shapeshift Exchange Hack
Shapeshift
200
469
05/13/16
Gatecoin Hack
Gatecoin
112
250
08/02/16
Bitfinex Security Breach
Bitfinex
72,000
120,000
10/14/16
Bitcurex Exchanges hack
Bitcurex
1,500
2,300
04/26/17
Yapizon Exchange Hack
Yapizon
7,600
3,831
06/29/17
Bitthumb Hack and PII Leak
Bitthumb
985
390
12/06/17
NiceHash Exchange hack
NiceHash
60,000
4,700
04/13/18
Coinsecure Exchange Hack
Coinsecure
3,300
438
09/20/18
Zaif Exchange Hack
Zaif
38,000
5,966
11/21/18
NicholasTruglia SIM Swapping
Hack
Individual
1,000
230
11/26/18
Bulgaria Crypto Hack
Individual
5,000
1,370
12/27/18
Electroneum Wallet Hack
Electroneum
800
250
INTL. J. APPL. Res. MANAGE. & ECON., 3 (1):10-20, 2020
19
6.2. The regression result A
The regression result from Stata of the price one day before the hack (and the price two days
before the hack) on the trading volume (ch2_0 and adv, ch3_0 and adv).
INTL. J. APPL. Res. MANAGE. & ECON., 3 (1):10-20, 2020
20
6.3. The regression result B
The regression result from Stata of the price one day before the hack (and the price two days
before the hack) on the actual amount of the lost value. (ch0_1 and StolenAmountK, ch0_2 and
StolenAmountK).