In recent years, the number of cryptocurrency-related financial institutions has been accelerating. Cryptocurrencies are slowly being incorporated into investment management by asset management companies. Although they have some things in common with more traditional assets, they have some unique properties, and they are still in the process of being understood as assets. Therefore, it is very important to summarize the existing research papers and results on cryptocurrency trading, including available trading platforms, trading signals, trading strategy research and risk management. This article provides a comprehensive overview of cryptocurrency trading research, covering 126 research papers on various aspects of cryptocurrency trading (such as cryptocurrency trading systems, bubbles and extreme conditions, volatility and return forecasts, crypto asset portfolio construction, and crypto assets , Technology transactions, etc.). This article also analyzes data sets, research trends, and the distribution between research objects (content/attributes) and technologies, and finally summarizes some promising opportunities that still exist in cryptocurrency trading.
Although cryptocurrency is a new concept, its rapid development has been widely accepted by the market. Many hedge funds and asset management companies have begun to incorporate cryptocurrency-related assets into their investment portfolios and trading strategies. The academic community has also made considerable efforts in studying cryptocurrency transactions. This article aims to provide a comprehensive overview of the research on cryptocurrency trading, that is, any research aimed at promoting and establishing cryptocurrency trading strategies.
As an emerging market and research direction, cryptocurrency and cryptocurrency trading have made considerable progress, and people's interests and activities have also risen significantly. From Figure 1, we observe that since 2018, more than 85% of the papers have been published, proving the emergence of cryptocurrency trading as a new research field in financial transactions.
According to cryptocurrency transactions, these documents are distributed in six different areas:
Cryptocurrency trading software system (ie real-time trading system, turtle trading system, arbitrage trading system);
System trading, including technical analysis, pair trading and other system trading methods;
New trading techniques, including econometric methods, machine learning techniques and other newer trading methods;
Portfolio and cryptocurrency assets, including symbiotic cryptocurrency and crypto asset portfolio research;
Market conditions research, including bubble or collapse analysis and extreme situations;
Other miscellaneous cryptocurrency trading research.
In this survey, our purpose is to compile the most relevant research in these fields and extract a set of descriptive indicators that can give an idea of the maturity level of research in this field.
There have been related work discussions or partial investigations in the literature related to cryptocurrency transactions. Kyriazis et al. investigated the efficiency and profitable trading opportunities of the cryptocurrency market. Ahamad et al. and Sharma et al. conducted a brief survey of cryptocurrencies. Ujan et al. briefly introduced the cryptocurrency system. Ignasi et al.  conducted a bibliometric analysis of Bitcoin literature. These work results focus on specific areas of cryptocurrency, including cryptocurrency and cryptocurrency market introduction, cryptocurrency system/platform, Bitcoin literature review, etc. But as far as we know, no one has provided a comprehensive survey before, especially focusing on cryptocurrency transactions.
In summary, this article has made the following contributions:
definition. This article defines cryptocurrency trading and divides it into: cryptocurrency market, cryptocurrency trading model and cryptocurrency trading strategy. The core content of this survey is the cryptocurrency trading strategy, and we cover all aspects of it.
Multidisciplinary investigation. This article has conducted a comprehensive survey of 126 cryptocurrency trading papers, covering different disciplines such as financial economics, artificial intelligence, and computer science. Some papers may involve multiple aspects and will be investigated for each category.
analysis. This article analyzes the research distribution, data sets and trends of cryptocurrency trading literature.
Vision. This article points out the challenges faced by cryptocurrency transactions and future research directions, aiming to promote further research.
Figure 2 depicts the paper structure, which is determined by the review mode used. For more details on this, please refer to Section 4.
This section introduces cryptocurrency trading. We will discuss blockchain technology, cryptocurrency market and cryptocurrency trading strategies.
Introduction to Blockchain Technology
Blockchain is a digital ledger of economic transactions. It can be used not only to record financial transactions, but also to record any objects with intrinsic value. The simplest form of blockchain is a series of immutable data records with time stamps, managed by a group of machines that do not belong to any single entity. Each of these data blocks are protected by the principle of encryption and connected to each other in a chain (see the workflow in Figure 3).
Cryptocurrencies like Bitcoin are manufactured on a peer-to-peer network structure. Each peer has a complete history of all transactions, thereby recording the balance of each account. For example, a transaction that says "A pays X bitcoins to B" is signed by A using his private key. After signing, the transaction will be broadcast on the network. When a peer node discovers a new transaction, it will check to ensure that the signature is valid. If the verification is valid, then the block will be added to the chain.
From blockchain to cryptocurrency
Confirmation is a key concept in cryptocurrency; only miners can confirm transactions. Miners add blocks to the blockchain; they retrieve the transaction in the previous block and combine it with the hash of the previous block to obtain its hash, and then store the derived hash in the current block . Miners in the blockchain accept transactions, mark them as legitimate transactions, and broadcast them over the network. After the miner confirms the transaction, each node must add it to its database. In layman's terms, it has become part of the blockchain, and miners are engaged in this work to obtain cryptocurrency tokens, such as Bitcoin. Unlike blockchain, cryptocurrency is related to the use of tokens based on distributed ledger technology. Any transaction involving purchase, sale, investment, etc. involves blockchain native tokens or sub-tokens. Blockchain is a platform that drives cryptocurrency and a technology that acts as a network distributed ledger. The network has created a means of trading, realizing the transfer of value and information. Cryptocurrencies are tokens used in these networks to transfer value and pay for these transactions. They can be considered as tools on the blockchain, and in some cases can also function as resources or utility tools. In other cases, they are used to digitize the value of assets. In short, cryptocurrency is part of an ecosystem based on blockchain technology.
What is cryptocurrency?
Cryptocurrency is a decentralized and decentralized medium of exchange, which uses cryptographic functions for financial transactions . Encrypted currency uses blockchain technology to obtain characteristics such as decentralization, transparency and immutability . Above, we discussed how blockchain technology is applied to cryptocurrencies.
Generally speaking, the security of cryptocurrency is based on cryptography, neither human nor trust. For example, Bitcoin uses a method called "elliptic curve" to ensure that transactions involving Bitcoin are safe. Elliptic curve cryptography is a public key cryptography that relies on mathematics to ensure transaction security. When someone tries to use brute force to bypass the above encryption scheme, they try 250 billion possibilities every second, and it takes one-tenth of the age of the universe to find a value match. As far as its use as currency is concerned, cryptocurrency has the same attributes as currency. Its supply is controllable. Most cryptocurrencies restrict the supply of tokens. For Bitcoin, the supply will decrease over time and will reach its final quantity around 2140. All cryptocurrencies control the supply of tokens through a timetable encoded in the blockchain.
One of the most important characteristics of cryptocurrency is the absence of financial intermediaries. No "middleman" will reduce the transaction cost of traders. In contrast, if a bank’s database is hacked or damaged, the bank will rely solely on its backups to restore any lost or damaged information. With cryptocurrency, even if part of the network is compromised, the rest can still verify transactions correctly. Another important feature of cryptocurrency is that it is not controlled by any central authority: the decentralized nature of the blockchain ensures that cryptocurrency is theoretically free from government control and intervention.
As of December 20, 2019, there are 4950 cryptocurrencies and 20325 cryptocurrency markets; the market value is approximately US$190 billion. Figure 4 shows historical data of global market capitalization and 24-hour trading volume . The total market value is calculated by adding up the dollar market value of all cryptocurrencies. From the graph, we can observe how cryptocurrencies experienced exponential growth in 2017 and experienced a huge bubble burst in early 2018. But in recent years, cryptocurrencies have shown signs of stabilization.
There are three mainstream cryptocurrencies: Bitcoin (BTC), Ethereum (ETH) and Litecoin (LTC). Bitcoin was born in 2009 and gained huge popularity. On October 31, 2008, a person or a group of individuals with the pseudonym Satoshi Nakamoto published the Bitcoin white paper and described it as: "A purely peer-to-peer version of electronic cash, which can be used to transfer via the network without going through a counterparty. One party pays to the other party. Ethereum was launched by Vitalik Buterin in 2015. It is a special blockchain with a token called Ether (the ETH symbol on the exchange). Ethereum is a very The important feature is the ability to create new tokens on the Ethereum blockchain. The Ethereum network was launched on July 30, 2015, and 72 million Ethereum was pre-mined. Litecoin is a peer-to-peer cryptocurrency created by Charlie Lee It was created according to the Bitcoin protocol, but it uses a different hashing algorithm. Litecoin uses the memory-intensive proof-of-work algorithm Scrypt.
Figure 5 shows the percentage of the total market capitalization of cryptocurrencies; Bitcoin and Ethereum account for the vast majority of the total market capitalization (data collected on January 8, 2020).
A cryptocurrency exchange or digital currency exchange (DCE) is a business that allows customers to trade cryptocurrencies. A cryptocurrency exchange can be a market maker, usually using the bid-ask spread as a service commission, or as a matching platform, only charging fees.
Table 1 shows the top or classic cryptocurrency exchanges by number compiled according to the "nomics" website . The Chicago Mercantile Exchange (CME), Chicago Board Options Exchange (CBOE), and BAKKT (backed by the New York Stock Exchange) are all regulated cryptocurrency exchanges. Fiat currency data also comes from the "Economics" website. The currencies supported by regulators and listed exchanges are collected from official websites or blogs.
Cryptocurrency trading is the act of buying and selling cryptocurrencies for profit.
The definition of cryptocurrency trading can be divided into three aspects: trading object, trading mode and trading strategy. The object of cryptocurrency transactions is the asset being traded, that is, "cryptocurrency". The mode of operation of cryptocurrency trading depends on the trading methods of the cryptocurrency market, which can be divided into "contracts for difference (CFD)" (a contract between two parties, usually called "buyer" and "seller", which stipulates that the buyer will be at the end of the position Pay the seller the difference between himself and “buying and selling cryptocurrency through an exchange.” The trading strategy in cryptocurrency trading is an algorithm developed by investors, which defines a set of predefined rules for buying and selling in the cryptocurrency market .
Advantages of cryptocurrency trading
The benefits of cryptocurrency trading include:
Violent fluctuations. The volatility of cryptocurrencies may generally attract speculative interest and investors. The rapid fluctuation of intraday prices can provide traders with a huge opportunity to make money, but it also contains more risks.
24-hour market. The cryptocurrency market is open for trading 24 hours a day, 7 days a week, because it is a decentralized and decentralized market. Unlike traditional trading of stocks and commodities, the cryptocurrency market does not conduct physical transactions in one place. Cryptocurrency transactions can occur between individuals, in different places around the world, as long as they can connect to the Internet.
Almost anonymous. The use of cryptocurrency to purchase goods and services is done online, without the need to disclose their identity. With increasing concerns about identity theft and privacy, cryptocurrency can provide users with some privacy advantages. Different transactions have specific "Know Your Customer" (KYC) metrics used to identify users or customers. KYC's commitment on the exchange allows financial institutions to reduce financial risks while maximizing the anonymity of wallet owners.
P2P peer-to-peer transactions. One of the biggest benefits of cryptocurrency is that it does not require intermediary of financial institutions. As mentioned above, this can reduce transaction costs. In addition, this feature may appeal to users who do not trust traditional systems. In this case, the over-the-counter (OTC) cryptocurrency market provides peer-to-peer transactions on the blockchain. The most famous cryptocurrency OTC market is "LocalBitcoin".
Programmable "smart" function. Some cryptocurrencies can bring other benefits to holders, including limited ownership and voting rights. Cryptocurrency may also include partial ownership interests in physical assets (such as artworks or real estate).
Cryptocurrency trading strategy
Cryptocurrency trading strategies are the focus of this investigation. There are many types of trading strategies, which can be roughly divided into two categories: technical strategies and basic strategies. The similarity between them is that they both rely on quantifiable information that can be back-tested based on historical data to verify their performance. In recent years, the third trading strategy, which we call the quantitative strategy, has received more and more attention. This trading strategy is similar to a technical trading strategy in that it uses information on exchange trading activities to make buying or selling decisions. Quantitative traders use quantitative data to establish trading strategies, which are mainly extracted from prices, trading volumes, technical indicators or ratios, and use market inefficiencies to be automatically executed by trading software. The cryptocurrency market is different from traditional markets in that it has more arbitrage opportunities, higher volatility and transparency. Due to these characteristics, most traders and analysts prefer to use quantitative trading strategies in the cryptocurrency market.
Cryptocurrency trading software system
The software trading system allows international transactions, processing customer accounts and information, accepting and executing trading instructions. The cryptocurrency trading system is a set of pre-programmed procedures that allow transactions between cryptocurrencies and fiat currencies and cryptocurrencies. The cryptocurrency trading system aims to overcome price manipulation, cybercriminal activities, and transaction delays. When developing a cryptocurrency trading system, we must consider the capital market, basic assets, investment plans and strategies. Strategies are the most important part of an effective cryptocurrency trading system. These strategies will be introduced below. There are several commercial cryptocurrency trading systems, such as Capfolio, 3Commas, CCXT, Freqtrade and Ctubio. Through these cryptocurrency trading systems, investors can obtain professional trading strategy support, fairness and transparency from professional third-party consulting companies and fast customer service.
System trading is a way to define trading objectives, risk control and rules. Generally speaking, system trading includes slower investment types such as high-frequency trading and system trend tracking. This article divides systematic cryptocurrency transactions into technical analysis, pairing transactions, etc. The technical analysis in cryptocurrency trading uses historical patterns of transaction data to help traders evaluate current and predict future market conditions in order to conduct profitable transactions. The price and volume chart summarizes all the trading activities of market participants on the exchange and influences their decisions. Some experiments have shown that the use of specific technical trading rules can generate excess returns, which is very useful for cryptocurrency traders and investors to make the best trading and investment decisions. Pair trading is a systematic trading strategy that considers two similar assets with slightly different spreads. If the spread widens, short high stocks and buy low stocks. When the price difference narrows to a certain equilibrium value again, profit will be generated. The papers presented in this section involve the analysis and comparison of technical indicators, matching and informed trading strategies.
Emerging trading technologies
The emerging trading strategies of cryptocurrencies include strategies based on econometrics and machine learning techniques.
Econometric methods use a combination of statistics and economic theory to estimate economic variables and predict their value. Statistical models use mathematical equations to encode information extracted from data. In some cases, statistical modeling techniques can quickly provide sufficiently accurate models. Other methods can also be used, such as sentiment-based predictions and predictions based on the classification of long-term and short-term fluctuations. The prediction of volatility can be used to judge the price fluctuation of cryptocurrency, which is also valuable for the pricing of cryptocurrency-related derivatives .
When using econometrics to study cryptocurrency transactions, researchers apply statistical models to time series data, such as generalized autoregressive conditional heteroscedasticity (GARCH) and BEKK (named after Baba, Engle, Kraft, and Kroner, 1995) Model to evaluate the volatility of cryptocurrencies . Linear statistical model is a method of evaluating the linear relationship between price and explanatory variables . When there are multiple explanatory variables, we can use multiple linear models to simulate the linear relationship between the explanatory variables (independent variables) and the response variables (dependent variables). The linear statistical model commonly used in time series analysis is the autoregressive moving average (ARMA) model.
Machine learning technology
Machine learning is an effective tool for developing trading strategies for Bitcoin and other cryptocurrencies , because it can infer data relationships that humans usually cannot directly observe. From the most basic point of view, machine learning relies on the definition of two main parts: input features and objective functions. The definition of input features (data sources) is where the basic and technical analysis knowledge comes into play. We can divide the input into several groups of features, for example, based on economic indicators (such as GDP indicators, interest rates, etc.), social indicators (Google Trends, Twitter, etc.), technical indicators (price, volume, etc.), and other seasonality The characteristics of the indicator (time, day of the week, etc.). The objective function defines the fitness criterion for judging whether the machine learning model has learned the current task. Typical forecasting models try to predict invisible results by numbers (such as prices) or categories (such as trends). Machine learning models are trained by using historical input data (sometimes called samples), and the patterns are reduced to invisible (out-of-sample) data to (approximately) achieve the goal defined by the objective function. Obviously, in terms of trading, our goal is to infer trading signals from market indicators, which help predict the future return of the asset.
Generalization error is a common problem in practical applications of machine learning, and it has extremely important significance in financial applications. Before we actually use the model to make predictions, we need to use statistical methods (such as cross-validation) to verify the model. In machine learning, this is often referred to as "verification." The process of using machine learning technology to predict cryptocurrency is shown in Figure 6.
According to the form of the main learning loop, we can divide machine learning methods into three categories: supervised learning, unsupervised learning and reinforcement learning. Supervised learning is used to derive prediction functions from labeled training data. Labeled training data means that each training instance contains input and expected output. Usually, these expected outputs are generated by the supervisor and represent the expected behavior of the model. The most used tags in transactions come from future asset returns in the sample. Unsupervised learning attempts to infer structure from unlabeled training data. It can be used to discover hidden patterns in exploratory data analysis or to group data based on any predefined similarity measure. Reinforcement learning utilizes trained software agents to maximize utility functions that define their goals; this is flexible enough to allow agents to trade short-term returns for future returns. In the financial sector, some transaction challenges can be expressed as a game, in which the agent’s goal is to maximize returns at the end of the period.
The application of machine learning in the research of cryptocurrency transactions includes the connection between data source understanding and machine learning model research. Further specific examples will be given in the following section.
Portfolio theory advocates diversification of investment, through the strategic allocation of assets, to maximize returns under a given level of risk. The well-known mean-variance optimization is an outstanding example of this method. There are some common ways to build a diversified portfolio of crypto assets. The first method is cross-market diversification, that is, mixing various investments in the cryptocurrency market's investment portfolio. The second method is to consider industry ecological market segments, that is, avoid investing too much money in any one category. The diversified investment of the cryptocurrency market portfolio includes a cross-cryptocurrency portfolio and a cross-global market portfolio, including stocks and futures.
For cryptocurrencies, market conditions research is particularly important. A financial bubble refers to a sharp rise in asset prices without any change in its intrinsic value. Many experts pointed out that the price of cryptocurrencies increased by 900% in 2017, and a cryptocurrency bubble appeared. In 2018, Bitcoin faced a plunge in value. This significant volatility has prompted researchers to study bubbles and extreme situations in cryptocurrency trading.
Paper collection and reporting mode
This section introduces the scope and methods of our paper collection, basic analysis and survey structure.
This article uses a bottom-up approach to study cryptocurrency trading, from systems to risk management techniques. For the basic trading system, the focus is on optimizing the structure of the trading platform and improving computer science and technology.
At a higher level, researchers focus on designing models to predict returns or fluctuations in the cryptocurrency market. These techniques are very useful for the generation of trading signals. On the next level of the above prediction model, the researchers discussed technical trading methods for trading in the real cryptocurrency market. Bubbles and extreme situations are hot topics in cryptocurrency trading because, as mentioned above, these markets have shown a high degree of volatility (and volatility decreased after the crash). Investment portfolio and cryptocurrency asset management are effective methods to control risks. In risk management research, we group these two areas into one group. Other papers in this survey included topics such as pricing rules, dynamic market analysis, and regulatory implications. Table 2 shows the general scope of cryptocurrency transactions in this survey.
Since many trading strategies and methods in cryptocurrency trading are closely related to stock trading, some researchers transplant or use the latter's research results to the former. When conducting this research, we only consider those papers that focus on cryptocurrency markets or compare these markets with other financial markets.
Specifically, we use the following standards when collecting documents related to cryptocurrency transactions:
1 This article introduces or discusses the general concepts of cryptocurrency trading or related aspects of cryptocurrency trading.
2 This article proposes a method, research or framework with the goal of optimizing the efficiency or accuracy of cryptocurrency transactions.
3 This article compares different methods and perspectives of cryptocurrency trading.
The "cryptocurrency transaction" referred to here refers to one of the terms listed in Table 2 and discussed above.
Some researchers conducted brief investigations on cryptocurrency, cryptocurrency systems , and cryptocurrency trading opportunities. Compared with our surveys, the scope of these surveys is quite limited, and they also include discussions of the latest papers in the field; we want to point out that this is a rapidly developing research field.
In order to collect papers in different fields or platforms, we used keyword searches on the two most popular scientific databases, googlescholar and arXiv. We also chose other public repositories, such as SSRN, but we found that almost all academic papers on these platforms can also be retrieved through Google Scholar; therefore, in our statistical analysis, we count these as Google Scholar hits. We chose arXiv as another source because it allows this investigation to be kept up to date with all the latest findings in the region. The keywords used for search and collection are listed below. [Crypto] refers to the cryptocurrency market, which is our research interest, because different markets may have different methods. Just before October 15, 2019, we conducted 6 searches on these two repositories.
[Crypto] + Trading
[Crypto] + Trading system
[Crypto] + Prediction
[Crypto] + Trading strategy
[Crypto] + Risk Management
[Crypto] + Portfolio
In order to ensure high coverage, we use the so-called snowballing method for every paper found through these keywords. We checked the documents introduced from the snowball method to meet the above criteria until we reached the end.
Table 3 shows the detailed results of our essay collection. The keyword search produced 126 papers in the six research areas of interest in Section 4.1.
Figure 7 shows the distribution of papers published in different research locations. Among all the papers, 45.24% were published in the Journal of Financial Economics (JFE), Cambridge Center for Alternative Finance (CCAF), Finance Research Letters, and the Center for Economic Policy Research (CEPR) ) And "Journal of Risk and Financial Management" (JRFM) and other financial journals; 4.76% of the papers were published in "Journal of Financial Economics" (JFE), "Cambridge Center for Alternative Finance" (CCAF), "Financial Research Express" ( Finance Research Letters), “Centre for Economic Policy Research” (CEPR), “Risk and being a Public Science Library No. 1 (PLOS one), Royal Society Open Science and SAGE; 15.87% of the papers were published in Intelligent engineering and data mining fields, such as the Computational Intelligence Symposium (SSCI), Intelligent Systems Conference (IntelliSys), etc., Intelligent Data Engineering and Automated Learning (IDEAL) and International Conference on Data Mining (ICDM); 4.76% of the papers were published in Physica Class A physical/physician sites (mainly physical sites); 10.32% of the papers were published in artificial intelligence and complex system sites, such as complexity and the International Federation of Information Processing (IFIP); 17.46% of the papers were published in areas containing independent published papers and Other places for dissertations; 1.59% of theses are published on arXiv. The distribution of different venues shows that most cryptocurrency transactions are issued in financial venues, but there are big differences in other aspects.
According to Table 4, we will discuss the contributions of the collected papers and the statistical analysis of these papers in the rest of the paper.
The papers we collected are organized and presented from six perspectives. Section 5 introduces several different cryptocurrency trading software systems. Section 6 introduces system transactions applied to cryptocurrency transactions. In Part 7, we introduced some emerging trading technologies, including cryptocurrency econometrics, machine learning technology, and other emerging trading technologies in the cryptocurrency market. Section 8 introduces the research of cryptocurrency pairs and related factors and the research of crypto asset portfolio. In Section 9, we discuss research on the state of the cryptocurrency market, including bubbles, crash analysis, and extreme situations. Section 10 introduces other research not involved in cryptocurrency trading.
What we want to emphasize is that the above six headings focus on a specific aspect of cryptocurrency trading; we have organized a complete organization of the documents collected under each heading. This means that papers covering multiple aspects will be discussed in different chapters, once for each angle.
In Section 11, we analyzed and compared the number of research papers on different cryptocurrency transaction properties and technologies, and summarized the data set and research timeline of cryptocurrency transactions.
Based on this review, we summarize some future research opportunities in Section 12.
With the development of computer science and cryptocurrency trading, many cryptocurrency trading systems/robots have been developed. Table 5 compares the existing cryptocurrency trading systems on the market. The table is sorted according to URL type (GitHub or official website) and GitHub star likes (if appropriate).
Capfolio is a proprietary system for cryptocurrency trading, a professional analysis platform, and an advanced posterior test engine. It supports five different cryptocurrency exchanges.
3 Commas is a proprietary cryptocurrency trading system platform that can accept profit and loss instructions at the same time. The system is compatible with 12 different cryptocurrency exchanges.
CCXT is a cryptocurrency trading system with a unified API out of the box and optional standardized data, supporting many Bitcoin/Ethereum/Altcoin exchange markets and merchant APIs. Any trader or developer can create trading strategies based on these data and access public transactions through API. The CCXT library is used to connect and trade with global cryptocurrency exchanges and payment processing services. It provides quick access to market data for storage, analysis, visualization, indicator development, algorithmic trading, strategy backtesting, automatic code generation and related software engineering. It is designed for programmers, skilled traders, data scientists and financial analysts to build trading algorithms. Current CCXT functions include:
I. Support many cryptocurrency exchanges;
II. Fully implement public and private API;
III. Optional standardized data for cross-trading analysis and arbitrage;
IV. Unified API out of the box, very easy to integrate.
Blackbird is a C++ bitcoin arbitrage trading system that automatically executes long/short arbitrage transactions between bitcoin exchanges. It can produce a market-neutral strategy that does not transfer funds between exchanges. The motivation behind Blackbird is to naturally profit from these temporary spreads between different exchanges while maintaining market neutrality. Unlike other Bitcoin arbitrage systems, Blackbird does not sell Bitcoin, but actually sells Bitcoin short on short exchanges. This feature provides two important advantages. First, the strategy is always market agnostic: fluctuations in the Bitcoin market (up or down) will not affect the return of the strategy. This eliminates the huge risks of this strategy. Second, this strategy does not require the transfer of funds (USD or BTC) between Bitcoin exchanges. Buying and selling transactions are conducted in parallel on two different exchanges, and there is no need to deal with transmission delays.
StockSharp is an open source trading platform that can be traded in any market in the world, including 48 cryptocurrency exchanges. It has a free C# library and a free trading chart application. Manual or automated trading (algorithmic trading robots, conventional or HFT) can be run on this platform. StockSharp consists of five components that provide different functions:
I. S#. Designer-free general algorithm strategy application, easy to create strategy;
II. Data free software, which can automatically load and store market data;
III. S# terminal-free trading chart application (trading terminal);
IV. Shell—a ready-made graphics framework, which can be changed as needed, and has a completely open source code in C#;
V. API-a free C library for programmers using visual studio. Any trading strategy can be created in S#.API.
Freqtrade is a free and open source cryptocurrency trading robot system written in Python. It is designed to support all major transactions and is controlled by telegram/telegram. It includes backtesting, mapping and money management tools, as well as strategy optimization through machine learning . Freqtrade has the following characteristics:
I. Persistence: Persistence is achieved through SQLite technology;
II. Optimize strategy through machine learning: Use machine learning to optimize your trading strategy parameters and real trading data;
III. Marginal position size: calculate the winning rate, risk-return ratio, optimal stop loss and adjust the position size, and then conduct position transactions for each specific market;
IV. Telegram management: Use telegram to manage robots.
V. Trial run: run the robot at no cost;
CryptoSignal is a professional technical analysis cryptocurrency trading system. Investors can track more than 500 Bitcoins, Bitcoins, Starcoins, Geminis and more. Automatic technical analysis includes momentum, RSI, Ichimoku Cloud, MACD, etc. The system provides alerts, including e-mail, Slack, telegram, etc. CryptoSignal has two main functions. First, it provides modular code for easy implementation of trading strategies; second, it is easy to install with Docker.
Ctubio is a low-latency (high-frequency) cryptocurrency trading system based on C++. This trading system can place or cancel orders through supported cryptocurrency exchanges in less than a few milliseconds. In addition, it also provides a chart system that can visualize the status of trading accounts, including transaction completion, fiat currency target positions, etc.
Catalyst is the analysis and visualization of cryptocurrency trading systems . It makes trading strategies easy to express and back-test on historical data (daily and minute resolution), providing analysis and insights on the performance of specific strategies. Catalyst allows users to share and organize data, and establish profitable, data-driven investment strategies. Catalyst not only supports transaction execution, but also provides historical price data (from minute to daily resolution) of all crypto assets. Catalyst also has back-testing and real-time trading capabilities, allowing users to seamlessly transition between two different trading modes. Finally, Catalyst integrates statistics and machine learning libraries (such as matplotlib, scipy, statsmodels and sklearn) to support the development, analysis, and visualization of the latest trading systems.
Golang Crypto Trading Bot is a Go-based cryptocurrency trading system. Users can test the strategy in a sandbox environment simulation. If the simulation mode is enabled, the fake balance of each token must be specified for each exchange.
Amit et al. developed a real-time cryptocurrency trading system. The real-time encrypted currency trading system consists of a client, a server and a database. Traders use a web application to log in to the server to buy and sell encrypted assets. The server collects cryptocurrency market data by creating scripts that use Coinmarket API. Finally, the database collects balance, transaction and order information from the server. The author tested the system through an experiment that demonstrated a user-friendly and safe experience for traders on the cryptocurrency trading platform.
The original Turtle Trading System was a trend-following trading system developed in the 1970s. The idea was to generate buy and sell signals for stocks for short-term and long-term breakthroughs, and use average true distance (ATR) Measured impairment conditions. The trading system will adjust the asset size based on the volatility of the asset. Basically, if Turtles accumulate positions in a highly volatile market, it will be offset by low volatility positions. The extended turtle trading system has been improved to have a smaller time interval, and the exponential moving average (EMA) rule has been introduced. Three moving average values are used to trigger the "buy" signal: 30 moving average (fast), 60 moving average (slow), and 100 moving average (long). The author of back-tested and compared two trading systems (original turtle and extended turtle) on 8 well-known cryptocurrencies. Through experiments, the original Turtle Trading System has achieved an average net profit margin of 18.59% (net profit as a percentage of total revenue) and an average profitability of 35.94% (winning bids accounted for the total revenue) in 87 transactions in the past year. Percentage of transactions). Expansion Turtle Trading System achieved an average net profit rate of 114.41% and an average profitability of 52.75% in 41 transactions in the same time period. Studies have shown that in cryptocurrency trading, how the extended turtle trading system is improved compared to the original turtle trading system.
Christian introduced an arbitrage trading system for cryptocurrencies. Arbitrage trading aims to discover possible price differences when there are differences in the supply and demand levels of multiple exchanges. Therefore, traders can buy from one exchange and then sell at a higher price on another exchange, thus realizing fast and low-risk profits. Arbitrage trading signals are captured by automated trading software. The technical differences between data sources require the organization of a server process for each data source. Relational databases and SQL are reliable solutions because they have a large amount of relational data. The author used the system to capture arbitrage opportunities in 787 cryptocurrencies on 7 different exchanges on May 25, 2018. The research paper lists the best 10 trading signals generated by the system from the existing 186 trading signals. The results show that the system has captured the "BTG-BTC" trading signal, arbitrage buying on crypto exchanges and arbitrage selling on binary exchanges, and a profit of 495.44% can be obtained. On May 25, 2018, three other good trading arbitrage signals were found (the profit expected by the author is about 20%). The arbitrage trading software system introduced in this article introduces the general principles and implementation methods of the cryptocurrency market arbitrage trading system.
To be continued............