Work in Progress- Cryptonomics: Investment behaviour in the cryptocurrency market

-Work in Progress- Cryptonomics: Investment behaviour in the cryptocurrency market By Mohamad Saleh Abstract This paper continues the work established by others in the research field observing the behaviour of investors in the cryptocurrency market. Through a thorough review of the related literature, the paper will propose no new impeccable evidence but highlight the mountain of circumstantial ones. The paper offers a detailed technical background and an overview of the related behavioural literature. Policy implications and regulations are also briefly discussed. Keywords: cryptocurrency, blockchain, Bitcoin, cryptomarket II List of Tables and Figures Table 1 Public and private blockchains (Coinbase, 2017) ............................................ - 9 - Table 2 Top 10 cryptocurrencies and ICO Tokens price changes. ............................. - 10 - Table 3 Cost of mining a single Bitcoin in 2017 per country ...................................... - 11 - Table 4 The benefits and drawbacks of cryptocurrencies ............................................ - 13 - Table 5 Behavioural biases.......................................................................................... - 16 - Table 6 Cryptocurrency exchange fees, location and adherence to laws .................... - 26 - Table 7 Frequency of trading in cryptocurrency ......................................................... - 29 - Table 8 Bitcoin price changes based on next week predictions. ................................. - 29 - Table 9 Confidence in beating the market in the next 3 months ................................. - 30 - Table 10 T-test of counter-reactions after overreaction days for Bitcoin prices. ........... - 31 - Table 11 Estimated illegal users count ........................................................................... - 32 - Table 12 Proposed determinates of cryptocurrency prices ............................................ - 35 - Table 13 Weak form efficiency in Bitcoin results. ........................................................ - 42 - Table 14 Determinates of cryptocurrency value ............................................................ - 43 - Fig. 1 Bitcoin mining pools (BTC.com) ........................................................................ - 12 - Fig. 2 Actual exchange market-share compared to survey respondents’ choices. ........ - 24 - Fig. 3 Cryptocurrency regulation around the world ...................................................... - 46 - Equation 1 Bitcoin hash function .................................................................................... - 7 - Equation 2 Scrypt hash function ...................................................................................... - 7 - Equation 3 Average past returns .................................................................................... - 36 - III Table of content List of Tables and Figures ........................................................................................................ III 1. Introduction ....................................................................................................................... - 1 2. Technical Background....................................................................................................... - 6 2.1 The Blockchain ....................................................................................................... - 6 2.2 Cryptocurrency ........................................................................................................ - 9 2.3 Initial coin offerings (ICOs) .................................................................................. - 13 3. Behavioural background ................................................................................................. - 15 3.1 Representativeness .................................................................................................... - 16 3.2 Overconfidence bias .................................................................................................. - 18 3.3 The disposition effect ................................................................................................ - 19 3.3 Regret aversion and the fear of missing out .............................................................. - 20 3.4 Other biases ............................................................................................................... - 21 5. Method ............................................................................................................................ - 22 6. Literature review ............................................................................................................. - 23 6.1 Market characteristics and their potential effect on investors ................................... - 23 6.2 Investor characteristics and observed behaviour ....................................................... - 28 6.3 Market sentiment, Privacy, Crime and Volatility ..................................................... - 32 7. Discussion ....................................................................................................................... - 36 8. Policy implications .......................................................................................................... - 45 9. Conclusion ....................................................................................................................... - 47 References ........................................................................................................................... - 49 Appendix ............................................................................................................................. - 58 - IV 1. Introduction This paper is an attempt to bridge the behavioural finance findings and research in cryptocurrency as a new medium and asset class for investment. By illustrating similarities between traditional individual investors in the stock market to investors in the cryptocurrency market, this paper can serve as an introduction or theoretical guide for future studies on investors’ behaviour and bias. Behavioural finance has long been established as the evolution of modern finance (Haugen, 1999; Nofsinger, 2016; Shefrin, 2006; Thaler, 1999). Modern finance relies on key principles such as the homo economicus assumption, which implies that the average investor is only interested in self-enrichment (Andrikopoulos, 2005), the efficient market hypothesis (Schwert, 2002), the rationality of the individuals while making investment decisions and the lack of bias in individuals’ future predictions (Andrikopoulos, 2005; Nofsinger, 2016). Behavioural finance provides robust empirical evidence for the existence of psychological biases in investors’ decisions, which ultimately creates market anomalies, such as crashes and bubbles. A price bubble typically describes a sharp increase in an asset’s price, this increase cannot be explained by an increase in the intrinsic value of the asset. There are many historical examples of market bubbles and subsequent crashes, for example, the Tulips market in the Netherlands, the Mississippi shares in France, the South Sea shares in the UK, the market bubbles that caused the crashes in the American stock exchange in 1923-1932, the Mexican, Gulf and Asian stocks crashes in the 1970s, 1980s and 1990s. The Dotcom fad of early 2000 and most recently the 2008 financial market crisis (Menschel, 2002, p. 47). Behavioural finance literature proposes ways to lower investors’ susceptibility to the different biases. Thus, lowering the risk of investors making unsound decisions based on emotions rather than rational reasoning (Nofsinger, 2016, p. 6). The selected biases discussed in this paper are as follows: the representativeness heuristic (Kahneman & Tversky, 1972), the overconfidence bias (Shefrin, 2006), the disposition effect, regret aversion (Chen, Kim, Nofsinger, & Rui, 2007; Nofsinger, 2016) herding behaviour and the overreaction (Menschel, 2002). Throughout the course of this paper, a correlation between the aforementioned biases and behaviours exhibited by investors in the crypto market such as holding or panic selling will be established. Psychological biases have been observed in varying degrees in all demographics, experienced and inexperienced, individual and institutional investors. Evidence suggesting that increased awareness of the bias can help minimize its effect on the individual. There is evidence that increases in experience and awareness can help investors make more rational decisions -1- (Chen et al., 2007, p. 429). This is particularly relevant for investors in the crypto markets, as preliminary evidence found in the literature and presented in this paper show that, on average, investors in the crypto markets may be lacking the necessary awareness and experience to minimise bias. Investors in the crypto market are likely to be more prone to psychological biases compared to stock market investors, as Shefrin (2006, p. 170) notes that nearly only 22% of all invested funds are done directly by individual investors while 60% are purchased through a broker, insurance agent, financial planner, bank representative or through a fiduciary. A fiduciary is a trustee that administers or manages investments, property or funds. Their behaviour is different when they do not invest with their personal funds. Fiduciaries are prone to different sets of biases when they invest with their own funds and when they invest with others’ money. Risk assessment and the level of risk aversion differs according to the source of the money invest. Engaging in risky investments may be caused by the pressure from the need to perform well for their clients (Brooks, 2008, pp. 6–7). Despite the importance of fiduciaries in institutional investment, there is no evidence that cryptocurrency investors rely on fiduciaries for their investments. Cryptocurrency exchanges are designed to enable direct investment by being as accessible as possible. They typically employ no barriers to entry. This is indicated by the fact that the majority of investments in the crypto market appear to be performed directly by individual investors. Until recently, most traditional financial institutions had shown little interest in or voiced concern around the negative impact of cryptocurrency and the high risk associated with investing in them. This changed in December 2017, when CME group and the Chicago Board Options Exchange (CBOE) launched the first Bitcoin futures contracts (Coindesk, 2017). This move marks a clear indication of a slow but growing interest in cryptocurrency from established financial institutions. The characteristics of the crypto markets will be discussed later in this paper to highlight the high risk faced by investors from unregulated, insecure and fraudulent investments, coin offerings and entire market platforms. The effect of pump-and-dump1 on smaller cryptocurrencies, insider trading, market manipulation and increased regulation that may hinder innovation will also be discussed. Cryptocurrencies are self-contained and unique financial systems, transacting tools and value stores. They are also considered a medium of exchange, as units of account which serve as immutable assets that cannot be counterfeited (Alvarez-Ramirez, Rodriguez, & Ibarra1 A form of securities fraud. P&D groups artificially inflate the prices of stocks or tokens in order to sell cheaply obtained assets at a much higher price. Token prices spike then crash after the dump is completed. -2- Valdez, 2018, p. 955). Unlike fiat money or commodity currencies, cryptocurrencies are neither controlled by a central authority nor tied in value to a tangible substance (Macdonell, 2014). They were characterized in the past as fiduciary currency, deriving their values solely from exchanges with fiat currency or on the belief that others would accept them as payment for a product or service (Venegas, 2017). Cryptocurrencies have also been described as tradable assets and as long-term securities, for example, by New Zealand regulators (Coindesk, 2018). Cryptocurrencies can also be considered a hedge against central bank monetary policies or as digital gold with counterfeit protection. Additionally, they offer a solution to privacy and a protection from identity theft (Mokhtarian & Lindgren, 2017, p. 23). The term cryptocurrency is generally used to describe all coins or tokens that rely on cryptography for the security of their networks. Cryptographic encryptions secure the transactions and networks from hostile attacks in the form of double-spending. Ownership and transaction records are time-stamped and stored on a public record called the blockchain. Ultimately, there are three archetypes of tokens according to the purpose they were designed to serve as currency, utility and investment tokens (Hacker & Thomale, 2017, p. 23). A more accurate use for the term cryptocurrency would be limited to describing cryptographic digital currencies, which employ the functions of secure value exchange and long-term storage. Bitcoin, Monero, Litecoin and Bitcoin Cash, despite being designed to serve mainly as currencies, are still used as a speculative digital asset, media of exchange, in addition to payment systems and they may still serve other non-monetary purposes (Hileman & Rauchs, 2017, p. 26). Other cryptocurrencies that expand the definition of the term, such as Ethereum, Lisk, Cardano and Eos tokens, are more versatile in the functions they serve on their blockchain-powered programming platforms. These platforms enable a wide range of uses, such as distributed applications (dApp), digital asset management, cross-border money transfers and smart-contracts. In 2017, dApp tokens are used by 400 projects on the Ethereum platform alone (Hileman & Rauchs, 2017, p. 27). Smart-contracts are decentralized, computerized, automated contracts that serve to include all contractual agreements, including terms of agreement and enforcement, in a peerto-peer agreement, which is then recorded on the public blockchain without the need for mediators. These peer-to-peer smart-contracts are initiated without the need for or involvement of a third-party (Sklaroff, 2017). Theoretically, the need for a central authority, such as a law or court to secure and verify the contracts is therefore eliminated (Mokhtarian & Lindgren, 2017, p. 7). Smart-contracts enable companies to autonomously negotiate directly with anonymous customers and fulfil their obligations automatically, by delivering goods or services once the payments are fulfilled (Ciaian, Rajcaniova, & Kancs, 2018, p. 174). Initial coin -3- offerings (ICOs), rely entirely on smart-contracts between the issuing company and the buyers (Hacker & Thomale, 2017, p. 10). Another category of cryptocurrencies are the tokens issued by distributed ledger technology (DLT) companies. These tokens are used to operate either on permissioned blockchains or on public blockchains, mainly offering corporate services such as R3’s Corda, which is a DLT used for complex transactions with restricted access to transaction data (Coindesk, 2017) or Ripple, which facilitates real-time gross settlements for banks (Conley, 2017, p. 1). This paper focuses on all tradeable cryptocurrencies listed on exchanges, regardless of their use-cases or functionality, therefore the terms token and currency will be used interchangeably. The concept of a decentralized currency powered by a blockchain was first published in a white paper during the financial crisis in 2008. The paper was published under the pseudonym Satoshi Nakamoto (Nakamoto, 2008). In the paper, the authors propose a new electronic or virtual peer-to-peer currency, Bitcoin, which would rely on cryptographic encryption to establish a model of market trust. This method would be a solution to the double-spending problem by time-stamping all transactions. Double-spending is the act of multiplying, replicating and spending the same funds more than once (Dwyer, 2015, p. 82), akin to counterfeiting. A mathematical model and proof were provided to illustrate the improbability of the success of a hostile attacker on the network (Nakamoto, 2008, p. 7). Peer-to-peer transactions would eliminate the need for traditional third-party systems to serve as mediators to the transactions. Unlike bank transactions, which rely on a central system component to maintain security and trust. The proposed system would depend on a network of nodes, sometimes referred to as full nodes or miners, to verify every transaction in exchange for a small reward. This reward functions as a form of incentive for users to actively participate in the network. A maximum supply of Bitcoin was set at 21,000,000 units. With diminishing mining rewards over time, the maximum supply should be reached by the year 2140. The maximum supply and mining reward would serve as a control mechanism for price inflation. Records of funds and transactions would be time-stamped, and then stored in a public record called the blockchain (Nakamoto, 2008). The limitations of Bitcoin, such as transaction cost and time are among the biggest problems facing its mainstream adoption (Back, Corallo, & Dashjr, 2014, p. 4). At its current implementation, the Bitcoin blockchain is limited to processing around 7 transactions per second at its theoretical maximum capacity, the actual average has fallen from 300,000 transactions per day in 2017 (Hanl & Michaelis, 2017, p. 364) to 200,000 transactions per day -4- in June 20182 (Blockchain). Other cryptocurrencies such as Litecoin’s 56 and Ethereum’s 15 transactions per second are far below Visa’s maximum capacity of 56,000 transactions per second (Visa, 2015). While there are solutions for the scalability limits, such as off-chain transactions, increased block size and sidechains for Bitcoin (Wright, 2017), some of these solutions have already been adopted by other cryptocurrencies, such as the upgrade to larger block size by Bitcoin Cash. The problem remains that no single cryptocurrency has, thus far, been able to capture the same attention as Bitcoin or provide the capacity to offer the same efficiency and convenience of traditional centralized financial systems such as Visa or Swift (Bano, Al-Bassam, & Danzis, 2017). The success and appeal of Bitcoin paved the way for different implementations of the same technology to take place. The source code for Bitcoin is publicly available free of charge in an open-source format for anyone to change or improve upon. Thousands of similar cryptocurrencies exist in the market today3. These alternative cryptocurrencies are commonly referred to as Altcoins. Some of these Altcoins function similarly to Bitcoin, others offer extra benefits, such as increased anonymity and privacy. Some cryptocurrency innovations focus on the blockchain technology itself for completely different applications, such as distributed storage, e-voting and smart-contracts. The majority of Altcoins serve no clear purpose and offer no innovation or novelty (Hileman & Rauchs, 2017, p. 15). Bitcoin is often compared to gold as a value-store or as an asset. This comparison is warranted by the purchases of Bitcoin as a long-term investment, value-store, or retirement fund (Hileman & Rauchs, 2017, p. 26). By studying Bitcoin’s transaction and network volumes, Courtois (2016) concludes that Bitcoin’s users tend to consider it an asset rather than a currency. Zach Pandl, a Goldman Sachs analyst and strategist, wrote a report in January 2018 describing Bitcoin as a long-term major asset, stating that “…digital currencies should be thought of as low/zero return or hedge-like assets, akin to gold or certain other metals.” (Fortune). A unit of Bitcoin is composed of 100,000,000 Satoshi, just as a Euro is composed of 100 Cents, and can be traded in fractions of a single unit (Hayes, 2017, p. 1310), for example, 1 unit of ETH = 0.07950030 Bitcoin (Coinmarketcap, accessed on 06.06.2018). In addition to the transaction costs, the reward for mining Bitcoin was 12.5 units per block in June 2018, valued at nearly $100,000. At its highest value, the mining reward for Bitcoin in December 2017 was $250,000 when the unit price reached an all-time high of $20,000 (Coinmarketcap). This mining reward is divided amongst active miners based on their share in the mining effort. Miners participate in the verification of transactions by competing to find the solution to 2 Offchain transactions that take place on exchanges are not recorded on the blockchain. 3 Coinmarketcap.com lists 1651 unique cryptocurrencies (accessed on 08.06.2018). -5- the encryption puzzle. When a miner finds the solution to the puzzle, they broadcast the solution to the network for confirmation. Upon confirmation, all transactions are time-stamped and stored in a new block in the chain. The time required to create a new block is called block time, it is the average time after which all confirmed transactions in a time interval are processed and stored in a block, for Bitcoin, every 10 minutes a new block is added to the chain. Sequential blocks make up the chain of blocks, hence the name, blockchain (Kuo Chuen, Guo, & Wang, 2017, pp. 18–19). The main purpose of the blockchain is to maintain a public permanent record of all verified and confirmed transactions. Some cryptocurrencies operate on public blockchains while others have their own. Blockchains enable trustless transactions, where a mediator is required to assure trust while offering a level of anonymity, often not offered by centralized financial systems. Cryptocurrency prices are known for their price volatility, the blockchain technology is often misrepresented and misunderstood. The crypto market experiences recurring bubbles and crashes. This suggests that amongst the many proposed determinates of volatility, a lack of rationality in investors’ decisions cannot be excluded (Cheah & Fry, 2015, pp. 34–35). The first part of the paper will provide a technical background on cryptocurrencies, blockchains and initial coin offerings. The second part is an overview of psychological biases adopted from the behavioural finance literature. This will be followed by a review of related literature and a discussion of the potential implications for both the investors and policymakers. The paper concludes with suggestions for further research to establish preliminary findings to answer the following questions: Are investors in the cryptocurrency market affected by behavioural biases? Do the cryptocurrency market characteristics (design and functionality) increase investors’ susceptibility to behavioural biases? Do crime, privacy and market sentiment increase cryptocurrency price volatility? 2. Technical Background This part discusses the technical background of the blockchain technology, the differences between cryptocurrencies and other tokens. Highlighting the potential uses of the technology. 2.1 The Blockchain The blockchain is a distributed ledger that stores all records of transactions and funds. When a user initiates a transfer of funds, a transaction hash is created by the wallet software or by the exchange market. The transaction hash is then broadcast to the entire network and all miners begin competing for the cryptographic solution in order to verify the transaction. The -6- verification process is done by performing a large series of calculations. Once a miner or node in the network finds the key to unlocking the hash, it broadcasts the solution back to the network for confirmation. The miner then creates a new block with all confirmed transactions and the block reward is issued. Each block is inextricably linked or chained to the previous block. Sovbetov (2018, p. 4) presents the SHA-256 hash function used in Bitcoin as 𝐻𝑎𝑠ℎ 𝑜𝑓 𝑐𝑢𝑟𝑟𝑒𝑛𝑡 𝑏𝑙𝑜𝑐𝑘 = 𝑓(𝜃, 𝜙, 𝑍) Equation 1 Bitcoin hash function where θ is the hash of the previous block, ϕ is current mining difficulty and Z is a random key specific to the new block. The required time to successfully mine a single block is called block time. For Bitcoin, the entire network should spend roughly 10 minutes to find the cryptographic solution. The mining difficulty adapts to the ever-increasing computational power of the network by increasing the difficulty of the cryptographic puzzle. Alternatively, the Scrypt hash function used by Litecoin is 𝐷𝑒𝑟𝑖𝑣𝑒𝑑 𝑘𝑒𝑦 = 𝑓(𝑃, 𝑆, 𝑁, 𝑟, 𝑝, 𝑑𝑘𝐿𝑒𝑛) Equation 2 Scrypt hash function Where P is predetermined passphrase, S is randomization factor short for salt, N is the computational cost spent must be larger than 1, r is block size, p is parallelization parameter and dkLen is the intended output key length. Mining cryptocurrency is not meant to find particular results but rather a particular output. Computing machines compete to calculate the hash output, which must begin with a set series of zeros. This is considered a waste of energy spent on empty calculations (Norvill, Pontiveros, State, Awan, & Cullen, 2017). When more than one miner finds the same decryption solution at the same time, two blocks are created and added to two different chains. This is called a fork. The way this issue is resolved is that the longer chain will always be the one that new blocks would be added to. Longer chains imply that more computational and mining effort was spent on creating them. The cryptographic code powering the Blockchain is the real asset behind every cryptocurrency. Users are assigned pairs of encryption keys, public and private keys that are used in the process encryption and decryption. The public key is considered an address and can be shared safely with others, much like an account’s number. The private key is similar to account passwords or PIN codes in securing the account from unauthorized access. Transactions are only authorized when the users provide their private and public keys. Additional security measures are also employed by most wallet providers, such as additional phone verification, two-step verification and email verification. An encrypted hash of the public keys of the sender and receiver as well as the senders private key is created when a transaction is initiated. Once the transaction has been verified and confirmed by the network, it is time-7- stamped and added to the blockchain. The transaction cost is not affected by the size of the transaction, which is an advantage to large value transactions and a drawback when the value is possibly smaller than the transaction costs. The cost of a Bitcoin transaction was around $1 in June 2018. However, on December 23rd, 2017 the transaction cost reached an all-time high of $55 (Bitinfocharts). The mining or verification step relies on different protocols4 to organize the network consensus. Protocols are the set of rules, conditions and instructions upon which the network functions. The most widely implemented mining protocol is called Proof-of-Work (PoW) (Kornmesser, 2008, p. 3). It was first proposed by Nakamoto (2008, p. 3) and is used by most cryptocurrencies including Bitcoin and Ethereum. PoW mining requires immense computational and electric power, for example, Bitcoin’s estimated annual electricity consumption is 71.12 Terra Watts per hour (Digiconomist). Environmental groups have widely criticized this as an essentially useless waste of computational power5. An alternative protocol named Proof-of-Stake (PoS) has been proposed and planned to replace PoW on some networks to minimize the environmental impact. Members of the network that use the PoS protocol acquire the right to generate new coins by committing some of their funds to the network, effectively stacking their assets (Eyal, 2017, p. 46). There are no block rewards, but transaction fees can serve as an incentive to active miners. There are several other protocols, such as Proof-of-Importance employed by NEM and the distributed open source consensus ledger used by Ripple. Not all cryptocurrencies can be mined, among the largest 10 cryptocurrencies, 6 are not mineable Ripple, EOS, Stellar, Cardano, IOTA and TRON. Their tokens cannot be mined and can only be acquired through direct purchase from an ICO or an exchange (Coinmarketcap). The blockchain technology offers many opportunities for innovation, the limits of which cannot yet be determined. The Bitcoin code is offered in its open-source form for all interested developers to build upon. Hundreds of blockchain technology start-ups emerged in the cryptocurrency market since 2016. The way they utilize the blockchain technology varies, for example, bank settlements, cross-border money transfers, computer resource utilization, distributed storage, predictions and wagering, education, voting and more (Conley, 2017). There are essentially two types of blockchains that offer different functionalities, as detailed in Table 1. Bitcoin along with the majority of cryptocurrencies operate on public 4 Cryptocurrencies such as Bitcoin and Ethereum can be considered the protocols upon which the system functions. Mining protocols are relevant only to miners. Since mining costs are arguably the biggest driver of value it is important to highlight the basic concepts. 5 All mining efforts spent by unsuccessful miners is waste without reward. A crucial element in the protocol is assuring that enough computational power is spent over a time interval. This assures that hostile miners cannot take over the network by mining faster, as they would be required to redo all preceding calculations. -8- blockchains. Private blockchains are more tailored to corporate and institutional uses (Coindesk, 2017, 2018). Public blockchain Private blockchain Network Decentralized Centralized Trust New models Accepted models Access Permissionless Permissioned access only Security Open network Approved participants Identity Anonymous or pseudonymous Known identities Asset Native assets Any asset Speed Slow Fast Table 1 Public and private blockchains (Coinbase, 2017) 2.2 Cryptocurrency Cryptocurrencies utilize one function from the blockchain technology, which is the storage and transfer of value with low cost and faster transaction times (Creyts & Trbovich, 2018). The term cryptocurrency is widely misused to describe all blockchain applications and protocols. In reality, only tokens that serve as value-stores and media should be recognised as a cryptocurrency. Instead, most cryptocurrencies employ many of the characteristics of a commodity or security (Hileman & Rauchs, 2017, p. 97; Mokhtarian & Lindgren, 2017, p. 11). Based on their particular use, they can serve as an alternative currency, a commodity or a speculative asset (Bartos, 2015; Mokhtarian & Lindgren, 2017). Unlike fiat currencies, which are legal tenders issued by a government or a central authority, such as the Euro, cryptocurrencies do not require a central authority to act as the trusted middleman. Records of ownership and transaction history are stored on the distributed public ledger, that is the blockchain. This entails a lack of central authority, control, management, a repository of information and most importantly a lack of a central point of failure (Bouri, Gupta, Tiwari, & Roubaud, 2017, p. 87). In 2018, Bitcoin was still unable to function as a daily general-purpose payment currency. The main challenges are its limited scalability, relatively small transaction capacity, increasing transaction costs and slow processing speed. The demand for Bitcoin did not slow down over the years, driving the price of a single Bitcoin to an all-time high of $20,000 in December 2017 (Coinmarketcap; Athcoinindex). The use of Bitcoin as a speculative asset and commodity may have had an effect on increasing its price volatility, in turn, this increases the challenges it faces in being adopted as an alternative currency (Bukovina & Martiček, 2016, p. 10). Price volatility, frequent market -9- bubbles and crashes are a real concern for traders accepting cryptocurrencies as payment for services and goods, see Table 2 for detailed price and market capitalization changes. As the price and demand for Bitcoin increased, so did transaction costs, which increased from a few cents to almost $50 per transaction (Coinmarketcap). The use of Bitcoin as a daily transacting currency especially for transactions of small amounts became considerably more expensive. Several companies that previously accepted Bitcoin suspended payments for their services with cryptocurrencies (Coindesk, 2018). Market cap Market cap YTD 2018 Bitcoin 220.9 130.3 Ethereum 69.77 Ripple Bitcoin Cash Name Price Price YTD 2018 58.99 13170.18 59.14 84.76 82.20 25.17 41.53 Price Δ ATH 7633.7 58.99 20,089.00 721.66 592.43 84.76 1,432.88 30.62 2.12 0.64147 30.62 3.84 18.42 44.36 2459.32 1073.52 44.36 4,355.62 4.95 13.27 268.08 8.64 14.81 268.08 22.89 Litecoin 12.00 6.99 58.27 220 123.13 58.27 375.29 Cardano 18.03 5.89 32.64 0.695418 0.227029 32.64 1.33 Stellar 5.76 5.53 96.07 0.322342 0.297671 96.07 0.93814 IOTA 9.57 5.24 54.76 3.44 1.88 54.76 5.69 TRON(ICO) 2.38 4.02 168.74 0.036185 0.061062 168.74 0.30036 EOS(ICO) Table 2 Market cap Δ Top 10 cryptocurrencies and ICO Tokens price changes. Note. All prices are in US Dollar on 03.05.2018. YTD is Year-To-Date price, ATH is all-time high. Market capitalization values are in the billions. Cryptocurrencies use cryptographic algorithms to maintain the security of their networks and transactions, in addition to the mining protocols, they also employ mining algorithms, for example, SHA-256 in Bitcoin, Ethhash in Ethereum and Scrypt in Litecoin. Each node in the network is assigned a unique address with unique private and public encryption keys. These keys are used to digitally sign every transaction the peer performs. Due to the extreme complexity of the encryption, reverse-engineering the encrypted hashes remains impractical. Users of cryptocurrencies can obtain units of the currency by purchasing them from exchanges, or through the process of directly mining of the currency when possible. When Bitcoin was first conceived as a peer-to-peer payment system, any individual could use their personal computer to participate in the network mining efforts. Miners contribute their computational power and electricity costs and their reward is newly issued coins. This has since changed due to the increase in network difficulty. This difficulty is - 10 - calculated by the number of processing units working concurrently on the network. More specialized hardware was developed to streamline the mining process, earning early adopters thousands of coins with millions of dollars in value (Bohr & Bashir, 2014, p. 97). Country Cost China 3,172 Georgia 3,316 United States 4,758 Mean (World) 7275,9 Max (South Korea) 26,170 Min (Venezuela) 531 Mean (Weighted) 3998 Annual consumption (TWh) 71.12 global mining revenues $5,697,530,703 global mining costs $3,440,394,685 Cost percentage 60.38% Per-transaction cost KWh 973 Market price 7548 Table 3 Cost of mining a single Bitcoin in 2017 per country Note. The weighted average is calculated as the average cost of mining in China, the United States and Georgia. Research from Cesco based on 3 miners AntMiner S9, AntMiner S7 and Avalon 6 requiring 14641.884, 37803.9816 and 45552.276 Kilowatts to mine a single Bitcoin in December 2017 (Coindesk, 2018; Digieconomist; Cesco; Coinmarketcap, accessed on 06.06.2018). The mining requirements for Bitcoin changed dramatically over the years. Initially, Bitcoin was mined using the computational capability of the central processing unit (CPU) of the computer, then graphical processing units (GPU) were used for their increased processing power. Soon after Application-specific integrated circuits (ASICs) were developed, the mining difficulty increased significantly such that neither CPU nor GPU miners were able to mine Bitcoin profitably. Bitcoin mining is highly centralized in countries with low electricity cost, such as China, the U.S. and Georgia (Coindesk, 2018; Cesco), detailed in Table 3. Mining centralization is a problem for any cryptocurrency, especially Bitcoin, as it restricts the flexibility of adapting new innovations or changes, this can also lead to the slow response to the changing needs and demands of their users. Large mining operations in China dominate Bitcoin mining (IEEE). In order to stop this from happening to their networks, other cryptocurrencies, such as Ethereum and Monero, restricted the use of ASIC miners to avoid a similar eventuality. - 11 - Fig. 1 Bitcoin mining pools (BTC.com) Some miners consider mining to be a source of income, not an investment. Miners are especially affected by price trends, mining difficulty and increased regulations. Large and smaller miners alike started to pool their resources together in what is referred to as a mining pool. Bitcoin mining pools are also very centralized, 81% are in China, 10% in the Czech Republic, 2% in Iceland, 2% in Japan and 1% in Russia. Further detailed in Figure 1 and Table A.8. More than 1600 cryptocurrencies are currently available on the market (Coinmarketcap), most of them lacking novelty or innovation. The term Altcoin can be used to describe cryptocurrencies that lack innovation (Hileman & Rauchs, 2017, p. 15), although it is also used to describe all cryptocurrency beside Bitcoin (Ciaian et al., 2018). There are quite a few cryptocurrencies that provide privacy as their primary point of appeal, these are usually referred to as privacy coins. They facilitate encrypted and anonymous transactions while protecting the identities of their users. The growing fear from privacy coins is supported by their increased use in ransomware attacks, illicit trade, criminal activities, and financing terrorist and extremist groups, such as ISIS and neo-Nazi groups. Privacy coins provide identity and privacy protection from government and corporate intrusions, they are seen as a tool for aiding democracy, freedoms and dissidents against tyranny or dictatorship (Guardian). Their importance is similar to the onion router (Tor), which provides encrypted and anonymous access to the internet. Tor is crucial to connect to the unindexed or hidden part of the internet called the darknet, which exists within the deep web. Tor can be used for all sorts - 12 - of browsing activities, legal or illegal. Table 4, summarizes the benefits and drawbacks of using cryptocurrencies. Benefits Drawbacks § Relatively low transaction costs. § § Transactions can be anonymous or quasi-anonymous. § Protection against identity theft and credit card fraud. § Enable illicit and criminal activity. § Available on a global scale. § Extreme price volatility. § Nearly zero barriers to entry. § Lack of institutional support. § Trust in mediators is no longer needed. § Markets are constantly targeted by hackers (Hileman Increased regulation is sending negative signals to both developers and investors. & Rauchs, 2017, p. 37). § Secure transactions. § Wide range of uses. § Irreversible transactions. § Supply is controlled through network consensus. § A threat to traditional financial systems. § Open source. § The cost and negative environmental impact of mining § Promotes democracy. § High risk of losing funds or access to funds if access are immense. codes are lost or forgotten. § Anonymous transactions cannot be tracked or stopped. Transactions are publicly available, is a benefit as it provides a public record to protect the users. At the same time, this is a drawback as it exposes users’ balances and transaction history to criminals. The benefits and drawbacks of cryptocurrencies Table 4 Note. The presented points are mainly adopted from Hurlburt and Bojanova (2014, p. 13), which focuses only on Bitcoin but most of the points apply to other cryptocurrencies. New coins are created at stable rates to control inflation, which means that scarcity and supply shortages are not an issue to worry investors. networks are set to automatically scale the mining difficulty according to the increasing computational power of the network. There are also milestones that change the network reward schedule, for example, Bitcoin halves its block reward every 210.000 blocks. When the maximum supply of Bitcoin (21,000,000 Bitcoins) is reached, which is anticipated to happen in the year 2140, the mining rewards in the form of new units will stop. Transaction fees will be the only incentive to continue mining. By factoring in the increasing mining difficulty and its probable effect on price, cryptocurrency prices are likely to increase significantly whenever mining rewards decrease. 2.3 Initial coin offerings (ICOs) ICOs are seen as a method of fundraising used by blockchain technology and fintech start-ups to generate capital and awareness of their proposed services and products. Issued coins are often referred to as tokens, which further increases confusion around the technology. Token functionalities vary according to the services they provide, often exceeding the basic functions of a currency. ICOs are similar in concept to initial public or private offering of stocks and - 13 - securities (Mokhtarian & Lindgren, 2017, p. 8). According to Conley (Conley, 2017, pp. 1–2), start-ups use ICOs as their approach to raising funds. ICOs became the equivalent to crowdfunding for blockchain projects in recent years. Instead of relying on venture capital for financing the proposed projects, start-ups can directly sell their tokens to interested investors through smart-contracts. Ethereum is by far the most utilized platform for conducting ICOs. Start-ups publish white papers containing all the relevant technical and financial proposals as well as the number of tokens to be issued. The companies also detail the pricing model and target goal. As of 2018, ICOs raised a cumulative $12 billion in funding. $6.3 billion was raised in the first quarter by 202 ICOs, the largest of which is Telegram’s TON. This is significantly higher than the total funding raised by 342 ICOs in 2017. Venture capital has also invested around $3 billion in ICOs, $885 million of which was invested in 146 ICOs in the first quarter of 2018. ICOs are widely unregulated and have proven to be highly successful in raising funds. Platforms such as Cardano and EOS raised billions by selling tradable tokens. Neither has shown any viable products yet. The number of ICOs is staggering and despite efforts from governments to quell their spread, for example, China, South Korea and Japan, most social media sites opted to ban related advertisements to protect users from potential scams, these include Facebook, Google and YouTube. ICOs can be completed by individuals with the required technical skill. This relative ease led to the creation of thousands of fraudulent ICOs (Corbet, Larkin, Lucey, Meegan, & Yarovaya, 2017). Similar to cryptocurrencies, tokens can generally be divided into three categories: currencies, distributed applications (dApps) and smart-contract protocols (Corbet et al., 2017). The fields in which these innovations are used are diverse, for example, storage, proxy, dynamic routing and payments services proposed by TON, which is developed by Telegram (Telegram, 2018). NEO, known as the Chinese Ethereum offers a platform for decentralized smartcontracts. NEM is an enterprise-grade platform that facilitates Smart Asset Systems. WAVES offer decentralized exchanges amongst other functions of its networks platform. Another example is Stellar, which is a cross-border financial service that payment and financial institutions offer their customers. Tokens serve various purposes on their native platforms, validation, transaction, spam protection, voting rights, proof-of-stack, and providing additional privilege access rights to the platform to develop additional tools, as is the case on the Lisk development platform for example. Another important application for tokens is to serve as an investment for venture capital, such as decentralized autonomous organization (DAO) tokens. Hacker and Thomale - 14 - (2017, p. 25) also mention that in a sample of 253 ICOs, 26% of their tokens entitled their owners to profit rights. Ethereum is regarded as a global decentralized and distributed computer, or what is referred to as a “Turing complete”, which is an all-purpose virtual machine capable of executing any kind of algorithm or process. This includes executing smart-contract protocols (Kuo Chuen et al., 2017). The first example of a decentralized investment fund or crowdfunding was the DAO. The DAO tokens were then used to vote on investment proposals (Venegas, 2017). Tokens are identical to coins in principle, the difference being that sometimes they can be used to purchase services on the platform exclusively. Since tokens are used in the same manner as coins, they can be purchased, traded, saved and used as a long-term investment despite the fact that they are riskier in principle. Unlike coins, tokens are tied to unlaunched products or services. On more than one occasion, buyers have not been protected from scams. ICOs sell fake tokens at launch to collect money then disappear, which constitutes a high risk to unaware investors (Kotas, 2018, p. 28). With the exception that some tokens can be mined, ICO tokens are usually directly issued or purchased from the issuing company or through a distributed application platform, for example, Ethereum. The majority of the cryptocurrencies can be mined (Coinmarketcap). A mineable cryptocurrency has an inherent value attached to the cost of mining it, which is composed of computational power spent, electricity and maintenance costs. The cost of mining a single Bitcoin averages 60% of its market price, around $4000 in 2018 (Digiconomist), see Table 2. Whether unmineable cryptocurrencies have an intrinsic value aside from their exchange potential is debatable. similarly, ICO tokens also rely on market movements and investors’ demand to assign them value. Tokens are sold on the promise that a service or a product will be available in the future. Similar to stocks that are difficult to value, prices in the crypto market are prone to be greatly affected by market sentiment (Kuo Chuen et al., 2017). 3. Behavioural background In the following section, an overview of behavioural biases is presented. The importance of studying cognitive biases and their connections to investment decisions has long been established. Chen, Kim, Nofsinger and Rui (2007, pp. 426–428) mention how behavioural biases can skew investors’ choices towards making less optimal or less rational decisions. In their paper, they focus mainly on overconfidence, the disposition effect and representativeness biases. Ramiah, Xu and Moosa (2015, p. 91) and Pompian (2012) review and study many behavioural biases and how they lead to momentum trading, bubbles, panic and crashes as can be seen in Table 5. - 15 - Anchoring and Adjustment Availability Cognitive dissonance Conservatism Confirmation Overconfidence Representativeness Self-attribution Illusion of control Overreaction Ambiguity aversion Endowment effect Self-control Optimism Mental accounting Underreaction Hindsight Loss aversion Recency Regret aversion Framing Status Quo Belief perseverance Herding The Disposition effect Table 5 Behavioural biases. Shefrin (2006, p. IX) writes, psychology is the basis for many of the errors committed by people and for understanding the driving forces that lead people to over- and underestimate market indicators. These indicators maybe “perceptual illusions, overconfidence, overreliance on rules of thumb, and emotions”. Ramdiah, Xu and Moosa (2015, p. 91) discuss how behavioural finance can explain some of the questions left unanswered by “neoclassical” finance. Such as why irrational anomalous behaviour persists. They also note how investors are affected by personal taste, preference and other psychological factors. This leads to differences in investment opportunities, despite herding behaviour’s tendency to unify tastes. Investor biases are many and their behaviour can often be explained by more than one bias, the selected biases were most commonly used to relate to the cryptocurrency market. 3.1 Representativeness Representativeness heuristic describes making misjudgements based on the misperception or misinterpretation of a problem or a situation, for example, how historical price movements, either for winning or losing, stocks may serve as indicators for the continuation of the price trends in the future. In other words, a past winner is expected to continue to yield high returns while past losers are expected to continue to perform poorly (Shefrin, 2006, pp. 82–84). In the stock market, investors tend to believe that previously winning stocks will continue to - 16 - remain winners and that losing stocks will remain losers. In fact, previous loser stocks tend to yield larger gains after periods of downturns, as observed over a three-year period by De Bondt and Thaler (1985). Kahneman and Tversky (1972) illustrated how people extrapolate false information from sentences6. Additionally, some investors suffer from the gambler’s fallacy. They expect previous trends to reverse in the future. they base their decisions on the expectation that past random occurrences constitute patterns with predictable changes7. Representativeness can lead to biased decisions, either overreaction or underreaction to extrapolated information (Shefrin, 2006, p. 52). The overwhelming optimistic sentiment of most cryptocurrency investors is an indication of their belief that winners will continue to win, and downturns must reverse despite or because of the volatile conditions in the crypto market. This belief is enforced by the apparent one-directional and upwards historical price trendline. Since the inception of Bitcoin, the market has continued to grow and prices to increase with few pauses. Market bubbles and crashes do occur on a more frequent basis, thus enforcing the sense of perpetual unending growth. Nofsinger (2016, p. 87) describes momentum investing as the reaction of magazines and newspapers readers to reports of high performing stocks and how these stocks receive more interest afterwards. The author’s explanation is that some investors chase winners. Momentum investors look for past winners that performed well in the past quarter or the past hours. Most crypto exchanges show price changes in real-time as well as providing price movements and trade volumes. Some exchanges provide a view of the past few weeks others provide the entire trade history of the cryptocurrency. Odean (1998, pp. 1893–1894) discusses how, regardless of the source of information, individuals would still draw inferences from any new information. The author notes that the accuracy, reliability, and whether the source is based on facts or opinions or whether the information is up-to-date has little effect on the perceived predictive value of the information. Individuals also underestimate the significance of sample sizes when considering the findings and the general tendency of extreme outcomes’ regression to the mean. Individuals also regard information that confirms their beliefs more highly than information that might contradict those beliefs or convictions. This behaviour is consistent with the confirmation bias and increases investors’ tendency to overreact or underreact to new information based on the information’s appeal and level of ambiguity. Representativeness can also describe the “illusion of seeing 6 “Linda is 31 years old, single, outspoken, and very bright. She majored in philosophy. As a student, she was deeply concerned with issues of discrimination and social justice, and also participated in anti-nuclear demonstrations.” (Barberis & Thaler, 2002, p. 1066). When asked to choose the more likely answer between A) Linda is a bank teller and B) Linda is a bank teller and is active in the feminist movement, most respondents choose answer B. 7 A fair coin toss that lands 4 times on the same face would be predicted to land on the other face in the following attempts. - 17 - patterns in a random walk” (Andrikopoulos, 2005, p. 10) or the tendency to base assessments on superficial and unrealistic probabilities (Brooks, 2008). 3.2 Overconfidence bias Self-deception occurs when individuals perceive their personal skills to be as good as, or better than, the average (Chen et al., 2007). Investors suffering from overconfidence would attribute past successes to their personal abilities and losses to bad luck. Overconfident investors tend to make riskier investments, especially after periods of consecutive successes. This can result in a reduction of realized welfare as cumulative trading costs increase with high investment turnovers. Realizing profits on investments further increases investors’ overconfidence, which in turn increases investment aggressiveness, leading to riskier choices in a cycle that ultimately harms the investor. Increases in overconfidence are correlated with increases in trade volumes, especially in high return stocks. This increase can lead investors to overreact. Overreaction to increased trade volumes can, in turn, reinforce overconfidence over the short run by overstretching market trends (Hirshleifer, 2015, pp. 136–138). Barberis and Thaler (2002, p. 1066) argue that overconfidence is the result of two underlying biases, self-attribution bias and hindsight bias. Self-attribution bias is the tendency to credit oneself with successes, regarding personal talent or skill as the cause of said success and attributing failures to bad luck. This can be compared to managers’ tendencies to attribute good performance to their personal abilities (Hirshleifer, 2015, p. 140). Overconfidence is enforced by their continued seemingly good performance, which may increase investment in riskier stocks. In the crypto market, high returns on investments cannot always be attributed to personal talent or skill. Due to the recurrence of price bubbles and extreme volatility, high returns may be a side effect of a market trend. It is also worth noting that the majority of investors in the crypto market are overwhelmingly male (Coindesk, 2018) since male investors are significantly more overconfident than female investors (Shefrin, 2006). Men are found to be trading 45% more often than women and earn lower returns (Barber & Odean, 2001; Hong, 2007, p. 8; Nofsinger, 2016, p. 13). Overconfidence is also enforced by the hindsight bias, which describes the process of believing that a given event could have been accurately predicted after the fact. Investors suffering from hindsight bias believe they had correctly predicted changes. However, this happens as they misremember making such predictions (Odean, 1998, p. 1893). Believing in one’s ability to predict the future may lead to exaggerated optimism, underestimation of risk, illusions of control and foresight superiority. This is often observed in - 18 - financial experts’ opinions and advice during extreme market movements, as assurances are given to other investors to stymie panic (Ireland, 2018, p. 16). 3.3 The disposition effect The disposition effect describes the tendency for investors to hold on to losing stocks for too long. Investors hope that they can resell the stocks during market upswings to recoup their losses. Investors miss out on loss minimization and purchase opportunities at lower price points while they wait for the prices to rebound. Waiting too long and refusing to react to changes may render the investment a sunk cost (Hirshleifer, 2015, p. 137). This inability to react according to price changes is not limited to inexperienced investors or traders. The disposition effect was observed in highly experienced traders as well as amateur investors (Hirshleifer, 2015, p. 140). The disposition effect also explains the tendency to sell winning stocks much quicker, as investors tend to sell winning stocks much faster than losers. Once an investor realizes a stock or a trade is profitable, they are more likely to commit to the sale sooner, regardless of current market trends. This can lead to an opportunity loss in the form of potentially missing out on longer upswings where a stock’s price might have continued to rise. The two parts of the disposition effect may be observed separately in different investors. As different investors may be more prone to exhibit only one side of the behaviour, for example, they may sell both winners and losers too soon or hold on to winners and losers for too long (Chen et al., 2007). Furthermore, the disposition effect can be explained in part by regret and risk aversion. As in the need to “get even” in order to avoid the regret of having made the wrong decision. Shefrin (2006) attributes prospect theory’s risk assessment for the disposition effect, citing users’ risk-seeking behaviour over losses and risk aversion over gains. Investors’ anticipation of the market to repeat past trends or regress to previous means may explain the fear of selling too soon. Some investors consider shorting the market as a riskier gamble and instead hold on to their losing assets. The disposition effect explains why investors would keep holding losing stocks for a long time while selling winning stocks much faster or as soon as they rebound. The explanation is the hope of investors getting even before getting out, or as described by Shefrin (2006, p. 107) as “getting evenitis”. A common phrase used by cryptocurrency traders is “Hold on dear life”, which is the common advice to hold on to one’s assets when the market appears to be heading downwards. The rational intuition would be to sell the assets and later recoup the losses by purchasing the same or other assets when the price bottoms. Holding in the crypto market appears to be the dominant strategy of the large majority. Many investment experts would advise against holding - 19 - and setting levels for stop-loss selling points to minimize short-term losses. Many investors keep holding losing investments for too long (Barberis & Thaler, 2002, p. 1104; Hong, 2007). The fact that the market has historically always rebounded, is accepted as evidence that holding is the rational investment strategy by holders. Anchoring may also have a role in encouraging early adopters of cryptocurrencies to hold. For example, Bitcoin was initially very cheap or even free to mine or purchase (Coinmarketcap). For early adopters, current prices constitute immense gains and the market uptrend promises bigger gains in the future. Promoting holding strategies becomes more prominent during market downswings, where the general mood is pessimistic and fearful sentiments start to spread through the cryptocurrency community, it is unclear whether this is done to hinder the spread of panic or from underreaction to changes. 3.3 Regret aversion and the fear of missing out Regret aversion describes individuals fear of making decisions that may prove to be suboptimal than the status quo. By avoiding making a decision, individuals avoid taking responsibility and feeling regret. Investors holding on to losing investments for too long is done to avoid admitting past mistakes. Regret aversion also explains the hesitation to invest in undervalued opportunities after periods of negative investment outcomes. Additionally, it may stop investors from selling winners even when signs of imminent market reversals exist. The fear of making the wrong decision can hinder the investors’ ability to act rationally and may push timid investors towards investing in good companies rather than good opportunities (Pompian, 2012, pp. 227–231). Regret aversion can also cause herding behaviour as the number of investors contributing to the “mass consensus” can limit the feeling of personal responsibility for future regret. Market bubbles are often created when investors lose their grounded perception of potential gains and engage in high-risk investments hoping for quick and high rewards. “Collective excitement” leads investors to choose portfolios of stocks based on sentimental hopes rather than on facts or grounded information (Krafft, Della Penna, & Pentland, 2018). During market bubbles, the real value of the traded items is completely detached from the inflated market price. During market bubbles, trading volumes increase rapidly, and investor overreactions shortly follow. Increased media attention and the activity of first-time investors may increase the level of excitement. The fear of missing out on the opportunities of making big gains can drive investors to make irrational decisions. Enforced by a collective consensus, regret aversion becomes the drive for making riskier investments. As the prices become more artificially inflated, the chance - 20 - of making bigger profits increase. As long as the investors realize their winnings before the collapse of the market they should enjoy their profits. On the other hand, unfortunate uninformed investors will either panic or hold on to losing assets for too long in the hope that the downturn to revert. 3.4 Other biases Bubbles occur more often in the crypto market due to the higher degree of volatility and unparalleled optimism about the future adoption of cryptocurrency (Leclair, Voia, Khalaf, Strathearn, & Leclair, 2018, p. 14). In December 2017, the price of Bitcoin reached the $20,000 mark for the first time. This price was more than 5 times Bitcoin’s average mining costs at the time. Similar to historical market bubbles, this price exaggeration may be enforced by irrational herding, overreaction and several other factors. Overreaction Ramiah et al. (2015, p. 93) describe overreaction as investors’ overestimation of present information and assigning less weight to past information. Overreaction is explained by individuals’ overestimation of their beliefs and skills, leading them to increase their trading volumes. Overreaction is connected to representativeness where investors would overreact to positive news and underreact to bad news in a bullish market. Overreaction was first recorded by De Bondt and Thaler (1985;1987). The authors found past winning portfolios performing worse than the market while under-performing portfolios over-performing the market over a three-year period. This implies a pattern of price correction and a possible regression to the mean (Caporale & Plastun, 2018). Caporale and Plastun (2018) also find evidence for overreaction in price movements in the crypto market. They also observe positive and negative changes in prices following days of overreaction. They fail to compose a consistently profitable strategy to exploit price overreaction in the market. This can be due to the additional transaction costs or due to an error in the coding of the trading bots. Their realized profits however, were not statistically different from randomized trades. Individual investors are more prone to overreaction than institutional investors. Since most investments in the crypto market are made directly by individual investors it is important to note the possibility of overreaction in the crypto market. - 21 - Herd behaviour and peer influence The expectation that the value of the asset is going to increase is the motive for investment. Without arbitrage, there would be no benefit for investors to engage in seemingly risky trades. When exploitable opportunities exist on a wider scale in the market, investors start to engage in investments based on trending practices. For example, collective excitement about price movements can perpetuate and artificially inflate the market price by increasing the demand for the asset (Menschel, 2002). Herding and peer influence can push investors in the wrong direction, for example, panic selling at the lowest price point during the financial crisis (Pompian, 2012, p. 231). Pack mentality can also affect individuals’ motivations for investment, by relying on the “consensus of the herd” the individual may stop rationally assessing every investment decision they make (Pompian, 2012, p. 235) In the crypto market, holding (Hodl), or holding on dear life translates to holding on funds regardless of price changes. This investment strategy shares many similarities with the disposition effect. Fear, Uncertainty and Doubts (FUD), is a concept used to describe periods of downturns and price crashes, inexperienced investors are prone to panic during these periods, which leads to increased trade volumes, overreactions and further declines in prices. The fear of missing out (FOMO) may be related to regret aversion. Investments made during periods when prices increase too rapidly and become detached from rationality are made to avoid missing an opportunity. When the opportunity cost is too high or too ambiguous, risk-averse investors may prefer not to engage in any trades. Peer influence and herding behaviour can be greatly reinforced by increased trading momentum, a characteristic of crypto market investors (Angelovska, 2016, p. 13). These behaviours can be found in both the stock market and the crypto market. 4. Method The research methodology is a review of the relevant literature on cryptocurrency users and market characteristics, cryptocurrency price determinants, market sentiment and market bubbles. Relevant sources were selected to establish a relation to behavioural finance findings. The reviewed literature is supplemented by additional sources from cryptocurrency reports, websites and discussion forums, for example, CoinDesk, Reddit, Coinmarketcap and Tradingviews. News reports on cryptocurrency criminal activity and regulation were obtained from various sources on the internet and cited accordingly. Behavioural finance books were obtained from the university’s library and are considered the starting point of this research. Research papers were found and accessed through Web of Science, SSRN, RePEc, Google Scholar, and the PUMA databank at the University of Kassel. - 22 - Prices, volumes and trading data were obtained either directly from exchanges websites or from Coinmarketcap.com. Tables and graphs were formatted in Word, Excel and Google Spreadsheet. All statistical summaries were calculated in Stata 14. Survey data was acquired with permission from Daniel Verweij (2018), a bachelor student at Tilburg University. This data was collected from three online social media sources. The total was 478 respondents from Reddit (347), Twitter (116) and Facebook (15). Additional surveys are cited directly from their original sources. All frequency and descriptive statistics were personally calculated. All tables and charts are personally compiled unless when explicitly stated otherwise. 5. Literature review This section is divided into three parts covering market characteristics, investor characteristics, and determinants of cryptocurrency price volatility. 5.1 Market characteristics and their potential effect on investors Hileman and Rauchs (2017) divide the cryptocurrency industry into four categories: wallets, payment service providers, mining operations and exchanges. Kotas (2018, p. 24) adds digital asset management and other services. Wallets are digital applications or services that facilitate the storage and transfer of tokens. Establishing a wallet account or address is often the first step a new user makes. Wallets securely protect users’ funds. Most cryptocurrencies offer their own basic wallet application that works on most operating systems. In addition, there are more sophisticated wallet applications that can manage multiple accounts and are compatible with multiple tokens, for example, Jaxx and Exodus. Some wallets integrate a built-in exchange or a payment system as an extra functionality for the users. The integrated exchange can be a centralised exchange or a brokerage service, a third-party exchange, or an entirely decentralized exchange. The estimated number of unique and active wallet users in 2017 was between 2.9 and 5.8 million addresses (Hileman & Rauchs, 2017, p. 10). Payment service providers perform a set of services with cryptocurrencies or with a combination of fiat and cryptocurrency. These markets serve as a point of sale, they facilitate cross-border money transfer services, offer merchant services such as shopping-cart integration, or provide general-purpose services such as instant, bills or payroll payments to their clients. Some of these payment services provide their users with wallet addresses as well as perform exchange services8. 8 One example is Cryptopay.me, which offers a wallet, prepaid cards as well as Bitcoin exchange services. - 23 - The mining sector is responsible for the maintenance of the integrity of the cryptocurrency’s network and blockchain. Large mining companies have a huge impact on the mining effort and can be considered an integral part of the stability of the cryptocurrency and its prices (Hileman & Rauchs, 2017, p. 21). Mining operators range from smaller miners, often collaborating in mining pools, cloud mining operators and mining giants that manage arrays of mining farms. The largest Bitcoin mining company in the world is the Chinese Bitmain, which operates its own mining farms as well as its Antpool mining pool. Mining pools are web portals that facilitate the pooling of mining resources between miners around the world as long as they have the necessary equipment. Some mining pools provide users with wallets and some incorporate exchanges or payment services (Vries, 2018, p. 804). Due to the increased difficulty of mining Bitcoin due to the introduction of highly specialized mining equipment, most small miners instead rely on mining other tokens before exchanging or selling them off. Full nodes and some miners maintain a full copy of the blockchain. They also actively process network transactions and participate in the confirmation of processed transactions. As the demand for mining increased, miners started to pool their resources in mining pools. These mining pools are semi-automated hubs. A mining pool user can easily connect their mining software to the hub and be part of the pools’ mining effort. Rewards are distributed to the participants of the pool according to different schemes. Some divide the rewards equally or according to spent computational power, or according to breakthroughs in the mining process. Mining pools usually charge a participation fee of around 2% of earnings. Fig. 2 Actual exchange market-share compared to survey respondents’ choices. Note. Left is top cryptocurrency exchanges by trade volume. Right is exchange data collected by Verweij (2018), the most frequently used exchange data is detailed in Table A.6 - 24 - Cryptocurrency exchanges are the primary, and often the sole point of purchase, sale and trade of cryptocurrencies and tokens. Additional services provided by exchange platform include order-book exchanges and brokerage services. The exchanges perform these services in exchange for fiat currencies or other cryptocurrencies. Hileman and Rauchs (2017, p. 30) define exchanges as the marketplace for cryptocurrencies as they offer a place for “trading, liquidity and price discovery”. Trading platforms differ in their adherence to government regulations, 52% of small exchanges hold licenses compared to 35% of large exchanges9. Most exchanges accept deposits in fiat currencies, mainly US dollar, Euro, Chinese Yuan and Japanese Yen. The total number of unique cryptocurrency exchanges is likely to exceed 138 exchanges (Coinmarketcap). Exchanges that do not operate with fiat currencies opt to use a peg token tied to a fiat currency, such as Tether, which is a cryptocurrency that serves as a facilitator for crypto exchanges, especially ones that are having difficulties with establishing banking relationships. Tether is pegged to the US Dollar and believed to have 100% backing by actual reserves10 (Griffin & Shams, 2018, p. 8). Figure 2 is an illustration of the largest cryptocurrency exchanges by number of trade volumes compared to survey responses (Blockchain; Verweij 2018). Most exchanges operate around the clock in a fully automated fashion and with no geographical barriers. Investors from almost anywhere are welcome to engage in trades without limitations or restrictions. Depending on the locations they operate in, some of the exchanges are required to adhere to anti-money laundering (AML) and know your customer (KYC) rules, forcing their users into a comprehensive verification process before allowing them access to the trading tools. Other exchanges, such as decentralized exchange Crypto Bridge, do not require any form of verification from their users. Some exchanges provide anonymous trading services to attract traders and investors who have privacy concerns. Some exchanges go as far as eliminating all identification requirements. An example is Shapeshift, which allows secure, anonymous instant and direct exchange of a selection of cryptocurrencies, Table 6 offers some examples. As for the location of the exchanges, China has the highest number (18), followed by the UK (15), US (8), Japan (6), South Korea (5) and only 4 decentralized exchanges (Cryptocompare). Aside from maintenance downtimes, crypto markets are open for trade around the clock. This design characteristic means that market movements are not tied to opening and closing cycles. 24-hour trading also allows trades to take place concurrently around the world across 9 Only percentages were provided in the original report. Total 51 exchanges. Issues related to Tether are subject to speculation and remains under investigation 10 - 25 - cultures, languages, time zones and borders. These individual differences may have an effect on cryptocurrency prices, trade dynamics and investor decisions as they have on traditional investors (Karolyi & Stulz, 2003, p. 1003). The crypto market is also more affected by global events, for example, news regulations in China on trading Bitcoin can have an adverse effect on prices, which may affect investors around the world. Any investor can sign up to most exchanges and start trading in minutes, except when a verification process is required, which can take from anywhere from a few minutes to weeks. Having a large number of exchanges removes any age, regional or policy barriers. Exchange Fees Location Trading Deposit Withdrawal Bitfinex 0.00%-0.20% $20 $20 Bitstamp 0.10%-0.25% $7.5 $15 HitBTC 0.10% $9 Kraken 0.10%-0.35% $20 Table 6 Incorp. Regulation adherence Management AML-KYC Id. UK Yes Withdrawal UK UK, USA, LUX Yes Funding $9 . . Tiered Withdrawal $20 USA USA Yes Funding Virgin Islands Cryptocurrency exchange fees, location and adherence to laws Note. Place of incorporation and management. Identification requirement and adherence to anti-money laundering and know your customer laws (Pieters & Vivanco, 2017, p. 8). Credit card and bank transfers are only accepted in exchange for a fee. Although withdrawal or trade in fiat is not always offered by the exchange, and recently exchanges have been having difficulties in offering credit card transactions. Most exchanges allow the free depositing of cryptocurrency but charge a withdrawal as well as a per-trade fee. Trade fees vary across exchanges, for example, Binance charges 0.1% while Coinbase charges 1.45%. Other exchanges apply the fees depending on Maker and Taker bids. The trade costs scale according to increases in traded volumes as exchanges offer discounts to large traders. In addition to the withdrawal fee paid to the exchange, there is the transaction fee that is paid to the network’s miners to verify and confirm the transaction. Crypto markets facilitate the trade of cryptocurrencies as well as tokens. Most trading platforms only allow trades paired with or valued against one of the main cryptocurrencies Bitcoin, Ethereum, Bitcoin Cash and Litecoin (Coinmarketcap). This trading mechanism essentially makes the use of these big coins, especially Bitcoin, the centre of the cryptocurrency trading world. Most tokens can only be traded into Bitcoin before being sold for fiat or other cryptocurrencies. Unlike stocks, which can be sold directly for money, some exchanges trade - 26 - tokens exclusively through Bitcoin. Even when the exchange offers direct exchange, purchase or sell they use Bitcoin as a proxy. Tokens are valued in Bitcoin, converted to Bitcoin then exchanged for another token or fiat money. This design decision ensures Bitcoin’s dominance and provides a minimal level of stability to smaller tokens. The main drawback is that the entire market is affected by the smallest changes in Bitcoin and other large cryptocurrencies’ prices and news (Sovbetov, 2018, p. 11). Exchanges face difficulties in making agreements with banks to facilitate cash payments, for example, banks in Chile shutdown accounts associated with cryptocurrency exchanges. Similarly, top Indian banks have suspended or restricted exchange accounts citing fears of money laundering and dubious transactions. This trend of shutting down accounts continues in Bulgaria, Norway, Thailand, Poland and China. In the United States and Canada, the largest banks banned cryptocurrency purchases with credit cards, this move was also followed by banks in the UK, Australia and South Africa (Cointelegraph). Krafft et al. (2018) experiment with trading bots. The authors used trading bots to trade on an exchange in 217 different cryptocurrencies. Using randomly uniform trade volumes, the trading bots would commit to a purchase or sale. The process was repeated over random time intervals while recording the details of the most recent committed trade. The total number of observations was 310,222 accomplished over the course of 6 months. The results indicate strong evidence for peer influence. An observed total 1% increase in the probability of a trade overall. A 2% increase in the probability of a buy in the following 15 minutes after a trade is performed, a significant increase from 28% in the control to 30% in the buy-intervention group. Additionally, the study found an aggregate 7% average rise in total purchases after a bot purchase. An increase of $16,000 or 83 Bitcoins (1513 Bitcoins compared to 1430 Bitcoins in control) were traded in the intervention trials compared to the control. Considering that the total size of the intervention is approximately 0.14 Bitcoin. The effect of the intervention is 500 times larger than the actual trade size. An important point mentioned in the study is that peer influence effects were not observed after sell-interventions, also no long-term effects on market trends were found after 30 minutes after the intervention. Another point raised by Krafft et al. (2018, p. 1) is the effect of cryptocurrency exchanges on exciting investors. Certain aspects of the design and functionality of the exchange interfaces are intended to “promote collective excitement”. Other factors that may have an effect is peer influence and increased early adoption of new cryptocurrencies, typically in the hope that prices would increase rapidly. Small investors are affected by the choice of big investors and endorsements from popular sources or authorities. This is consistent with the data - 27 - collected by Verweij (2018), analysis shows that 67.8% (324) of the investors would refrain from committing to a purchase following advice from peers. The use of trading bots on cryptocurrency exchanges can have an effect on other investor’s choices. In addition to the large impact on trade volumes, trading bots’ decisions can be falsely identified as human behaviour. This can enforce the investors’ irrational decisions. By influencing inexperienced investors towards imitating their choices, trading bots can increase the trade momentum of an overpriced cryptocurrency or advertise obscure cryptocurrency by increasing trade activity. This sort of advertisement can lend a sort of unwarranted legitimacy to potentially dubious tokens. The level of sophistication of programming these bots receive defines their ability to conduct successful trades. Trading bots may have a large effect on the market, as their decisions can have a noticeable effect on trading volumes and investors’ responses (Krafft et al., 2018). Some exchanges place advertisements for the most traded currencies at their front pages, such as Bittrex. Another example is Livecoin, which additionally displays the most pumped, most dumped and sponsored tokens. These advertisements can possibly serve as a nudging mechanism. Other exchanges and cryptocurrency related websites offer investment advice or live chat functionality which can further increase peer influence, for example, Hitbtc and Tradingview. There is enough evidence to suggest that market characteristics (design and functionality) may have an effect on the individual investors, this effect might increase or decrease according to their personal characteristics. On the other hand, market design and functionality could influence irrationality and lead susceptibility to investment biases. 5.2 Investor characteristics and observed behaviour The exact number of cryptocurrency users is hard to estimate precisely, users are free to use multiple unique wallet addresses or multiple wallet applications. Access to wallets can be done through computers, phones and tablets. The option to print out a paper version of the wallet for offline storage may also be available. Some users may leave their funds directly on the exchanges without the use of a personal wallet. Exchange wallets usually pool users’ funds together to minimize security risks. Exchange accounts assign each user with a unique address. The risks associated with relying on exchange wallets comes from the risk of exchange hacks, the potential for loss of access to the exchange, and the fact that most exchanges restrict users’ access to private keys. Token storage is an important issue with the increase in cryptocurrency adoption. According to Hileman and Rauchs (2017, p. 27), more than 10 million people would have owned Bitcoin by 2016. - 28 - Sovbetov11 (2018, p. 9) studies the factors influencing cryptocurrency price. Original findings are detailed with updates based on data from Google Analytics. The study indicates that 96.57% of those involved with cryptocurrency market are Male compared to only 3.43% Female12. The largest age groups 45.71% is 25-34 years and 30.62% between 35-44 years. A recent survey conducted by Verweij (2018) found the average age of investors to be 28.4 years (Min 13 and Max 68). The large majority being male 82% and female 18%. The sample size is 478 respondents, the majority of which are from the United States 182, the United Kingdom 39, Netherlands 38, Canada 29 and Germany 28. 68% of the respondents identify as Caucasian. The sample has a high number of respondents with higher education degree (259) 54% or are in the process of acquiring a degree (82) 17%. The field of study being mainly Engineering (138) 29%, Business or Economics (105) 22% and other (131) 27.4%. When respondents were asked to state the first time they made an investment in cryptocurrency, the majority (279) 58% answered 2017 and (59) 12.3% answered 2018. Frequency of trade Responses Percent Daily 50 10.46 Weekly 158 33.05 Monthly 208 43.51 Yearly 62 12.97 Total 478 100 Frequency of trading in cryptocurrency Table 7 How sure are you of your answer? Bitcoin price change Do not know Sure Not sure Total Increase 9 93 87 189 Decrease 6 51 43 100 No idea 27 146 16 189 Total 42 290 146 478 Table 8 Bitcoin price changes based on next week predictions. Signs of overconfidence can be inferred from the observed high frequency of trade as well as high confidence in price predictions, seen in Tables 7 and 8, this can also serve as an indicator of a high turnover rate that would imply overconfidence (Chen et al., 2007, p. 444). 11 The source of the data mentioned in the paper is www.coin.dance, observed age groups were limited to over 18-year olds. potential involvement from younger age groups cannot be determined. 12 In June 2018 the ratio had slightly changed to Male 91.22% Female 8.78%, age groups 25-34 years to 48.43%, and 35-44 years to 24.96%. - 29 - Unsurprisingly, more than half of the respondents rated their investment performance as average (276) 57.74% while (164) 34.31% rated it as better than average, an indication of overconfidence (Odean, 1998, p. 1892), optimism and wishful thinking (Barberis & Thaler, 2002, p. 1066). Additionally, an overwhelming majority of respondents indicated that they are confident or very confident of their ability to beat the market in a 3-month period, detailed in Table 9. Confidence Frequency Percent Cumulative No confidence 128 26.78 26.78 Yes, some confidence 300 62.76 89.54 Yes, much confidence 50 10.46 100 Total 478 100 Table 9 Confidence in beating the market in the next 3 months Similar results can be found in Coindesk’s (2018) findings, 55% of the near 3000 respondents bought cryptocurrency for the first time in 2017. The majority of them (68%) in the fourth quarter of 2017, 45% of all respondents bought cryptocurrencies at the highest price point in late 2017, which can be a sign of herding. Only 52.8% sold cryptocurrency back compared to 40.2% who continued to hold13. Respondents check the prices every day (56.5%) or every hour (37.3%). When asked whether they believe that the market prices are in a bubble, a cumulative 40% answered “No” (slightly no 13%, moderate no 17% and extreme no 10%) compared to a cumulative 50% for “Yes”. When asked at which level would they stop holding, 60% of the respondents indicated they would continue to hold even if the token lost all its value, 12% indicated that the point would be at a loss of 50% of value and 12% indicated that point at 100%. 86% did not sell due to price dips and 78% bought additional tokens because of the dips. 70% believed that holding on to tokens is more important than spending them. This may indicate that some investors may be suffering from the disposition effect. More than a cumulative 93% predicted the prices of all cryptocurrencies to either go up a bit or go up a lot. 27% hold between $10,000 and $50,000 in cryptocurrency investment compared to 10% $0 to $1,000, 16% “$1,000 to $5,000”, 14% “$5,000 to $10,000” and a combined 33% holding more than $50,000. 37% believed they will become millionaires in 2018 compared to 7% who actually became millionaires in 2017. Bohr and Bashir’s (2014) study is considered to be the first attempt to analyse Bitcoin’s community demographics, behaviour and political leanings, using a dataset from responses to an online survey. They found that the average age of the respondents was 33, nearly half of the 13 It is unclear whether the same respondents performed both the buy and sell trades, no distinction was made in the report. - 30 - respondents lived in the United States. 95% of the respondents were males. 47% identified as libertarian, 17% progressive, 9% socialist, 8% centrist, 7% both green and anarchist and 5% conservative. Evidence from the paper shows that age is significantly correlated with Bitcoin accumulation. Sovbetov (2018) reports billions of dollars have been invested in assets with no history of producing revenue. He further explains that the rise in cryptocurrency prices in 2017 was driven by the herding influence of the increased investment from “other people”. Increased attention brought to the market new investors with vast sums of money. The cryptocurrency total market capitalisation reached an all-time high of $830 billion in January 2018 before quickly imploding in the following weeks. Period 5 10 20 50 Parameter Normal OR Normal OR Normal OR Normal OR Mean 2.13% 2.98% 2.01% 3.53% 2.00% 3.71% 1.92% 4.19% SD 3.48% 5.27% 3.22% 6.00% 3.23% 6.18% 3.20% 6.30% N 1303 296 1327 267 1342 241 1326 227 t-criterion 2.65 4.02 4.19 5.31 t-critical 1.96 1.96 1.96 1.96 Table 10 T-test of counter-reactions after overreaction days for Bitcoin prices. Note. Periods are in days. OR is Overreaction. All reported results are significant (Caporale & Plastun, 2018, p. 14). Caporale and Plastun (2018) study price overreaction after a one-day extreme price change in the crypto market. Using data on Bitcoin, Litecoin, Ripple and Dash, spanning 2013 to 2017, the authors ran several statistical tests on two sample groups. The first group contains data on prices with normal price changes and the other group includes observations after days with abnormal or extreme price changes. The null hypotheses for both research questions regarding the existence and direction of overreaction in the market could be rejected (Caporale & Plastun, 2018, p. 8). The authors could not develop a profitable investment strategy to exploit their findings. Their findings are detailed in Table 10. Based on the reviewed results it is worth considering that investors in the cryptocurrency market are prone to suffer from biases. Investors often overreact, exhibit signs of the disposition effect, herding behaviour and are overconfident. They also overestimate their ability to predict future market changes. - 31 - 5.3 Market sentiment, Privacy, Crime and Volatility An estimated $72 billion worth of illegal activity is performed with the aid of cryptocurrency (Foley, Karlsen, & Putniņš, 2018), for example, in drug trade (Rudesill, Caverlee, & Sui, 2015), money laundering and illegal trades (Jones, 2018) and financing terrorism (Berton, 2015). The demand for secure and anonymous transactions may have been aided by crime. Similar to fiat currencies, the use of cryptocurrency is not limited to criminal activity. It can be used in legitimate as well as illegitimate transactions. According to Foley et al. (2018, p. 14), more than 6 million Bitcoin user holders, or about 5.86% of all users, are estimated to be using their holdings for illegal activities. These users are responsible for 32% of all network transactions14 as can be found in Table 11. User count Transactions Value Addresses Volume All users 106,244,432 605.69 2,964.66 221.71 1,862.51 Illegal users 6,223,337 196.11 1,342.43 58.38 241.46 (5.86%) (32.38%) (45.28%) (26.33%) (12.96%) 100,021,095 409.58 1,622.23 163.33 1,621.05 (94.14%) (67.62%) (54.72%) (73.67%) (87.04%) Other users Table 11 Estimated illegal users count Note. Volume is in billion. Transaction count, holding value and addresses are in million. The push for more regulation on Bitcoin from policymakers forced other cryptocurrencies to implement additional security measures to pre-empt possible government overreach. Monero and Zcash for example encrypts and anonymizes every transaction on their networks. Transactions are almost untraceable15. The addresses of the sender and receiver, the amount and the record of the transaction are encrypted and securely recorded on the blockchain, and is only accessible by the authorized or involved parties (Foley et al., 2018, p. 8; Miller, Moeser, Lee, & Narayanan, 2017). According to Berton (2015, pp. 1–2) cryptocurrencies were allegedly used by Isis to fund terrorist activities, purchase hacked Paypal accounts, acquire stolen credit cards and other forms of electronic payments. Darknet markets such as Silkroad, Agora, Evolution and AlphaBay offer illegal services such as selling drugs and weapons. The transactions are made anonymously, and payments are made with the cryptocurrency of choice, Bitcoin, Monero, 14 Most cryptocurrency trade is performed off-chain and is never recorded on the blockchain. This means that the observed volume of trade found on the blockchain only encompasses the aggregate trades performed only when tokens are deposited or withdrawn from exchanges or moved directly between addresses. 15 Early Monero transactions are traceable due to limitations to the mixing policy (Miller et al., 2017). - 32 - Ethereum or Litecoin. The purchased weapons or drugs are then shipped by regular mail services. Fake residency permits, passports and identity cards are also available for purchase. The number of illegal markets and trade volumes has increased significantly since the closure of the original Silkroad by the FBI in 2013 (Bhaskar, Linacre, & Machin, 2017). Safe, secure and anonymous access to illegal drug markets enabled $1.2 billion in sales by July 2013 to 150,000 buyers (Rudesill et al., 2015). That number has increased significantly as similar drug markets can easily be found on the dark web. While drug trade is inherently risky, there may be evidence that the safety and security of decentralization may have reduced violent crime. An estimated 4.5% - 9% of all Bitcoins moved through the original Silkroad (Bohr & Bashir, 2014, p. 95). The use of cryptocurrencies in the darknet has created a new phenomenon in the form of tainted tokens. As tracing the source of every coin is easily done by looking it up in the blockchain, some exchanges refuse to accept tokens that were used in illegal activities on the dark web. New services offer the mixing of tainted tokens. Mixing services are akin to money laundering. In exchange for a fee, tainted tokens are traded in specialized exchanges or by payment services in exchange for clean or freshly mined tokens. When asked which sector of blockchain applications they are most bullish on, respondents to the Coindesk’s (2018) survey chose decentralized exchanges (28.1%), privacyfocused transactions (22.6%) and asset management tools (20.6%). This highlights the increased demand for more privacy and security following increased hacking attacks and attempts as well as increased government regulations and scrutiny. According to Ali (2017), not only is the use of cryptocurrency thought to aid criminals, there has also been a wave of cryptocurrency related crimes. Extortion, theft and kidnapping crimes occurred in order to coerce the transfer of large amounts of cryptocurrency funds. These funds are mostly untraceable and unreturnable even once the criminals have been arrested. Wallets are encrypted to protect their owners, short of their wilful consent, access to stolen funds may be near impossible. Similarly, loss of access to funds is a common problem either due to forgetting the private key, passport or possible hardware failure, for example, $150 million worth of Ethereum tokens remain inaccessible in what is called the Parity Multi-Sig frozen funds (Blockgeeks). Another problem stems from a new form of cybercrime called Ransomware, which is a computer virus that delivers a code which encrypts the contents of the computer. To receive a decryption key to unlock the contents, the victim is asked to pay a ransom. Payments are usually demanded in Bitcoin (Gonzalez & Hayajneh, 2017). This has increased both awareness and contempt of Bitcoin’s anonymity and has left authorities unable to track the criminals. This - 33 - increase in cybercrime had the indirect effect of increasing corporate demand for Bitcoin. This is done to defend against possible ransomware attacks (Chan, Chu, Nadarajah, & Osterrieder, 2017, p. 1). Hacking is the main threat to crypto markets as well as traditional institutions, traditional banks have fallen victim to such hack attacks, for example, JP Morgan Chase was hacked in 2014, in addition to the stolen funds, the private information about 83 million account holders were compromised (Reuters). Decentralisation of cryptocurrency means that the number of potential targets has increased significantly: exchanges, wallets, ICOs, mining pools, mining companies as well as hosting companies. For example, Nicehash, a cloud mining rental company, was hacked and over 4,700 Bitcoins were stolen. At the time, this was valued at $60 million. 120,000 Bitcoins worth $72 million in 2016 were stolen from Bitfinex. The DAO hack’s $150 million worth of stolen tokens is another example. Reports of hacks are found to have an effect on market prices (Sklaroff, 2017, p. 300). Group-IB, a Moscow based cybersecurity firm, reported that successful ICO projects are attacked an average of 100 times per month. The report also showed that many companies conducting ICOs lacked proper security measures to be safe from cyber-attacks (Group-IB). There are many factors that contribute to the price volatility of the cryptocurrency, internal factors, such as block rewards and mining difficulty constitute of a baseline for the prices, and external factors, such as hacks, security breaches, government regulations and restrictive laws. Bitcoin’s price crashed in 2014 following Bank of China’s banning financial institutions from using Bitcoin (Fry & Cheah, 2016, p. 350). Similarly, the hack of MtGox exchange in 2013 resulted in $473 million being stolen which led to its bankruptcy and the following seizure of its assets (Donier & Bouchaud, 2015), the closure of the original Silkroad in 2013 and the Ethereum network’s DAO hacking in 2016 (Hileman & Rauchs, 2017), were all followed by price crashes in the crypto market. In addition to Sovbetov’s (2018) summary of the price factors, additional factors should also be considered, such as the price of Bitcoin and other top cryptocurrencies (Gkillas, Bekiros, & Siriopoulos, 2018). The list of possible cryptocurrency price determinants can be found in Table 12. Valuing ICO tokens is another challenge. Without the underlying cost of mining, tokens’ assigned values are harder to evaluate. Conley (2017) attempts to estimate the value of tokens by using several different models, the quantity theory of money, present value, metagame value that is value at maximum potential, the efficient market theory and finally a behavioural model that should incorporate all irrational biased behaviour. As participating in an ICO helps in the funding of start-up, he concludes that tokens should be worth the present value after deducting the cost of the services provided by the company. Conley (2017), also - 34 - notes that most tokens continue to be held by investors and speculators and are not used as intended to purchase products or services. The high volume of trade in ICO tokens can inflate their prices. This may be an unwanted side effect of speculation. In 2018, the Lisk token, which uses a proof-of-stake consensus protocol, was structured in a way that enabled delegates to be selected by staking tokens as votes. After the price reached $35, the company had to address the increase in token price by restructuring their voting mechanism and effectively devaluing their token. Internal factors External factors Supply and demand Market factors Traditional Macroeconomic+ Mining rewards Attractiveness+++ markets Regulations Mining costs Popularity+++ Stock markets Restrictions Transaction cost 24hr Trade volume+++ SP500+ Adoption Trade volume+++ Market volatility+++ Exchange rates Market capitalisation Sentiment and speculation Changes to the network Correlation with top New forks, ICOs Crime and privacy Table 12 coins+++ Gold Interest rates Proposed determinates of cryptocurrency prices Note. + indicates significant evidence for Bitcoin, +++ indicates significance for all cryptocurrencies (Gkillas et al., 2018, p. 16; Sovbetov, 2018, p. 7). Kuo Chuen et al. (2017, pp. 26–27) consider the use of media and news sentiments and overnight returns as a proxy to measure investor’s sentiment to be inadequate. News sentiment describes the effect of using certain keywords in news reports. The use of keywords such as “earnings” or “profits” associated with the performance of a stock or company can have an effect on actual earnings. Available word lists designed to measure sentiment in the stock market cannot be directly used in the crypto market. The authors argue that the accuracy of the analysis remains inadequate, despite including cryptocurrency and blockchain related keywords. Overnight returns describe the tendency of some investors to place buy or sell orders after work hours or during the night. These orders would be filled at the start of the following work day. This can lead to a significant increase in the trade volume from the large number of opening orders. However, cryptocurrency exchanges do not have neither closing nor opening times. Exchanges operate constantly regardless of the time or date, therefore, overnight returns and Monday effects are not observed. However, Bitcoin’s day of the week effects are observed in another study by Caporale and Plastun (2017). Past average return was used as a proxy instead, the initial expectation is that investors would buy or sell cryptocurrencies with high or low past returns. This can then be used as an - 35 - indicator of optimism or pessimism about the cryptocurrency. Their mathematical model, quoted here, uses the average of past returns to eliminate the effect of individual daily returns. 𝑆𝑒𝑛𝑡𝑖𝑚𝑒𝑛𝑡C,D IJK ΣFGH 𝑃𝑎𝑠𝑡 𝑟𝑒𝑡𝑢𝑟𝑛C,DJF = 𝑁 Equation 3 Average past returns j is cryptocurrency, n is the number of days, N is the “formation-period” for sentiment build up and was set in their analysis at 10 days. They find that after days of high or low sentiment events, cryptocurrency returns drop or spike before reversing the return on the following day. They also observe that after the next trading day, returns do not seem to further change, which indicates a lack of price correction and leaves the cryptocurrency mispriced. This section highlighted market sentiment, privacy and crime as determinates of cryptocurrency price. Additional attention should be paid to the very recent findings from Griffin and Shams (2018) regarding the effect of market manipulation. 6. Discussion In this section, previously detailed findings and their implications on investors are discussed, in addition, arguments from both opponents and proponents of investing in the crypto market are presented. Finally, limitations to the findings in the literature are discussed. Individual biases have long been used to explain investors’ irrational behaviour in the stock market (Pompian, 2012). Applying the same approach to the crypto market should explain some of the differences in behaviour. This can help investors, researchers, policymakers and developers of the technology make more informed decisions. The crypto market is relatively young. Therefore, it is expected that investors in the crypto market are on average inexperienced and inexperienced individual investors are also more prone to suffer from behavioural biases (Chen et al., 2007, p. 441). Inexperienced and unaware investors are more likely to rely on feelings and emotions rather than rational planning and facts when they make investment decisions. Their limited capacity for processing complex market information combined with overestimating their skills may increase their overconfidence. This overconfidence may also be enforced with long periods of market uptrends when the prices continue to increase consistently. Moreover, investors may exhibit skewed and selective perception of market trends depending on their beliefs and level of optimism. Their incapacity to deal with the information overload may lead investors to make the wrong decision at the wrong time. The brain relies on a series of shortcuts and rules of thumb when faced with complex problems. Often generating inaccurate estimates of a solution - 36 - before all relevant information is acquired and processed. This typically leads to inaccurate conclusions or lucky guesses (Nofsinger, 2016, p. 83). Investors in the crypto market can be considered early adopters. They may overestimate the value of their knowledge about the blockchain technology and ignore warnings from institutional or financial experts. Similarly, opponents of the technology may underestimate its potential and entirely miss out on investment opportunities. This might stem from an inherent ignorance of the advantages and limitations of the blockchain technology. An example is JP Morgan’s boss Jamie Dimon calling Bitcoin a fraud in September 2017, then later retracting his statement by saying he is simply “not interested in the subject at all” (Financial Times). There is no doubt that the craze for fintech or financial technology is similar to the Dotcom fad in the early 2000s albeit on a much smaller scale (Menschel, 2002, p. 47). In the cryptocurrency market, there are many cryptocurrencies with “Bitcoin” in their names, such as Bitcoin Cash, Bitcoin Gold, Bitcoin Private, Bitcoin Diamond and Bitcoin Dark. All unrelated to the original Bitcoin (Coinmarketcap). The collective market value of these iterations or imitations is not insignificant. The increased demand on these coins might be an indication of a correlation with the attractiveness of the name. Different iterations of the same cryptocurrency may coexist in the market. Varied changes in the code over the years create what is referred to as Forks. Soft-forks and Hard-forks are created whenever a significant change is applied to the rules governing the network, especially when network nodes split between using the new and the old codes. A prominent example of a hard fork is the split between Ethereum and Ethereum Classic (Eyal, 2017, p. 48). There is an implicit belief in the cryptocurrency community that most tokens will eventually lose all of their value and disappear. Most investors justify their continued investment by the high short-term returns they achieve. Fraudulent cryptocurrencies that exist merely to scam unaware investors are likely to disappear in the long-term whereas serious projects may persist. Kuo Chuen, Guo and Wang (2017, p. 22) find persistent evidence for low comovements between cryptocurrencies and traditional assets, leading them to conclude that cryptocurrencies are a new asset class that offer opportunities for portfolio diversification. The extreme volatility of the crypto market is deterring many experienced institutional investors from entering it due to the high risk associated with investing in it. New investors might not be mentally prepared for the degree of volatility in the crypto market. The market’s vulnerability to nation-state’s policies adds to its volatility, for example, China’s crackdown on ICOs in 2017, had a big negative effect on the market. Bitcoin lost nearly US$2000 in the following hours of the announcement. Similar dips in value are commonplace - 37 - in the crypto market. In June 2018, Bitcoin had dropped nearly 70% from its all-time high of $20,000 to under $6000 (Coinmarketcap). Unlike traditional markets where share prices tend to reflect the performance of a company (Shleifer, 2000), the price of a token is affected by the smallest of changes in public perception about the entire field. The mere mention of a cryptocurrency in the news significantly affect its value, one example is Dogecoin, which doubled its price overnight in 2017 after news reports. There was a massive increase in new cryptocurrency investors in 2017. Coinbase and GDAX, the largest Bitcoin exchanges in the United States, reported a trader base of 13 million users in November 2017, this number is expected to have doubled by January 2018. Binance, the largest cryptocurrency exchange with a staggering $9.5 billion daily trading volume, revealed that it has added more than 250,000 users in a single day in January 2018. In December 2017 most of the large crypto markets stop accepting new investors as they could not keep up with the requests. Teenage millionaires who made their fortunes from trading in cryptocurrency, might not have been allowed to trade on the stock exchanges due to age restrictions in certain countries, such is the case in some U.S. states and countries that enforce a know your customer (KYC) policy that requires users to submit a proof of identity, often not available to minors. The loss of control over the flow of capital is something that worries some central banks. This led some countries to completely forbid Bitcoin’s use. According to Coindesk (2018), Egypt’s clerical body issued a state-sponsored declaration forbidding the use of cryptocurrency, calling it harmful. Indonesia’s central bank issued a statement warning against the use of cryptocurrency and declaring Bitcoin payments illegal. Some countries however, fully embraced the new possibilities of the technology. For example, Japan’s friendly approach to integrating cryptocurrency’s use in daily activities was supported by approving four new exchanges in December 2017. Israel and Ukraine were seeking to ban Bitcoin firms from trading in the stock exchange while they were planning to launch state-sponsored cryptocurrencies. Venezuela issued an oil barrel-backed coin called the Petro. Saudi Arabia and UAE central banks were testing the use of cryptocurrencies (Coindesk, 2018). Cryptocurrency prices are not uniform across all exchanges, some exchanges have higher trading fees and others operate in specific currencies. Bitcoin prices tend to be within a margin of a few hundred dollars. This discrepancy allows theoretical cross exchange arbitrage (Borri & Shakhnov, 2018). An investor can potentially buy and sell cryptocurrencies across multiple exchanges for different prices and make profits from the price differences. The main limitation to this investment strategy is the transaction cost and time delay, in addition to - 38 - network, trade and exchange fees. Other fees for depositing or withdrawing of funds must also be accounted for. A possible scenario for the complete set of costs to a U.S. investor is as follows: US Dollar to other fiat exchange costs, deposit fee in primary exchange platform, trading fee incurred by the purchase of a cryptocurrency. network or transaction fee in addition to attached withdrawal fees to secondary exchange. Deposit fee in secondary exchange. Additional fees from trading the units for fiat or US dollars and finally, withdrawal fee from the secondary exchange. Borri and Shakhnov (2018, p. 17) give an example with Bitcoin16, but the same should apply to other cryptocurrencies across multiple exchanges. Another important issue to discuss is the immense environmental impact of mining cryptocurrency. Ethereum decided in 2017 that it will move away from the Proof-of-Work protocol to another form of transaction verification that requires a fraction of the energy currently needed. The current total energy consumption by Bitcoin and Bitcoin Cash miners alone is an approximate 71.12-terawatts per hour or similar to the consumption of Chile or 0.3% of the world total electricity consumption (Vries, 2018). Price clustering at round numbers is another phenomenon, which happens when prices of assets tend to cluster around whole digits, such as numbers ending in zero or double zeros. According to Urquhart (Urquhart, 2017a, p. 145), this phenomenon also exists in the stock market, the commodity market, betting markets and foreign exchange markets. Attraction to whole numbers maybe driven by ambiguity aversion. The author finds evidence for price clustering at round numbers or numbers ending with the 00 digits in Bitcoin closing prices on the Bitstamp exchange from 2012 to 2017. This occurred 10.81% of the time, additionally, the author found smaller evidence for the 50 and 99 digits It may be useful to draw a comparison between noise traders and some cryptocurrency investors (Ramiah et al., 2015, pp. 94–95). Noise traders are investors that appear to make investment decisions haphazardly. They make decisions without a clear underlying reason or rationale. Moreover, they base their decisions on bad information or on market trends rather than fundamental factors. Findings related to noise trading’s effect on increasing volatility in the stock market could help in studying volatility in the crypto market. Peer excitement about cryptocurrency news extends beyond the crypto market. Spikes in stock prices can be observed following news of cryptocurrency related cooperation. Although in-depth analysis remains an opportunity for future research to study, there are examples of the extreme overreaction to blockchain technology news in the stock market. Kodak announced a 16 Most exchanges do not charge fees on Bitcoin deposits, the largest exchanges charge a withdrawal fee instead (Borri & Shakhnov, 2018, p. 17). Additional deposit and withdrawal fees vary by cryptocurrency. - 39 - planned ICO on the 9th of January 2018, the share price increased from $3.1 on January 8th to $10.7 on January 10th (Reuters). Similar jumps can be found in the share prices of Veltyco, Overstock, Long island ice tea, Chinese social media Renren and Future Fintech Group following similar announcements of blockchain technology cooperation or investment (CNBC). Stock prices spiked directly after news was reported about the potential launch of an ICO or the use of a blockchain related technology or in the case of Long Island changing the company’s name to Long Blockchain. In many cases, stock prices fallen to their previous values in the following days (Cointelegraph). Crypto markets are filled with inefficiencies. Irrational decisions, optimistic sentiments and positive convictions about the future of the technology are commonplace. Beliefs not facts drive most investment decisions. Biases are exasperated by exchanges’ inability to combat or absorb the effect of mob movements. An example of this is the effect pump-and-dump groups have on smaller cryptocurrencies’ and tokens’ price. On a near daily basis, questionable tokens receive significant trade volumes that see their prices boosted to irrational highs before subsequently crashing. Exchanges often display recent major price movements on prominent positions on their front page, sometimes displaying traded charts that include most pumped and most dumped tokens. Unfortunate investors that buy artificially inflated tokens either out of hope, greed or fear are left with overpriced or worthless tokens. There have been cases of possible fraud, Ponzi schemes, insider trading and market manipulation (Griffin & Shams, 2018; Kotas, 2018, p. 27). The latest of which is related to the cryptocurrency Tether, which is used by many exchanges as a “peg” for the US Dollar. Tether is supposed to be backed with an equivalent amount of US Dollars on a one-for-one basis. Tether has a total market capitalization of over $2,6 billion. Griffin and Shams (2018) argue that Bitcoin price manipulation took place and may have resulted in 2017’s price increase to $20,000. The cryptocurrency exchange backing Tether, Bitfinex has been under ongoing investigation by U.S. regulators for allegations of issuing tokens without backing. The authors propose that Tether accumulated top cryptocurrency tokens while their prices were down from one exchange effectively inflating the prices of top cryptocurrencies, this was followed by selling these cryptocurrencies on other exchanges for US Dollars. The authors studied the effect of issuing new Tether tokens on Bitcoin and other top cryptocurrency prices. They found that on 87 occasions or 1% of their sampled time series, the printing of Tether tokens was correlated with 50% of Bitcoin’s compounded return and 64% of the returns of other large cryptocurrencies (Griffin & Shams, 2018). - 40 - Additional arguments Despite the findings of significant evidence for overreaction to Bitcoin price changes, Caporale and Plastun (2018) conclude that their evidence cannot be seen as evidence against the efficient market hypothesis (EMH). They base their conclusion on the inability of their primary investment strategies to maintain stable profitability. A secondary strategy that generated stable profits was found to be not different from random trades. One possible limitation to their method for measuring overreaction is its lack of distinction between positive and negative price movements. The authors test the direction of the overreaction but do not assign more weight to positive movements. The reason this may be relevant is derived from Angelovska’s (2016, p. 9) study of price overreaction in the Macedonian stock exchange. An interesting finding in the paper was that positive price shocks occur 25% more often than negative shocks. Courtois (2016, p. 11) evaluates whether Bitcoin’s price reflects all publicly available information and finds a one-sided bias towards Bitcoin’s increased popularity17. The study does not observe any correction effect after negative news. Including reports of thefts and hacks. The author concludes that bias in the investors’ evaluation of negative news is prevalent. The main limitation to using these proxies for popularity may be that investors do not rely on these sources for information. Blau (2018, p. 21) finds that speculative trading in Bitcoin cannot be attributed as the cause of price volatility. The proposed explanation for the causes of volatility provided in the paper is “The fact that Bitcoin cannot be borrowed and, therefore, cannot be shorted is part of the initial intention of Bitcoin since digital signatures are required as proof of ownership”. This argument is no longer valid or accurate, as a lending market for Bitcoin already exists, cryptocurrency exchanges provide lending services and enable short-sale possibilities, for example, Bitfinex and Poloniex offer lending services with varying interest rates. In addition, the first Bitcoin futures contracts went into effect in December 2017 (Coindesk, 2018). Fama (1970) details three types of tests for market efficiency, weak form for testing historical prices reflection of market information, semi-strong, which tests whether prices efficiently change according to publicly available information, and strong form tests for the existence of monopolistic access to relevant information. Urquhart (2017b) tests the weak form efficiency of Bitcoin. By running six different tests and splitting the time-series samples into two subsamples, there is strong evidence of inefficiency in Bitcoin pricing. The results are 17 Popularity here was measured by changes in Wikipedia.com’s Bitcoin traffic, event dummies, network and exchange volume changes. - 41 - adopted in Table 13. The author also concludes that, as cryptocurrencies mature there may become less volatile and more efficient Test Ljung-Box Runs Bartels AVR BDS R/S Hurst 2010-2016 (0.00) (0.00) (0.00) (0.01) (0.00) 0.353 2010-2013 (0.00) (0.00) (0.00) (0.00) (0.00) 0.363 2013-2016 (0.35) (0.00) (0.00) (0.64) (0.00) 0.406 Table 13 Weak form efficiency in Bitcoin results. Note. Ljung-Box test for autocorrelation of returns, null hypothesis is no autocorrelation. The runs and Bartels tests for the independence of returns, independence is the null hypothesis. Automatic variance test (AVR) using wild-bootstrapped AVR to test if the price process follows a random walk. BDS test for serial dependence in stock returns. Rescaled Hurst exponent for long memory of stock return test for persistence, strong persistence for values greater than 0.65 and strong anti-persistence for values less than 0.45 (Urquhart, 2017b, p. 81). Cheah and Fry (2015, p. 33) argue that due to the speculative nature of Bitcoin, market sentiment is the sole determinate of Bitcoin’s value. And that price volatility is undermining Bitcoin’s ability to function as a stable medium of value exchange. The authors conclude that the cryptocurrency is in a speculative bubble irrespective of the tested time intervals. Therefore, they claim that the fundamental value of Bitcoin is zero. A counter-argument here is that the cost of mining Bitcoin can still serve as a tangible baseline for its value, albeit in the form of computational labour (Hayes, 2017, p. 1310). Hayes (2017, p. 1310) also details some of the arguments against Bitcoin’s use as a currency, its intrinsic value and its lack of “moneyness” as it does not have a tangible tied asset. Therefore, it cannot be compared to gold. The author debates whether Bitcoin has no fundamental value except pure market valuation against other currencies18. Arguing that the limited supply of Bitcoin can also be used as a factor to explain the increased demand on the limited supply. Dormant accounts constitute a risk to the cryptocurrency market, as sudden increases in the supply of circulated coins would increase inflation and lowering the cryptocurrency’s value as can be interpreted from the author’s analysis of the impact of new tokens on market price. Another argument is that the underlying factor for Bitcoin’s value is the technological breakthrough attached to the blockchain’s ability to solve the “doublespending” problem. Double-spending happens if a hostile agent in a financial system is able to replicate their funds, thus enabling them to spend the same amount more than once. Another point discussed is the value of Bitcoin’s popularity and appeal. 18 Although due to its money-like asset-like properties, it might entail some value albeit without a tangible basis. - 42 - Log(price) log(hash power) (1) (2) 0.67** 0.685*** (0.064390) log(new blocks) -0.98*** -0.983*** (0.061908) Scrypt algorithm 7.43*** 7.461*** (0.875) Max supply ratio -0.57 Longevity 0.00067 Constant -9.68*** -9.527*** (0.780) N 66 66 R2 0.844 0.843 Adjusted R2 0.830 0.835 . -121.659 63.71 111.038 . 0.000 log (likelihood) F-test Prob(F-statistic) Table 14 Determinates of cryptocurrency value Note. Adopted from Hayes (2017, p. 1313,1320). Hayes (2017) proceeds to test the relation between cryptocurrency values and the computational power of the network, block rate, the ratio between circulating and max supply, other mining algorithm used in several other cryptocurrencies19 and finally the longevity of the cryptocurrency. He finds positive evidence for increased hashing power and the use of other algorithms and a negative correlation with the issuance of new tokens. His results are detailed in Table 14. Cheah and Fry (2015, p. 34), discuss the dormancy of an estimated 70% of all Bitcoins, these are tokens held in addresses that were never used after receiving funds. Ron and Shamir (2013) estimate that between 55% and 73% of all tokens in the system are either dormant or completely out of circulation, access to these tokens may be lost forever. The large difference in their estimate is due to the difficulty of estimating whether the funds are held in saving accounts or permanently locked in inaccessible wallets. In 2013, there were only 9,000,050 Bitcoin units in circulation. Despite the novelty and innovation in the blockchain technology, it is important to note that most of the proposed solutions can already be found in abundance from centralized 19 There have been many changes in the development of specialized ASIC hardware capable of mining the Scrypt algorithm. The main premise remains valid with increased mining costs and difficulty. - 43 - solutions, sometimes with higher quality and reliability. The perception of cryptocurrencies as the solution to all the problems in the financial system is erroneous, cryptocurrencies provide new and alternative solutions to existing problems. Methodical limitations in the reviewed literature There are some limitations to findings of the reviewed studies. Selected data sources included no randomized field or lab experiments. There were no control groups to serve as the baseline to compare the observed results with. The demographics of investors are changing over time as adoption increases. Therefore, the recorded findings cannot be generalized without a more balanced sample. Hileman and Rauchs’s (2017) study details findings from expert respondents that are affiliated with the cryptocurrency field. By profession, these respondents are more knowledgeable and experienced than the average cryptocurrency user and investor. Coindesk’s (2017; 2018) much larger sample of roughly 400 responses in 2017 and 3000 in 2018 is unbalanced. The Coindesk’s blockchain sentiment survey sample is made up of, 84% from North America and Europe. The survey respondents were also overwhelmingly Males 95% in 2017 and 97% in 2018 between the age 26-35 (32% in 2017 and 39% in 2018) and 36-45 (29%). The survey was conducted through their website and only the results were published in the reports. Verweij’s (2018) sample was also collected anonymously through social media websites. Respondents who answered the survey had a chance to earn 0.1ETH, or about $50. The survey was conducted on Google Forms. However, the possibility of repeated attempts by fraudulent respondents cannot be completely ruled out. The reliance on Google Analytica to observe trends in popularity by Sovbetov (2018) to acquire accurate indicators for demographics is acceptable for preliminary analysis but remains inadequate for generalizability. Despite being the largest search engine on the internet, the private nature of cryptocurrency may indicate that additional sources of data may exist elsewhere, for example, on the hidden part of the internet known as the deep web, which is the part of the internet that contains all websites not indexed by mainstream search engines and websites that can only be reached through encrypted connections or specialised software. The deep web houses nearly 96% of the content of the world wide web. Within the deep web lies the much smaller network called the Darknet, where most illegal activity related to the deep web take place (Foley et al., 2018, p. 6; Rudesill et al., 2015, p. 6). The private nature of cryptocurrency makes it also difficult to assure the randomness of any of the reviewed respondent samples. The overwhelming majority of respondents were - 44 - either Europeans or North American. In some cases, all respondents were from English speaking online communities (Bohr & Bashir, 2014). Although online surveys can help in gathering data with relatively low costs, there are many drawbacks to this methodology. Sample validity may be an issue in the reviewed papers. Reliance on online forums and discussion platforms, such as Reddit cannot guarantee the randomness of the survey respondents, despite the relatively large sample sizes, a coverage error could have occurred. The risk of a self-selected opinion poll is especially large in surveys conducted directly on websites, as the motivations of respondents cannot be controlled for. Respondents could be motivated solely by monetary incentives, they could have found survey by chance and may not have prior knowledge on the subject, are personally interested or affected by the results or were notified about the survey from other similar respondents such as friends and family members that share the same interests and opinions. Online surveys exclude nonparticipants and include more similar respondents thus introducing a double bias to the sample (Duda & Nobile, 2010, p. 57). Non-response sampling bias describes samples where some important members of the observed community refrain from responding to the survey. As the characteristics of non-respondents cannot be accounted for, results obtained from surveys cannot be generalized with much confidence. Another important issue with online surveys is the stakeholder bias and the existence of unverified respondents. Stakeholders have a personal interest in the results and could actively seek to influence the results by completing the survey multiple times or by advertising it to their friends that share their opinions. Maintaining strict access to verified unique individuals to the survey is difficult. Providing a monetary incentive or a participation reward could encourage hostile action from some respondents seeking to win the prize (Duda & Nobile, 2010, p. 58). 7. Policy implications The initial reaction to Bitcoin in 2009 was either a warm reception or scathing denouncement as fraud. Whether cryptocurrency prices are in a bubble or not, warnings about investing in the crypto market have been the response from institutional investors. Others criticised its use for criminal activities or as a waste of resources (Chiu & Koepp, 2017, p. 2). The increased popularity of cryptocurrencies and the recurrence of crimes involving them have driven most countries to regulate rules governing cryptocurrency different uses. The lack of control in the market leaves pump and dump groups free to deform the real values of the tokens. Fighting money laundering and funding for criminals and terrorists is a top priority for regulators and policymakers alike. - 45 - China and South Korea’s ban on ICOs stems from the possibility that investors can be easily defrauded. Fraudulent ICOs also share similarities with Ponzi and Pyramid schemes. China has declared plans to ban all cryptocurrency exchanges, although this remains to take effect. Japan is the biggest market for cryptocurrency and has thus far taken a positive position towards cryptocurrency adoption, Bitcoin is considered legal tender and exchanges are allowed to operate with a license in the country (Coindesk, 2018). ICOs as an equity-crowdfunding platform has remained relatively unregulated. In 2017, the US Securities and Exchange Commission (SEC) ruled that DAO tokens are security liable for taxation (Hacker & Thomale, 2017). The Internal Revenue Service (IRS) defines all virtual currencies, including cryptocurrencies, as property and not as a currency. This definition extends to mined or purchased cryptocurrencies. Increased regulations led token issuers to structure their sale in a way that would prevent US citizens from obtaining their tokens thereby evading US regulations entirely. EU regulations regarding ICOs, are that ICOs are legal as long as they adhere to antimoney laundering (AML) and know your customer (KYC) policies. The European Securities and Markets Authorities (ESMA), the European Banking Authority (EBA) and the European Insurance and Occupational Pensions Authority (EIOPA) issued a warning against buying or holding “virtual currencies” (Europa.eu). The same warning was repeated by several executives at the European Central Bank (ECB), such as Yves Mersch and Mario Draghi. The ECB has taken a negative stance towards cryptocurrency since 2012. Fig. 3 Cryptocurrency regulation around the world - 46 - Note. Classifications are legal tender, currency, asset or property and commodity (Coin.dance). According to Coin.dance (accessed on 07.06.2018), 106 countries allow the unrestricted use of Bitcoin, 10 countries have some restrictions and 10 consider it illegal. Classification of Bitcoin varies across countries as mentioned before. A sample of classifications can be found in Figure 3 and in more detail in Appendix Table A.10. Regulators around the world have classified cryptocurrencies as a currency, security or commodity based on their inherent functionalities. However, the American SEC has classified ICO tokens issued for the purposed of raising funds as securities and not as digital currencies (Mokhtarian & Lindgren, 2017, pp. 9–13). In some countries, there appears to be an overreaction from policymakers and regulators to cryptocurrency, as is the case in Saudi Arabia, Malaysia, Morocco and Bangladesh for example. The response in cases of policymakers is to design a one-size-fits-all policy to regulate or contain these new innovations. Regulators should instead determine the appropriate classification for each cryptocurrency or token on a case-by-case basis. Effort should be made to raise investors awareness of fraudulent investments and scams, while simultaneously avoiding the restriction of innovators’ abilities to develop alternative technology solutions to perceived problems either in the financial market or elsewhere. Blanket banning of cryptocurrency advertising, by Google, Facebook and Twitter, protects investors from scams but also prevents them from accessing useful information about promising investment opportunities (Coindesk). Hacker and Thomale (2017, pp. 25–40) provide an extremely detailed guide on the regulations imposed on cryptocurrency and ICO tokens. The study also considers tokens as an investment, utility or currency component. It then separately studies hybrid tokens that share two or more components. While they conclude that pure currency and utility tokens are exempted from the EU security regulations, they find that most tokens share security components, such as fungibility. This confusion regarding token classification has left developers worried about the prospects of crowdfunding via ICOs, which entails a large burden and cost to adhere to all legal requirements. 8. Conclusion The development of a decentralized distributed solutions for alternative finance has allowed cryptocurrencies to flourish. The use of blockchain technology has expanded in the development of many new technology solutions. Although many of the innovations offer alternative solutions to existing problems, centralized solutions may already exist and also be more efficient and cost-effective. Smart-contracts aim to limit the influence of mediators on the parties involved but they cannot completely eliminate the need for mediation. Decentralized - 47 - exchanges, market, and betting platforms offer a sort of security and anonymity that is not found in their traditional counterparts, but the risk of security failures remains pertinent. Cryptocurrencies offer a new method of transaction between individuals. The use of cryptocurrencies is usually private, secure, relatively quick and not expensive. Some currencies offer complete anonymity to protect the identities of their users. This anonymity also allows safe and untraceable participation in illegitimate trade or procurement of illegal drugs or weapons. New threats from criminals and cybercriminals have become an issue that is alarming governments and security agencies. The increased use of cryptocurrencies by criminals is likely to receive a response in kind from law enforcement. Cryptocurrencies use is not restricted to illegal activities. They offer a wide range of investment opportunities. This is supported by the rapid increase in the total market capitalization and the increased interest from individual and institutional investors. Institutional investment entities are slowly embracing the technology. Investors in the crypto market may be as prone to behavioural biases as individual investors in the stock market. Investors appear to suffer from overreaction, overconfidence, herding, representativeness biases and exhibit signs of suffering of the disposition effect, regret and ambiguity aversion. Additionally, prices in the crypto market show signs of inefficiency, such as persistent mispricing, extreme volatility and recurring price bubbles. Irrational decisions may also be influenced by an overreaction to news regarding crime, privacy breaches, changes in market sentiment. On average, the prices of mineable cryptocurrencies are considered to be largely driven by mining costs. The characteristics of the exchanges, their design and how they function, may have an additional effect on increasing user’s susceptibility to certain biases. The design can act as a nudging mechanism towards committing certain purchasing or selling decisions, which is not necessarily in the interest of the investor. Exchanges are programmed to function automatically, 24-hours a day, and remain open on holidays and weekends. They typically have no geographical or language barriers, although not restricted, the investor base of Japanese exchanges may be from Japan due to the interface’s language. Some exchanges do not force age limitations or user verification. On most exchanges, trades are performed completely anonymously. Decentralized exchanges do not have a central management to either maintain or enforce specific security or policy issues. Pump-and-dump groups remain a major disruption as their immense effect, especially on smaller cryptocurrencies, is increasing volatility and threating unaware and inexperienced investors. Market manipulation is a major risk and needs to be properly addressed by regulators and policymakers. - 48 - This paper also highlights the opportunities and risks facing cryptocurrency investors. Increasing awareness of potential investment pitfalls, caused by behavioural biases, can help investors make more rational decisions. Future research would benefit from obtaining data on investors decisions from a balanced randomized pool of respondents using indirect instruments other than surveys. The reviewed studies relied on the direct observation of survey respondents, which has the drawback of capturing specific points in time. Lab or field experiments may observe a more balanced representation of investors’ behaviour. Another suggestion would be to compose a nonprobability sample, for example, using snowball sampling (Sue & Ritter, 2011, p. 45). This can be done by identifying some respondents that exhibit clearly distinct characteristics and ask them to refer other potential respondents, the sample can be built with a randomization of the collected responses or tested in clusters. Cultural background and individual characteristics may cause differences in biases (Chen et al., 2007, p. 429). Research on cultural differences is another opportunity for future research as the literature on this part remains scant. References Ali, A. (2017). Ransomware: a Research and a Personal Case Study of Dealing With This Nasty Malware. Issues in Informing Science & Information Technology, 14, 87–99. https://doi.org/10.1080/13880290490480167 Alvarez-Ramirez, J., Rodriguez, E., & Ibarra-Valdez, C. (2018). Long-range correlations and asymmetry in the Bitcoin market. Physica A: Statistical Mechanics and Its Applications, 492, 948–955. https://doi.org/10.1016/j.physa.2017.11.025 Andrikopoulos, P. (2005). Modern Finance vs. Behavioural Finance: An Overview of Key Concepts and Major Arguments. SSRN Electronic Journal, 44(0), 0–21. https://doi.org/10.2139/ssrn.746204 Angelovska, J. (2016). Large Share Price Movements , Reasons and, (1936), 1–18. Back, A., Corallo, M., & Dashjr, L. (2014). Enabling blockchain innovations with pegged sidechains. URL: Http://Www., 1–25. Retrieved from http://newspaper23.com/ripped/2014/11/http-_____-___-_www___-blockstream___com__-_sidechains.pdf Bano, S., Al-Bassam, M., & Danzis, G. (2017). The Road to Scalable Blockchain Designs. ;;Login:, 42(4), 31–36. Retrieved from http://sheharbano.com/assets/publications/usenix_login_2017.pdf - 49 - Barber, B. M., & Odean, T. (2001). BOYS WILL BE BOYS: GENDER, OVERCONFIDENCE, AND COMMON STOCK INVESTMENT [*].(Statistical Data Included). Quarterly Journal of Economics, 116(1), 261. https://doi.org/10.2139/ssrn.139415 Barberis, N., & Thaler, R. H. (2002). A Survey of Behavioral Finance. SSRN Electronic Journal. https://doi.org/10.2139/ssrn.327880 Bartos, J. (2015). Does Bitcoin follow the hypothesis of efficient market? International Journal of Economic Sciences, IV(2), 10–23. https://doi.org/10.20472/ES.2015.4.2.002 Berton, B. (2015). The dark side of the web : ISIL ’ s one-stop shop ?, (June), 1–2. https://doi.org/10.2815/454889 Bhaskar, V., Linacre, R., & Machin, S. (2017). Dark web: The economics of online drugs markets. LSE Business Review Blog, 1–5. Blau, B. M. (2018). Price dynamics and speculative trading in Bitcoin. Research in International Business and Finance, 43(December 2015), 15–21. https://doi.org/10.1016/j.ribaf.2017.07.183 Bohr, J., & Bashir, M. (2014). Who Uses Bitcoin? An exploration of the Bitcoin community. 2014 12th Annual Conference on Privacy, Security and Trust, PST 2014, 94–101. https://doi.org/10.1109/PST.2014.6890928 Borri, N., & Shakhnov, K. (2018). Cryptomarket Discounts. Bouri, E., Gupta, R., Tiwari, A. K., & Roubaud, D. (2017). Does Bitcoin hedge global uncertainty? Evidence from wavelet-based quantile-in-quantile regressions. Finance Research Letters, 23, 87–95. https://doi.org/10.1016/j.frl.2017.02.009 Brooks, M. (2008). Behavioral finance: Theories and evidence. The Research Foundation of the CFA Institute, (1979), 1–26. https://doi.org/10.2470/rflr.v3.n1.1 Bukovina, J., & Martiček, M. (2016). Sentiment and Bitcoin Volatility. MENDELU Working Papers in Business and Economics. Caporale, G. M., & Plastun, A. (2017). the Day of the Week Effect in the Crypto Currency Market, (October). Retrieved from http://www.brunel.ac.uk/economics Caporale, G. M., & Plastun, A. (2018). Price Overreactions in the Cryptocurrency Market. Retrieved from https://ssrn.com/abstract=3113177 Chan, S., Chu, J., Nadarajah, S., & Osterrieder, J. (2017). A Statistical Analysis of Cryptocurrencies. Journal of Risk and Financial Management, 10(2), 12. https://doi.org/10.3390/jrfm10020012 Cheah, E. T., & Fry, J. (2015). Speculative bubbles in Bitcoin markets? An empirical investigation into the fundamental value of Bitcoin. Economics Letters, 130, 32–36. - 50 - https://doi.org/10.1016/j.econlet.2015.02.029 Chen, G., Kim, K. A., Nofsinger, J. R., & Rui, O. M. (2007). Trading performance, disposition effect, overconfidence, representativeness bias, and experience of emerging market investors. Journal of Behavioral Decision Making, 20(4), 425–451. https://doi.org/10.1002/bdm.561 Chiu, J., & Koepp, T. V. (2017). The Economics of crypticurrencies - Bitcoin and Beyond. Ciaian, P., Rajcaniova, M., & Kancs, d’Artis. (2018). Virtual relationships: Short- and longrun evidence from BitCoin and altcoin markets. Journal of International Financial Markets, Institutions and Money, 52, 173–195. https://doi.org/10.1016/j.intfin.2017.11.001 Coindesk. (2017). State of blockchain Q4 / 2017, 6. Retrieved from https://www.coindesk.com/state-blockchain-2018-slideshow/ Coindesk. (2018). State of the blockchain Q1/2018. Coindesk. Retrieved from https://www.coindesk.com/research/state-blockchain-2018/ Conley, J. P. (2017). Blockchain and the Economics of Crypto-tokens and Initial Coin Offerings. Vanderbilt University Department of Economics Working Paper Series, (June). Retrieved from http://www.accessecon.com/Pubs/VUECON/VUECON-1700008.pdf Corbet, S., Larkin, C., Lucey, B., Meegan, A., & Yarovaya, L. (2017). Cryptocurrency reaction to FOMC Announcements: Evidence of Heterogeneity Based on Blockchain Stack Position. SSRN Elecronic Journal, 1–39. Retrieved from https://ssrn.com/abstract=3073727 Courtois, N. T. (2016). Features or bugs: The seven sins of current bitcoin. New Economic Windows, 8437(4), 97–120. https://doi.org/10.1007/978-3-319-42448-4_6 Creyts, J., & Trbovich, A. (2018). Can blockchain help us to address the world’s energy issues? DeBondt, W. F. M., & Thaler, R. H. (1985). Does the Stock Market Overreact? Journal of Finance, 40(3), 793–805. https://doi.org/10.1111/j.1540-6261.1985.tb05004.x Donier, J., & Bouchaud, J. P. (2015). Why do markets crash? Bitcoin data offers unprecedented insights. PLoS ONE, 10(10), 23–25. https://doi.org/10.1371/journal.pone.0139356 Duda, M. D., & Nobile, J. L. (2010). The fallacy of online surveys: No data are better than bad data. Human Dimensions of Wildlife, 15(1), 55–64. https://doi.org/10.1080/10871200903244250 Dwyer, G. P. (2015). The economics of Bitcoin and similar private digital currencies. Journal - 51 - of Financial Stability, 17, 81–91. https://doi.org/10.1016/j.jfs.2014.11.006 Eyal, I. (2017). Blockchain Technology: Transforming Libertarian Cryptocurrency Dreams to Finance and Banking Realities. Computer, 50(9), 38–49. https://doi.org/10.1109/MC.2017.3571042 Fama, E. F. (1970). American Finance Association Efficient Capital Markets : A Review of Theory and Empirical Work Author ( s ): Eugene F . Fama Source : The Journal of Finance , Vol . 25 , No . 2 , Papers and Proceedings of the Twenty- Eighth Annual Meeting of the American, 25(2). Foley, S., Karlsen, J. R., & Putniņš, T. J. (2018). Sex, Drugs, and Bitcoin: How Much Illegal Activity Is Financed Through Cryptocurrencies? Ssrn. https://doi.org/10.2139/ssrn.3102645 Fry, J., & Cheah, E. T. (2016). Negative bubbles and shocks in cryptocurrency markets. International Review of Financial Analysis, 47, 343–352. https://doi.org/10.1016/j.irfa.2016.02.008 Gkillas, K., Bekiros, S., & Siriopoulos, C. (2018). Extreme Correlation in Cryptocurrency Markets, 1–19. Retrieved from https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3180934 Gonzalez, D., & Hayajneh, T. (2017). Detection and prevention of crypto-ransomware. 2017 IEEE 8th Annual Ubiquitous Computing, Electronics and Mobile Communication Conference (UEMCON), 472–478. https://doi.org/10.1109/UEMCON.2017.8249052 Griffin, J. M., & Shams, A. (2018). Is Bitcoin Really Un-Tethered? Retrieved from https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3195066 Hacker, P., & Thomale, C. (2017). Crypto-Securities Regulation: ICOs, Token Sales and Cryptocurrencies under EU Financial Law. SSRN Electronic Journal. https://doi.org/10.2139/ssrn.3075820 Hanl, A., & Michaelis, J. (2017). Kryptowährungen — ein Problem für die Geldpolitik? Wirtschaftsdienst, 97(5), 363–370. https://doi.org/10.1007/s10273-017-2145-y Haugen, R. A. (1999). The inefficient stock market: What pays off and why. Prentice Hall. Hayes, A. S. (2017). Cryptocurrency value formation: An empirical study leading to a cost of production model for valuing bitcoin. Telematics and Informatics, 34(7), 1308–1321. https://doi.org/10.1016/j.tele.2016.05.005 Hileman, G., & Rauchs, M. (2017). 2017 Global Cryptocurrency Benchmarking Study. SSRN Electronic Journal. https://doi.org/10.2139/ssrn.2965436 Hirshleifer, D. (2015). Behavioral Finance. Annual Review of Financial Economics, 7(1), 133–159. https://doi.org/10.1146/annurev-financial-092214-043752 - 52 - Hong, H. (2007). Behavioural finance: Introduction. European Financial Management, 13(3), 389–393. https://doi.org/10.1111/j.1468-036X.2007.00365.x Hurlburt, G. F., & Bojanova, I. (2014). Bitcoin: Benefit or curse? IT Professional, 16(3), 10– 15. https://doi.org/10.1109/MITP.2014.28 Ireland, D. (2018). Behavioral Economics , Regulatory Games and Errors , and Why the Next Global Financial Crisis Might Be Inevitable. Jones, C. (2018). Digital currencies and organised crime update. Financial Regulation International. Retrieved from http://eprints.uwe.ac.uk/35541 Kahneman, D., & Tversky, A. (1972). Subjective probability: A judgment of representativeness. Cognitive Psychology, 3(3), 430–454. Karolyi, G. A., & Stulz, R. M. (2003). Chapter 16 Are financial assets priced locally or globally? Handbook of the Economics of Finance, 1(SUPPL. PART B), 975–1020. https://doi.org/10.1016/S1574-0102(03)01025-2 Kornmesser, S. (2008). Theoretizität im logischen empirismus und im strukturalismus Erläutert am fallbeispiel des neurobiologischen konstruktivismus. Journal for General Philosophy of Science, 39(1), 53–67. https://doi.org/10.1007/s10838-008-9062-0 Kotas, C. (2018). Kryptowährungen als Digital Assets - eine Zwischenbilanz. Zeitschrift Für Das Gesamte Kreditwesen, 71(2), 78–82. Krafft, P. M., Della Penna, N., & Pentland, A. (2018). An Experimental Study of Cryptocurrency Market Dynamics. https://doi.org/10.1145/3173574.3174179 Kuo Chuen, D. L., Guo, L., & Wang, Y. (2017). Cryptocurrency: A New Investment Opportunity? SSRN Electronic Journal, 20(3), 16–40. https://doi.org/10.2139/ssrn.2994097 Leclair, E. M., Voia, P. M., Khalaf, L., Strathearn, T. A. M., & Leclair, E. M. (2018). Herding in the cryptocurrency market Herding in the cryptocurrency market. Macdonell, A. (2014). Popping the Bitcoin Bubble: An application of log---periodic power law modeling to digital currency, (December 2013), 1–33. Retrieved from https://economics.nd.edu/assets/134206/mac_donell_popping_the_biticoin_bubble_an_a pplication_of_log_periodic_power_law_modeling_to_digital_currency.pdf Menschel, R. (2002). Markets, Mobs & Mayhem. How to Profit From the Madness of Crowds. Miller, A., Moeser, M., Lee, K., & Narayanan, A. (2017). An Empirical Analysis of Linkability in the Monero Blockchain. Retrieved from http://arxiv.org/abs/1704.04299 Mokhtarian, E., & Lindgren, A. (2017). Rise of the Crypto Hedge Fund: Operational Issues and Best Practices for an Emergent Investment Industry, (i). - 53 - Nofsinger, J. R. (2016). The psychology of investing. Routledge. Norvill, R., Pontiveros, B. B. F., State, R., Awan, I., & Cullen, A. (2017). Automated labeling of unknown contracts in Ethereum. 2017 26th International Conference on Computer Communications and Networks, ICCCN 2017, 1–6. https://doi.org/10.1109/ICCCN.2017.8038513 Odean, T. (1998). Volume, volatility, price, and profit when all traders are above average. Journal of Finance, 53(6), 1887–1934. https://doi.org/10.1111/0022-1082.00078 Pieters, G., & Vivanco, S. (2017). Financial regulations and price inconsistencies across Bitcoin markets. Information Economics and Policy, 39, 1–14. https://doi.org/10.1016/j.infoecopol.2017.02.002 Pompian, M. M. (2012). Behavioral Finance and Wealth Management. Evaluation. https://doi.org/10.1002/9781119202400 Ramiah, V., Xu, X., & Moosa, I. A. (2015). Neoclassical finance, behavioral finance and noise traders: A review and assessment of the literature. International Review of Financial Analysis, 41, 89–100. https://doi.org/10.1016/j.irfa.2015.05.021 Ron, D., & Shamir, A. (2013). Quantitative analysis of the full Bitcoin transaction graph. Lecture Notes in Computer Science (Including Subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 7859 LNCS, 6–24. https://doi.org/10.1007/978-3-642-39884-1_2 Rudesill, D. S., Caverlee, J., & Sui, D. (2015). The Deep Web and the Darknet: A Look Inside the Internet’s Massive Black Box. SSRN Electronic Journal, (August). https://doi.org/10.2139/ssrn.2676615 Schwert, G. W. W. (2002). Anomalies and Market Efficiency. SSRN Electronic Journal. https://doi.org/10.2139/ssrn.338080 Shefrin, H. (2006). Beyond Greed and Fear: Understanding Behavioral Finance and the Psychology of Investing. Financial Management Association Survey and Synthesis Series, 1999. https://doi.org/10.1207/S15327760JPFM0102_7 Shleifer, A. (2000). Inefficient Markets: An Introduction to Behavioral Finance. Journal of Policy Modeling (Vol. 158). https://doi.org/10.1628/0932456022975402 Sklaroff, J. M. (2017). Smart Contracts and the Cost of Inflexibility. University of Pennsylvania Law Review, 166(1), 263–303. Retrieved from http://search.ebscohost.com/login.aspx?direct=true&db=a9h&AN=126415305&lang=es &site=ehost-live Sovbetov, Y. (2018). Factors Influencing Cryptocurrency Prices : Evidence from Bitcoin , Ethereum , Dash , Litcoin , and Monero. Journal of Economics and Financial Analysis, - 54 - 2(February), 1–27. https://doi.org/10.1991/jefa.v2i2.a16 Sue, V. M., & Ritter, L. A. (2011). Conducting online surveys. SAGE publications. Telegram. (2018). Telegram primer. Thaler, R. H. (1999). The End of Behavioral Finance. Financial Analysts Journal, 55(6), 12– 17. https://doi.org/10.2469/faj.v55.n6.2310 Urquhart, A. (2017a). Price clustering in Bitcoin. Economics Letters, 159, 145–148. https://doi.org/10.1016/j.econlet.2017.07.035 Urquhart, A. (2017b). The inefficiency of Bitcoin. Economics Letters, 148, 80–82. https://doi.org/https://doi.org/10.1016/j.econlet.2016.09.019 Venegas, P. (2017). Initial Coin Offering (ICO) Risk, Value and Cost in Blockchain Trustless Crypto Markets. Economy Monitor. Retrieved from https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3012238 Verweij, D. (2018). Shooting for the Moon. Tilburg University. Visa. (2015). Visa Inc. at a Glance, (August), 1. Retrieved from https://usa.visa.com/dam/VCOM/download/corporate/media/visa-fact-sheet-Jun2015.pdf Vries, A. De. (2018). Bitcoin ’ s Growing Energy Problem. Joule, 2(5), 801–805. https://doi.org/10.1016/j.joule.2018.04.016 Wright, C. (2017). Proof of Work as it Relates to the Theory of the Firm. NChain. Retrieved from https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2993312 Nakamoto, Satoshi (2008). Bitcoin: A Peer-to-Peer Electronic Cash System. Retrieved from http://bitcoin.org/bitcoin.pdf Jones, C. (2018) Digital currencies and organised crime update. Financial Regulation International. [In Press] Available from: http://eprints.uwe.ac.uk/35541 www.uk.reuters.com/article/eastman-kodak-stocks/eastman-kodak-unveils-cryptocurrencystock-soars-idUKL1N1P416N (Accessed on 02.06.2018) www.cnbc.com/2018/01/04/renren-stock-jumps-after-it-says-its-raising-money-throughico.html (Accessed on 02.06.2018) www.cnbc.com/2017/12/19/tiny-fruit-juice-company-soars-on-speculated-move-into-bitcointhat-likely-isnt-real.html (Accessed on 02.06.2018) www.cnbc.com/2017/10/25/overstock-com-goes-for-largest-digital-coin-offer-ever-at-500million.html (Accessed on 02.06.2018) www.group-ib.com/media/ico-100/ (Accessed on 02.06.2018) www.athcoinindex.com/ (Accessed on 02.06.2018) www.coindesk.com (Accessed on 11.04.2018) - 55 - www.bittrex.com (Accessed on 30.05.2018) www.livecoin.net/en/fees (Accessed on 30.05.2018) www.gdax.com (Accessed on 01.06.2018) www.tradingviews.com (Accessed on 02.06.2018) www.coinmarketcap.com (prices are recorded on 02.06.2018) www.economist.com/blogs/buttonwood/2018/01/tales-crypto-1 (Accessed on 14.05.2018) www.economist.com/blogs/buttonwood/2017/12/crypto-through-tulips (Accessed on 14.05.2018) www.btc.com/stats/pool?pool_mode=year (Accessed on 06.06.2018) www.cryptocompare.com/exchanges/#/overview (Accessed on 06.06.2018) www.esma.europa.eu/press-news/esma-news/esas-warn-consumers-risks-in-buyingvirtual-currencies (Accessed on 07.06.2018) www.theguardian.com/us-news/2018/jun/14/steve-bannon-cryptocurrencydeplorables-bitcoin (Accessed on 14.06.2018) www.ft.com/content/e04e359a-e9e9-3f8e-8e2f-3f4373e5efb0 (Accessed on 15.06.2018) www.ft.com/content/ba6c2c40-285f-11e8-b27e-cc62a39d57a0 (Accessed on 15.06.2018) www.spectrum.ieee.org/computing/networks/why-the-biggest-bitcoin-mines-are-in-china (Accessed on 16.06.2018) www.silkroad7rn2puhj.onion/?road (Accessed through Tor on 04.05.2018) www.fortune.com/2018/01/10/goldman-sachs-bitcoin-currency/ (Accessed on 06.06.2018) www.blockchain.info/charts/n-transactions?timespan=2years&daysAverageString=7 (Accessed on 18.06.2018) https://bitinfocharts.com/comparison/transactionfees-btc-eth-bch-ltc-sma7.html#log&1y (Accessed on 24.06.2018) https://cointelegraph.com/news/chilean-crypto-exchanges-go-to-court-to-fight-the-bankskilling-an-entire-industry (Accessed on 24.06.2018) https://cointelegraph.com/news/indian-banks-suspending-bitcoin-exchange-accounts (Accessed on 24.06.2018) https://cointelegraph.com/news/us-bank-wells-fargo-bans-crypto-purchases-with-its-creditcards (Accessed on 24.06.2018) https://blockgeeks.com/guides/cryptocurrency-hacks/ (Accessed on 24.06.2018) https://www.coindesk.com/google-to-ban-cryptocurrency-ico-ads-from-june/ - 56 - (Accessed on 25.06.2018) - 57 - Appendix Ethnicity Freq. Percent Cumulative African 4 0.84 0.84 Caribbean 5 1.05 1.88 Caucasian 326 68.20 70.08 East Asian 31 6.49 76.57 Hispanic or Latino 18 3.77 80.33 Middle Eastern 22 4.60 84.94 Mixed 19 3.97 88.91 Other 25 5.23 94.14 South Asian 28 5.86 100 Total 478 100 Table A.1 Ethnicity distribution (Verweij, 2018) Gender Freq. Percent Cumulative Male 468 82.11 82.11 Female 102 17.89 100 Total 570 100 Table A.2 Gender frequency (Verweij, 2018) Education Freq. Percent Cumulative High school 71 14.85 14.85 B.Sc./Ms 259 54.18 69.04 Incomplete B.Sc./Ms 82 17.15 86.19 Other 16 3.35 89.19 PhD 21 4.39 93.93 Special vocational/College 15 3.14 9707 Vocational School/Tech school 14 2.93 100 Total 478 100 Table A.3 Education distribution (Verweij, 2018) - 58 - Area of Study Freq. Percent Cumulative Business/Economics 105 21.97 21.97 Education 16 3.35 25.31 Engineering 138 28.87 54.18 Healthcare 34 7.11 61.30 Law 13 2.72 64.02 Media/Culture 19 3.97 67.99 No formal education 22 4.60 72.59 Other 131 27.41 100 Total 478 100 Table A.4 Area of study (Verweij, 2018) Year Freq. Percent Cumulative Before 2015 73 15.27 15.27 2015 18 3.77 19.04 2016 49 10.25 29.29 2017 279 58.37 87.66 2018 59 12.34 100 Total 478 100 Table A.5 Year of first investment in a cryptocurrency (Verweij, 2018) Exchanges Freq. Percent Cumulative Binance 267 55.86 55.86 Bitfinex 10 2.09 57.95 Bitstamp 9 1.88 59.83 Bittrex 25 5.23 65.06 Coinbase 63 13.18 78.24 Huobi 3 0.63 78.87 Kraken 27 5.65 84.52 Other 74 15.48 100 Total 478 100 Table A.6 Top cryptocurrency exchanges by number of responses (Verweij, 2018) - 59 - Exchange Price Binance $6,436.16 Coinbase $6,406.83 Bittrex $6,398.15 Bitstamp $6,399.46 Kraken $6,408.00 Bitfinex $6,471.50 Huobi $6,438.32 Mean $6408.00 Max(LakeBTC) $7,033.81 Min(MBAex) $6,308.77 Coinmarketcap $6,441.74 Table A.7 Cross exchange Bitcoin price differences Note. All prices are in US Dollar, accessed from online sources on May 30th 16:06 Mining Pool January February March April May BTC.com 22.04 25.98 24.38 27.38 25.36 AntPool 18.99 15.36 15.51 13.30 15.37 Other 13.65 14.20 16.57 13.01 11.79 SlushPool 10.52 10.41 11.02 10.76 10.96 ViaBTC 11.75 11.83 10.87 10.74 8.97 BTC.TOP 13.14 11.83 9.22 8.79 8.71 F2Pool 5.48 4.90 6.38 7.59 8.35 DPOOL 0.00 0.00 0.00 3.51 3.75 BTCC 2.41 3.80 4.10 2.60 2.57 BitFury 2.02 1.69 1.95 2.32 2.18 Bixin 0.00 0.00 0.00 0.00 1.99 Table A.8 Bitcoin mining pool market share. Seen in Figure 1 Note. All numbers are percentages (BTC.com) - 60 -