Bachelorarbeit, 2021
43 Seiten, Note: 1,3
1 Introduction
2 Background information
2.1 The cryptocurrency ecosystem
2.2 Regulation of relevant crypto exchanges
3 Data
4 Empirical Evidence
4.1 Benford's Law
4.1.1 General
4.1.2 Pearson’s Chi-Squared Test for Benford's Distribution
4.1.3 Statistical Results
4.2 Clustering at key psychological numbers
4.3 Volume Spike Analysis
4.4 Discussion of statistical results
5 Incentives, Perpetrators and Impact
6 Measures to reduce washtrading
7 Conclusion
8 Outlook on future research
9 Bibliography
This thesis makes use of statistical methods such as the Pearson’s Chi-squared analysis to find significant deviations in the first significant digit distributions in the historical transaction record of selected regulated crypto exchanges compared to Benford’s Law. The analysis of trade size clustering behavior at key round numbers is used in order to detect possible signs of washtrading, followed by the volume spike analysis, where the correlation between the four exchanges in terms of rise and fall of their volume is carefully observed.
Aloosh and Li (2019) and Lin et al. (2021) suggest a divergence between regulated and unregulated exchanges in regards to the washtrading activity, in the sense that most regulated exchanges seemed to confirm most statistical analysis while many unregulated crypto exchanges have shown signs of significant violations.
Opposed to these findings, the focus will lie on regulated crypto exchanges only, for which partly abnormal patterns are in fact found, at least regarding the first significant digit distribution. Furthermore, the various regulatory frameworks for the selected exchanges are illustrated, consisting of Gemini, Bitstamp, Kraken and Zaif. The centre of attention will then shift to showing off possible incentives for the various parties to engage in washtrading in the first place. The thesis lays out how these activities distort exchange ratings and the connected metrics as well as aid in creating illegal schemes such as pump and dumps.
List of Abbreviations
Abbildung in dieser Leseprobe nicht enthalten
With the rise of cryptocurrencies over the last decade a new set of crypto focused exchanges and brokers as well as countless new cryptocurrencies emerged. While the market cap of cryptocurrencies rose from a miniscule amount during the founding days to nowadays with an estimated market cap of over $2 trillion as of 25 August 2021 (Coinmarketcap (2021)), the surrounding ecosystem was benefiting immensely. Since early on there was little regulation, low exchange standards and low barriers to perform scams. Bad actors were therefore poised to benefit in the semi-anonymous and unregulated market (Alexander and Douglas (2020)).
One way these bad actors are seeking to benefit in illegitimate ways is via the practice of washtrading, where an actor is clearing his own limit orders so that he can gain an unfair advantage. With this, there was and still is a way to create large amounts of fake liquidity on either an exchange or in general specific altcoins traded on crypto exchanges. However, washtrading is strictly forbidden in the USA. The practice was outlawed in 1936 with the passage of the CEA. It is further defined under the US Securities and Exchange Act of 1934, administered by the SEC. Section 10(b) and matching SEC rule 10b-5 have broad anti-fraud provisions. The crypto currency ecosystem in general is the subject of Chapter 2.1.
Washtrading is not limited to the crypto space according to Imisker and Tas (2018). This is not surprising since traders may facilitate washtrading through exchanges without the knowledge or consent of the respective exchanges. The important difference between the crypto space and traditional stock brokers is still immense in both the volume of the facilitated washtrading as well as the amount of actors involved. There is also a profound difference in regulatory clarity and scrutiny, since many crypto exchanges are barely regulated and often don't even have any local headquarter or any sort of KYC process, including those which are regulated to some extent (the biggest crypto exchange Binance until recently being one of them). This thesis will only focus on regulated exchanges and compare those among themselves and it is therefore of great importance to exemplify in great detail the regulatory frameworks for the four selected exchanges in their respective jurisdictions as well show their regulatory compliance (Chapter 2.2).
Chapter 3 then outlines the origin of the corresponding data sets used as well as their robustness and adjustments made to it. The focus lies on the potential data footprints fraudulent actors generate with varying forms of washtrading based on this data. Trading data of several regulated exchanges (Kraken, Gemini, Bitstamp and Zaif) are made use of for the trading pairs BTC/USD and ETH/USD and subsequently undergo various statistical analyses to detect potential signs of washtrading and also the magnitude of it. This includes the inspection of whether the data’s first significant digit distribution follows Benford's law because this relation is reliably found in most big financial datasets (Benford (1938)). If the dataset in question has a statistically significant deviation from Benford’s Law then there is reason to suspect fraudulent activity such as washtrading. It is well known from countless data sets over the decades that Benford's Law does work for the proposed data sets but it is not ultimately proven why it works. It could be explained due to the logarithmic nature of data for example but it has yet to be proven mathematically. Bolton and Hand (2002, S.237) have also proven that it is very difficult to fabricate data so it appears to be a Data set complying with Benford's Law. Accordingly, the actual versus expected data will then be compared by means of a Chi-Squared-Test, determining if possible deviations can be explained by random chance or not (Chapter 4.1).
Furthermore, the observation that there is a spike of trades clustering at psychological and round numbers if the market has little manipulation and consists of mainly authentic (human) traders can be made. It is a well documented phenomenon consistently occurring in markets with low manipulation and high regulatory standards, caused among many other things by ill informed and irrational investors and the tendency to round numbers (Aitken et al. (1996), Mitchell (2001) and Lawrence (1991)). If this clustering does not occur at these price levels in any meaningful way, there appears to be an increased likelihood of manipulated trade prints which could be caused by washtrading among other things. This is because washtrading is normally facilitated by trading bots which have no irrational tendency to cluster their trades at psychological numbers. The more washtrading activity, the less clustering at these key psychological levels should be observable (Chapter 4.2). It should be mentioned here, that the lack of clustering can also be caused by other phenomena. The empirical part finally concludes in Chapter 4.3 with the contemplation of volume spikes across different exchanges. Since all four exchanges are deemed regulated in their respective jurisdictions, the daily trading volume should show a very high correlation when it comes to daily trading volumes, meaning if exchange X has a 90% spike in volume on a random day, then the other exchanges in general should exhibit similar behavior ceteris paribus. All empirical results are summarized in Chapter 4.4.
Incentives for conducting washtrading include the inflation of trading volume on exchanges to improve exchange rankings. Apparent liquidity created on these exchanges via washtrades will attract more customers who trade authentic volumes. Legitimate exchanges on the other hand have an unfair disadvantage if they do not engage in the same bad practices as their volumes look relatively lower compared to others. Exchanges do regularly also indirectly incentivise their customer base to conduct washtrades with so called trade games. Not only do exchanges have incentives to conduct washtrading but so do individuals with an interest in certain altcoins (especially those with low liquidity). The generation of trading volume on illiquid altcoins is often used to inflate the apparent liquidity of an altcoin and or prompt a buying frenzy. Bad actors in crypto washtrading may thus also be more than one different exchange to conduct multi-exchange washtrading schemes. Individuals with big shares of different (alt)coins or big capital are another potential party conducting washtrades since they may want to create false liquidity and volume for a small cap altcoin they hold substantial amounts of.
Chapter 5 sets forth with stating the various incentives of washtrading, the perpetrators and the overall negative impact it has on the ecosystem. Accordingly, tools and policies are proposed as possible solutions to reduce the impact of this problem (Chapter 6). Finally, Chapter 7 concludes the results of this thesis.
On 31 October 2008 in the wake of the financial crisis the bitcoin whitepaper named “Bitcoin: A Peer-to-Peer Electronic Cash System” was published by an individual known under the Pseudonym Satoshi Nakamoto (Satoshi Nakamoto (2008)). Soon thereafter on 3 January 2009 the Genesis block of the bitcoin blockchain was mined, which marks the start of what we now understand of the cryptocurrency ecosystem.
One of the initial goals of bitcoin in particular was to eliminate the middleman of electronic transactions, namely banks and payment processors. The fact that these systems rely mainly on trust in these third party institutions and that transactions are always reversible, driving up costs, was mentioned in the whitepaper by Satoshi as some of the main criticisms. This would’ve been enforced by the great financial crisis of 2008, during which many people lost trust in the conventional banking system. There were other criticisms on the conventional banking system such as the inflationary nature of most FIAT currencies but these don’t have a profound effect on the issue of washtrading.
With the invention of bitcoin, there was now a way to circumvent these middlemen on a global scale. No longer was there a need for trust but rather a “Proof of work” mechanism to determine which transactions are valid. In May 2021 cryptocurrencies broke a new record with a combined market cap of over 2,5 trillion USD without any crypto related stocks such as miners or exchanges taken into account. With the massive increase of new capital in the market in the past few years, many new parties got interested in participating in the ecosystem. Over time sophisticated cryptocurrency mining operations got started and the respective companies are today traded at exchanges world wide. Another set of crypto related companies were the exchanges over which much of the crypto trading is facilitated today. They also function as an onramp of new capital and thus paradoxically the traditionally centralized exchanges are quite contrary to the initial idea of decentralized cryptocurrencies.
In recent years there was also an emergence of so-called decentralized exchanges such as Uniswap, Sushiswap, Kyber or 1Inch. These can be understood as P2P marketplaces. In contrast to centralized exchanges, these platforms are non custodial, meaning the users still hold their private keys. There is no central authority whatsoever but rather smart contracts programmed to self execute requested trades.
The focus however is primarily on trading activity at centralized exchanges, since in addition to non affiliated individuals practicing washtrades, a large number of centralized crypto exchanges themselves facilitate washtrading in order to boost up their volume and exchange ratings. According to Le Pennec and Fiedler (2021), the trading volume of crypto exchanges with evidence of washtrading is exceeded by a factor of 25-50 compared to the trading volume of accurately reporting crypto exchanges. They also found that a very large percentage of the volume of these investigated exchanges was based on washtrading activity. Lin et al. (2021) found that the total volume on unregulated crypto exchanges is inflated by 70% on average. This can be explained due to unregulated exchanges often not having to report to any central regulatory authority in order to be able to keep their business running. This leads naturally to a substantially higher incentive for very small and unregulated exchanges to conduct in washtrading activity themselves in order to prop up their ratings and rankings - and that is exactly what has been observed for many years in the crypto market. The fact that there is a substantial volume of washtrading activity is broadly accepted in literature (Aloosh and Li (2019), Lin et al. (2021), Victor and Weintraud (2021)) although the actual magnitude of this washtrading activity is still the subject of debate. Furthermore, there is still the question which parties make up what proportion of the volume. On the other hand, there is also an incentive to lower this activity and be more transparent due to the possibility of attracting more professional traders or institutions and building a good reputation, benefiting the company in the long term.
The (lack of) regulation of cryptocurrency exchanges has been controversial for a long time with numerous exchanges being hacked without insurance such as the at the time biggest exchange by volume Mt. Gox in early 2014. Other regular issues include the lack of anti-money laundering regulation such as Know Your Customer. For example, until recently in 2021, the biggest crypto exchange in the world Binance did not have mandatory KYC requirements to trade and withdraw cryptocurrencies on their platform. Currently Binance is under investigation by the Internal Revenue Service and the Justice Department in the USA. Other countries such as Singapore, UK and the Netherlands are holding any business operations of Binance down due to a lack of regulation and concerns of money laundering. Binance is just one example of many exchanges in the crypto space not being properly regulated even though it is by far the biggest exchange in the space with estimates of over 50% of spot and derivative trading volume (Tokeninsight (2021)). There are countless examples of crypto exchanges conducting washtrading on a massive scale, resulting in trading volumes of 50% and more made up by pure washtrading. Fu and Wu (2019) identified more than 10 crypto exchanges, including HitBTC, EXMO, MXC, Biki, Bgogo, Bibox, DragonEx, LBank, CoinEgg and CoinMex, making up more than 70% of their total trading volume with washtrades. Another devastating report was pitched to the SEC in the same year by Fusaro and Hougan from Bitwise (2019), stating that the actual volume only made up 4,5% ($273 million) of the reported volume ($6 billion) and fewer than 5 exchanges (including Gemini, Bitstamp and Coinbase) reporting actual non inflated numbers. These research articles sent shockwaves through the industry, resulting in many of the bigger exchanges promising more transparency and open access to data. International regulators such as the SEC knew that something was to be done against this blatant fraudulent activity.
While the sheer percentage volume of washtrading may have gone down in recent years, the space is still far from free of washtrading. Another more recent example of a crypto exchange grossly exaggerating numbers, was south korean crypto exchange Coinbit, one of the back then biggest exchanges which manipulated up to 99% of its total transactions volume between August 2019 to May 2020 (Seoul Shinmun (2020)). In the case of Coinbit, this fraudulent activity netted them over 100 billion won ($84,26 million) in total.
Nowadays, several countries and states like New York have provided clear frameworks for the regulation of crypto exchanges seeking to operate in the State. New York in particular, has one of the strictest and thought-out regulations to date. In June 2015 the NYDFS issued the first virtual currency regulation, 23 NYCRR Part 200, under the New York Financial Services Law. Approved companies acquire a BitLicense with which they are able to conduct their business legally in New York, though under the strict guidelines of the regulation. There are several strict metrics which need to be met by the approved exchanges including a documented business continuity and disaster recovery plan which is independently tested annually, a independent exam by the NYDFS, annual penetration testing, annual audits, taking fingerprints and photographs of employees with access to customer funds, strict capitalisation requirements set at NYDFS’s discretion, full reserves of custodian assets, cyber security and anti money laundering requirements and the submission of audited financial statements including income statements, statement of assets/liabilities, insurance, and banking on a set basis. Gemini and Bitstamp both obtained a BitLicense in the State of New York in october of 2015 and april of 2019 respectively and can in turn both be classified as regulated exchanges. However, this regulatory framework is only relevant for crypto exchanges wanting to engage in trading operations in New York such as Gemini or Bitstamp.
Kraken is another exchange from which historical trade prints are extracted and thus, will also be of interest regarding their regulatory status and subsequent classification. Kraken has not obtained a BitLicense in New York so far. In turn, they also do not operate in the state of New York but in the rest of the USA apart from Washington state. The decision to not obtain a BitLicense to operate in the New York state can be explained by economic motives and should not be treated as a sign of non compliance. Staying compliant with the New York regulation represents high operational and compliance burdens as well as extra costs for the company which may exceed the potential payoffs from entering the highly competitive New York market. This can only be compensated with very high revenues in return which they may not be able to achieve. Since Kraken is not active in New York, we will have to look at other regulatory frameworks to determine if Kraken can be considered regulated in the framework of the thesis. In the USA, Kraken is licensed as a MSB under the Financial Crimes Enforcement Network, thus complying with legal and regulatory requirements for the areas in which they operate.
These requirements include maintaining long records of currency exchanges and money transfers over a certain amount, reporting suspicious activity of customers, reporting large currency transactions for cash-in and cash-out transactions, strict identification and verification requirements, various anti-money laundering obligations and identifying people with significant ownership or controlling stakes in the company. Furthermore, Kraken Bank Processing, a bank charter recognized under federal and state law, is the world's first Special Purpose Depository Institution, and regulated by the Wyoming division of banking. Going forward, Kraken will also be treated as a regulated exchange in this thesis.
The regulatory status of exchanges engaging in trading operations outside of the USA and New York like Zaif will have to be investigated in the respective country they operate in. Zaif is based in Osaka and one of the oldest crypto exchanges in Japan. Even before there was a regulatory framework in Japan for crypto exchanges, they helped the regulators establish one. It became one of the first licensed crypto exchanges in Japan in 2017 under the Japanese Financial Services Agency. For them to stay compliant, they have to fulfill similar strict metrics as companies wanting to obtain a MSB in the USA and Zaif is thus classified as a regulated exchange as well.
Much of the historical cryptocurrency trading data is accessible through free API’s. Cryptodatadownload.com provides much of the data needed to do robust statistical analysis for free, sourced from the websites of the cryptocurrency exchanges mentioned in the data sets or from the cryptocompare.com API. More data such as volume is sourced from tokeninsight, another well known and reliable data analysis firm in the crypto space.
No trading data is sourced from coinmarketcap.com since it was recently acquired by Binance and could suffer from bias. Data will also be obtained from crypto exchanges directly. For example Kraken offers their historical trade prints for free. For obvious reasons will this method of obtaining data only be offered on regulated and professional exchanges. It is to be expected that these kinds of datasets show less signs of manipulation and washtrading, though it can still occur if users of the platform do so. Data of several trading pairs is included in the analysis such as BTC/USD and ETH/USD with the goal to identify differences between them. The same goes for a range of different regulated exchanges which include Kraken, Bitstamp, Gemini and smaller ones such as Zaif.
The timeframe of the data sets is focused on the first quarter of 2020 since it is a relatively recent timeframe and march of 2020 back then had one of the highest trading volumes for bitcoin in history. This surge in volume and with it renewed interest in cryptocurrencies could’ve contributed to a higher than usual incentive to washtrade in order to prop up volume for exchanges or low liquidity altcoins.
The various exchanges are sorted by size and all of the chosen exchanges are examined regarding their regulatory status since it has to be ensured that the chosen exchanges do comply with regulatory standards. Size and scale alone isn’t a guarantee for high regulatory standards which is what can be seen for example with low regulatory standards for the biggest crypto exchange by volume, Binance, with a market share of about 50% at the current time (cryptocompare.com (2021)).
On the other a very small exchange such as itBit is overseen by the New York Department of Financial Services and registered as a Bank. These regulatory differences have an arguably bigger influence on the likelihood of finding evidence of washtrading, hence the inclusion as an extra metric. The trade files are split up into quarters due to the size of the data sets. They include the trade date, the trading pair, the trade price, the amount, the type of order (sell or buy) and the transaction ID.
However, small gaps in the dataset can and will occur due to downtime of the API’s provided by the exchanges. This can be explained by maintenance downtime for example. Since these gaps are very small and seldom this shouldn’t change the result of the statistical models, especially since they run over a dataset of a very long timeframe with hundreds of thousands trades.
The data is getting collected by polling the market data rest API which is made public in an exchange’s API documentation. Ignoring the small gaps already mentioned, every executed trade is listed in the data set. The specific exchanges are polled in regular time frames to ensure that every trade is collected. The dataset contains sell and buy orders so all sell orders were deleted to avoid duplicates. Furthermore, it also contains “block” transactions and transactions from auctions which were not deleted since they can be considered regular transactions and will most likely be of authentic nature. A further measure to avoid faulty results, is to delete all transactions under a value of 1 US Dollar.
[...]
Der GRIN Verlag hat sich seit 1998 auf die Veröffentlichung akademischer eBooks und Bücher spezialisiert. Der GRIN Verlag steht damit als erstes Unternehmen für User Generated Quality Content. Die Verlagsseiten GRIN.com, Hausarbeiten.de und Diplomarbeiten24 bieten für Hochschullehrer, Absolventen und Studenten die ideale Plattform, wissenschaftliche Texte wie Hausarbeiten, Referate, Bachelorarbeiten, Masterarbeiten, Diplomarbeiten, Dissertationen und wissenschaftliche Aufsätze einem breiten Publikum zu präsentieren.
Kostenfreie Veröffentlichung: Hausarbeit, Bachelorarbeit, Diplomarbeit, Dissertation, Masterarbeit, Interpretation oder Referat jetzt veröffentlichen!
Kommentare