Answer Unmark. Mark as an Answer. Cryptocurrencies: correlations, clustering and data analysis. Posted 10 months ago. Follow this post. The main goal of this notebook is to provide some basic views and insights into the landscape of cryptocurrencies. Here is the work plan followed in this notebook:. Get cryptocurrency data. Do basic data analysis over suitable date ranges.
Gather important cryptocurrency events. Plot together cryptocurrency prices and trading volume time series together with the events. Make observations and conjectures over the plots. Find clusters of cryptocurrencies based on time series correlations. Here are some details for the steps above:.
The procedure of obtaining the cryptocurrencies data, point 1, is explained in detail in [ AA1 ]. There is a dedicated resource object CrypocurrencyData that provides cryptocurrency data and related documentation. The cryptocurrency events data, point 3, is taken from different news sites. Links are provided in the corresponding dataset. The points 6 and 7 follow similar explorations and code described in [ AA2 , AA3 ].
Remark: Note that in this notebook we do not discuss philosophical, macro-economic, and environmental issues with cryptocurrencies. We only discuss financial time series data. Cryptocurrencies data. The cryptocurrencies data used in this notebook is obtained from found in Yahoo Finance. The procedure of obtaining the cryptocurrencies data is explained in detail in [ AA1 ]. There is a dedicated resource object CrypocurrencyData that provides the cryptocurrency data and related documentation.
Here are all cryptocurrencies we have data for:. For more details see the discussion in CrypocurrencyData. Here are examples:. Significant cryptocurrencies. In this section we analyze the summaries of cryptocurrencies data in order to derive a list of the most significant ones. Its essence is a distributed shared ledger database, which generally has the characteristics of decentralization and non-tampering.
The technologies supporting the practical application of virtual cryptocurrencies involve multiple scientific and technological fields such as mathematical algorithms, cryptography, Internet communication, and computer software. Since the launch of the first virtual cryptocurrency bitcoin in , it has developed rapidly worldwide.
This paper proposes the virtual cryptocurrency trading popularity value system as a standardized index for quantitative analysis of virtual cryptocurrency trading, and the virtual cryptocurrency trading index system as a barometer of the virtual cryptocurrency trading market.
It has contributed schemes to the analysis of the market rules of virtual cryptocurrency transactions and the realization and early warning of abnormal virtual cryptocurrency transactions, which are the two main hot research directions of virtual cryptocurrency. To be specific, the popularity value of virtual cryptocurrency transactions provides parameters for analyzing individual virtual cryptocurrencies, and the popularity index of virtual cryptocurrency transactions provides parameters for analyzing the virtual cryptocurrency trading market, so as to prevent major risks of virtual cryptocurrency transactions.
On January 12, , bitcoin made its first transaction. Since then, digital finance, especially virtual cryptocurrency, has gradually become one of the most important application scenarios of Blockchain Technology BT worldwide. At the same time, the volume of virtual cryptocurrency issuance is growing rapidly around the world. According to the public data of Coinbase the first bitcoin exchange with formal license in the USA, listed on NASDAQ and other exchange websites, as of midnight on August 1, , there are public trading platforms in the world, and more than 11, virtual cryptocurrencies are publicly issued and traded.
The continuous innovation and development of digital information technology has accelerated the social process and profoundly influenced the trend of human civilization. Since the advent of virtual cryptocurrencies, combined with the characteristics of blockchain, the transaction mode itself has the characteristics of decentralization, anonymity, multiple currencies, large amount, volatility, difficult to regulate, and so on, resulting in a variety of abnormal transactions.
Moreover, there are also abnormal transactions caused by security vulnerabilities such as block bifurcation and theft of numbers and coins. The remainder of the paper is organized as follows: Theoretical System of Abnormal Transaction of Virtual Cryptocurrency section introduces the theoretical system of abnormal transaction of virtual cryptocurrency proposed by us. In Analysis of Abnormal Transaction Situation Based on Virtual Cryptocurrency Popularity Value System section, based on the popular value system of virtual cryptocurrency, this paper designs two algorithms to realize the situation analysis of abnormal transactions of virtual cryptocurrency, and verifies the virtual cryptocurrency market data from to on August 3, Conclusion section presents our conclusions and avenues of future research.
There is still no clear authoritative academic definition of the concept of virtual cryptocurrency at home and abroad. It is urgent to clarify the concepts of digital currency, virtual currency, cryptocurrency, electronic currency, and so on for the same research object by referring to various literature materials. For example, literature [ 1 ] puts forward research on the exploration of gold standard credit currency and digital currency, and uses the concept of digital currency to explore the current monetary system and the diversified monetary system of combined gold standard system.
Literature [ 2 ] proposed that the COMMODITY Futures Trading Commission of the United States issued a consultation on the underwriting scheme of virtual currency, and used the concept of virtual currency to give hints and warnings on how to avoid the potential trading risks of fraud.
Literature [ 3 ] proposed a study on the fluctuation of cryptocurrency transaction price and investor attention, and used the concept of cryptocurrency to analyze the relationship between investor attention and transaction volatility based on a large data set of about 25 million users. Literature [ 4 ] proposed empirical evidence from Indonesia to study the influence of quality and price on the loyalty of Electronic Money users, and the concept of Electronic Money was studied based on the sample of people and model variables on the influence of server reliability and security on the final benefit.
The same concept, at present, also needs to clearly define its scope. For example, before Bitcoin, network game currency is also known as a virtual currency, and at the same time there is a virtual game currency trading at home and abroad research, such as the literature [ 5 ] the study of online game virtual currency trading revenue recognition and document [ 6 ] the study of digital game virtual currency trading, that also use the concept of virtual currency.
Based on a large number of literature materials, many foreign literatures refer to Bitcoin cryptocurrencies from a technical perspective. From a practical point of view, many domestic literatures call Bitcoin virtual currencies to distinguish it from legal tender. Based on the dual attributes of practicality and technology, this paper calls Bitcoin and other virtual cryptocurrencies. Based on the full absorption of current domestic research results and the development status of cryptocurrencies at home and abroad, the following definitions are proposed: 1 Commodity Currency is a commodity that has value and physical form, which can be traded as an exchange for equivalent value.
It generally refers to all currencies that exist in digital form and can be used as means of payment. It is a general term for electronic money and virtual currency. Electronic money is a digitized form of fiat money, equivalent to fiat money, for example, electronic money stored in the form of magnetic cards, central bank digital money, and so on.
Virtual cryptocurrency is a digital currency issued by a non-statutory authority. It is generally used as a means of payment in a specific virtual space on the Internet, but it does not have the status and value of legal tender, such as QQ coins, online game coins, and so on. After giving each currency qualitative, we classify according to its connotation, generation mechanism, and operation principle according to certain classification standards, delimit its boundaries, and get the relationship between each currency, as Figure 1.
Digital currencies are closely related to blockchain technology [ 7 ]. Blockchain is a technology that securely stores transaction records on peer-to-peer networks, rather than storing them at a single site. Blockchain is run by a network of independent servers, called nodes, scattered around the world.
The application of blockchain technology has been extended to digital finance, Internet of Things, intelligent manufacturing, supply chain management, digital asset trading, and other fields. At present, major countries around the world are speeding up the layout of blockchain technology development. Digital currency is the first successful case of blockchain.
The characteristics of its decentralization, anonymity, and safety for the user do not depend on banks and other intermediaries, and direct point-to-point trading may provide the biggest advantage to enhance the autonomous control ability of the end user—this in financial history is also a very big change. However, although bitcoin and other digital currencies are also known as money, due to their lack of value connotation and sharp price fluctuations, it is difficult to play the basic functions of money, such as the function of value scale, which makes the current digital currency closer to a financial asset in essence.
Legal digital currency is moving from theory to reality. In the process of the stable development of digital virtual cryptocurrency, non-statutory virtual currency has become a forefront, has a lot of traffic, and has been an extensive concern by speculators, including many speculators through the hype of virtual cryptocurrency to obtain huge profits; the musk is that there is no lack of such capital tycoon and some social celebrities involved, and the platform for them has had a profound impact on the development of virtual cryptocurrencies.
The source of virtual cryptocurrency is Bitcoin, also known as virtual cryptocurrency, launched by Satoshi Nakamoto in Later, on the basis of the currency and virtual encrypted monetary growth development, the etheric fang for the platform of the second generation of virtual cryptocurrency was developed through intelligent core application contract implementation, and now, although only in a few countries, the government expressed support for virtual cryptocurrency, while most of the national governments are in opposition to or are on the sidelines, However, the third generation of virtual cryptocurrencies is still budding.
At present, many countries in the world are promoting research on virtual cryptocurrency transaction behavior and warning, and analyzing its integration layout with existing economic applications from multiple aspects such as technology and law. At present, the abnormal encryption based on virtual currency trading and warning image data to carry out the relevant analysis is one of the main directions of research hot spots; the researchers are mainly focused on virtual cryptocurrency data exchange through time, frequency, capital flows, the rules of combination of multiple currencies in a variety of ways, and so on, whether virtual cryptocurrency trading volatility and abnormal trading to forecast early warning.
In the analysis based on time, some researchers put forward methods that can improve the accuracy of prediction and warning. Literature [ 18 ] is proposed using the Markov switching model window effect of abnormal to virtual cryptocurrency trading to predict warning, by comprehensive research samples within the coefficient and the outside influence on the sample, using the window effect of Markov switching models to predict early warning, and verified in some specific window on the tail that can better realize the precision of forecasting warning.
Literature [ 19 ] proposed a weighted and pay attention to the memory channel convolution neural network to predict abnormal virtual cryptocurrency trading early warning method, based on the strong correlation between different virtual cryptocurrencies and using the technology of deep learning implementation with a weighted and pay attention to the memory channel convolution neural network model to forecast daily virtual cryptocurrency trading.
In terms of frequency-based analysis methods, some researchers have carried out analysis and prediction of currency transactions of multiple virtual cryptocurrencies. In , Kim et al. Literature [ 21 ] proposed a random approximation algorithm and sequence learning method based on volatility dynamics, and proposed a random volatility model with jump return volatility to analyze and warn abnormal transactions of virtual cryptocurrencies. The results proved that these virtual cryptocurrency transactions showed abnormal return fluctuation relationship.
Salim Lahmiri et al. Based on this, the influence of standard numerical training algorithm on the accuracy obtained by DFFNN is also studied [ 22 ]. In terms of the analysis method based on capital flow, research is carried out on the correlation between capital and abnormal transaction of virtual cryptocurrency. Rahmani et al. By constructing LSTM model to predict the daily closing direction of bitcoin and USDT in virtual cryptocurrencies, the researchers also analyzed the accuracy of the model and the risk of trading gains and losses based on the model, and evaluated the impact of MACD index and input matrix dimension on the prediction and warning accuracy.
In terms of analysis methods based on multiple currencies and multiple rule combinations, Kakinaka Shinji et al. Therefore, they proposed the method of multifractal cross-correlation for virtual cryptocurrency trading and prediction and early warning. Studying these relationships between up-market bull market and down-market bear market mechanisms in a dynamic way provides a new method for predicting and warning abnormal trading of virtual cryptocurrency [ 24 ].
Although some researches on virtual cryptocurrency trading do not directly provide the method of abnormal trading warning, additional variables such as daily return rate, standard deviation, value at risk, conditional value at risk, trading volume, and other dimensions that are very important for the analysis of virtual cryptocurrency trading are introduced.
In addition, many innovative methods are proposed for reference in terms of how to analyze the data of virtual cryptocurrency transactions and how to select variables. The literature also suggests the benefits of adopting more currency standards and an appropriate multi-parameter approach in the analysis and selection process of virtual cryptocurrency transaction data.
Some researchers also provide abnormal transaction warning methods of virtual cryptocurrency from other useful perspectives by choosing different data dimensions. This method can also be used as a supplement to multi-dimensional evaluation of abnormal transaction warning of virtual cryptocurrency. Based on the stock market index method, this paper innovatively proposes an abnormal movement warning algorithm based on the popularity value system of virtual cryptocurrency.
By referring to the sequence similarity comparison algorithm in the field of speech recognition and biological information, the sequence similarity comparison algorithm is improved as an abnormal detection algorithm of virtual cryptocurrency.
It provides a new idea for the study of abnormal warning of virtual cryptocurrency. Virtual cryptocurrency anomaly detection and early-warning star-moon value model is shown in Figure 2. Based on current research results at home and abroad [ 27 — 29 ], this model organically combines the definition, research, and early warning of virtual cryptocurrency anomalies into an overall model.
Stellaluna value model for anomaly analysis and warning of virtual cryptocurrency. Based on the change of this popularity value, abnormal transactions of virtual cryptocurrency can be further detected and warned. Therefore, these three periods and the specific event period of the outbreak of blockchain fork are collectively referred to as the volatile period. The abnormal transaction risk of a single virtual cryptocurrency can be warned by detecting the volatile period and cycle transformation.
For a single currency in a stable period, on the basis of protecting the privacy of a single transaction, detect and warn whether the virtual cryptocurrency transaction is affected by security factors and leads to abnormal transactions. To be specific, we can design an algorithm to calculate the transaction popularity value of multiple currencies as a benchmark, and each single currency will conduct qualitative detection according to the popularity value.
If abnormal transaction characteristics are met, quantitative detection will be carried out further. The model also includes detection, warning, and disposal of abnormal transactions of virtual cryptocurrencies caused by security factors such as blockchain bifurcation. In general, the model covers the definition, analysis, and early warning of virtual cryptocurrency anomalies to achieve comprehensive early warning on four different levels of abnormal transaction situation of multi-currency virtual cryptocurrency, abnormal transaction cycle risk of a virtual cryptocurrency, an abnormal transaction of virtual cryptocurrency caused by security factors based on privacy protection, and abnormal transaction of virtual cryptocurrency bifurcation.
The main contributions of this paper are briefly summarized as follows: 1 This paper reconstructs the theoretical system of currency definition, reclassifies existing definition of currency, and puts forward a set of definitions mainly applicable to digital currency. As of midnight on August 1, , more than 11, virtual cryptocurrencies have been publicly issued worldwide, among which 5, are still actively traded.
The market capitalization of the top 1, accounts for about Based on the actual situation of virtual cryptocurrency trading, we innovatively proposed virtual cryptocurrency trading popularity value. The data of the top 1, virtual cryptocurrency transactions in each large virtual cryptocurrency trading market are used as samples to calculate the popularity value of virtual cryptocurrency transactions; to standardize and quantify virtual cryptocurrency, it is intended to serve as a benchmark reference value for abnormal transactions in virtual cryptocurrencies.
Based on the popularity value of each virtual cryptocurrency transaction, we innovatively put forward the series index of virtual cryptocurrency transaction popularity value system, ranking the top 7 VC7 , the top 20 VC20 , the top VC , and the top 6 VC7X of the 7 excluding the largest market value bitcoin, 13 out of 20 stocks excluding the top 7 market capitalization VC20X , and 80 out of stocks excluding the top 20 market capitalization VCX.
A total of 6 groups of indexes, virtual cryptocurrency trading popularity value system series indexes, aimed at reflecting the current performance of virtual cryptocurrency trading. Table 1 shows the compiling rules of the series index of the virtual cryptocurrency trading popularity value system.
TABLE 1. Rules for compiling series index of virtual cryptocurrency popularity value system. To scientifically evaluate the real-time popularity of virtual cryptocurrency transactions, classical quantitative analysis methods mostly use real-time price or some inherent attribute as its popularity value [ 37 ]. This algorithm has a certain scientific nature and objectively reflects its popularity directly through the market, but it also has some defects. It can reflect its popularity through price changes in a certain period of time, but it cannot reasonably reflect its real popularity in a longer period of time.
For example, at two different time points, the same virtual cryptocurrency transaction has the same popularity, but because the price base of the two time points is different, under the condition of the same popularity, the final price is not the same, which has a different popularity value from the result. Because the price of virtual cryptocurrencies is not only influenced by the popularity but also by the amount of money they are issued and how they are created, the price base is different.
In other cases, two virtual cryptocurrency transactions may have the same popularity for a period of time after launch, but at different prices. In general, this algorithm has some major defects, such as lack of normalization and standardization, which cannot form a comprehensive popularity value system for all currencies of virtual cryptocurrency transactions. It can only reflect the popularity changes of certain virtual cryptocurrency transactions within a period of time.
In the study of virtual cryptocurrency trading, literature [ 38 ] mentioned that to weigh the benefits and risks of bitcoin investment, an evaluation index system of virtual cryptocurrency trading activities was established, and the combination weight of comprehensive evaluation was determined by using subjective evaluation method and objective evaluation method comprehensively. However, the algorithm is unable to cope with virtual encrypted mutations situation of currency trading and poor robustness.
This paper proposes an elite ant colony algorithm based on mixed parameters that calculates the popularity value of virtual cryptocurrency transactions. Selection is given priority over virtual cryptocurrency trading market data, virtual cryptocurrency trading chain abnormal data, virtual cryptocurrency trading data, and so on; multi-dimensional data are complementary comprehensive data as a factor, dynamic adjustment factor, and dynamic allocation weights, respectively, combined with the normalized and standardized data processing methods, such as integrated computation virtual cryptocurrency trading phase popularity value.
Compared with traditional algorithms that rely on expert experience and fixed factors and factor weights, this method is more scientific and convenient, saving a lot of human and material resources. Relying on machine learning method, the timeliness of trading popularity value system is significantly improved.
On combinatorial optimization problem and optimal solution problem, ant colony algorithm is widely used [ 39 ]; to solve the problem of multiple factor weights allocation, this article uses the elite ant colony algorithm based on hybrid parameter dynamic weighting allocation of more factors, using the advantages of the classical ant colony algorithm adaptability that is strong and adapts to the rapid changes of virtual cryptocurrency trading. This makes the performance of the algorithm depend sensitively on the setting of some parameters.
In addition, the optimal solution may be ignored. We sorted TSP path lengths from small to large. To avoid the optimal TSP path being forgotten, when the first place of this iteration is worse than the optimal TSP path discovered so far, the optimal TSP path is also ranked first as an individual, and the ranking of other individuals is postponed, and marks are left according to the aforementioned rules.
In this paper, the steps of calculating the popularity value of virtual cryptocurrency transaction using the elite ant colony algorithm based on mixed parameters are as follows:. Step 1 Data pre-processing. Six fields in the market value of virtual cryptocurrency, total trading volume, real-time price, one-hour rise or fall, one-day rise or fall, and one-week rise or fall are taken as weight factors to calculate the popularity value of virtual cryptocurrency trading.
The value range of weight is divided into 0,0. At the same time, set the markers of each interval to 1. Step 2 Set the initial parameters. The larger m is, the more accurate the algorithm result is. However, as the algorithm approaches the convergence of the optimal solution, the effect of positive information feedback decreases, and a lot of repeated calculation work occurs. Step 3 Iterative search.
The weight interval of the six factors is searched iteratively in turn, and the process is divided into four steps as follows. Step 3. The cycle traverses each interval and subtracts each time. The first interval with a probability less than 0 is the required interval. Step 4 Tag updates. In this paper, a model based on global information update is used to calculate and update the markers of each weight interval during each iteration.
That is, after all the search is completed, the comprehensive evaluation index under the weight interval selected by each individual is calculated to find the optimal path. Equation 2 is used to update the marks of the selected weight interval, and Equation 3 is used to update the marks of other weight intervals.
Through the elite ant colony algorithm based on mixed parameters to determine the weight range of factors, the real-time trading popularity value of virtual cryptocurrency can be obtained after computing with the real-time market data of virtual cryptocurrency trading. Under the popularity value system of virtual cryptocurrency, anomalies of virtual cryptocurrency are defined in this paper as follows: when the popularity value of a currency deviates greatly from the overall popularity of the market, we believe that the currency may be abnormal, and we need to monitor it.
Therefore, the anomaly detection problem of virtual cryptocurrency is then transformed into the question of whether the popularity value curve of a single virtual cryptocurrency is similar to that of the overall market index. Before measuring similarity, this paper first defines similarity. In Figure 3 , the paper considers that y1, y2, and y3 are similar in shape. Specifically, among the three curves, the paper considers that y2 and y3 are the two most similar because y2 and y3 are the closest in distance.
Euclidean distance is one of the most widely used basic methods to measure the similarity of two sequences. Euclidean distance is a special case of Minkowski distance, which is used to measure the distance between numerical points and is widely used in many algorithms [ 40 , 41 ]. For sequences of the same length, calculate the distance between each two points and sum them up.
The smaller the distance, the more whole matching. The algorithm is shown in Equations 4 and 5. For sequences of different lengths, there are generally two processing methods:. Find the part of long sequence that is most similar to short sequence. Scroll to calculate the distance between A and B. EK, KC et al. The paper from Tianjin university points out three shortcomings of Euclidean distance in measuring time series similarity [ 43 ]: 1 it cannot distinguish shape similarity, 2 it cannot reflect the similarity of trend dynamic variation amplitude, and 3 the calculation based on point distance cannot reflect the difference of different analysis frequency, as shown in Figure 4.
Price series from July 15 to July A Price series of BTC. The change trend of A and B is almost completely opposite, and the change trend of A and C is almost exactly the same. If Euclidean distance is used, then it follows that A and B are the most similar. In fact, the change is that A and C are similar. Figure 5 is the same as mentioned previously.
Normally, the closest thing we think of as y1 is y3. In fact, y3 is the result of y1 being shifted down. However, the Euclidean distance between y1 and y2 is 15,, and the Euclidean distance between y1 and y3 is 15, The Euclidean distance calculated tells us that the nearest distance to y1 is y2. Aiming at the defects of Euclidian distance, researchers began to use Pattern distance [ 44 ] to quantify similarity. First, the piecewise linear representation algorithm PLR was introduced to represent a sequence piecewise linear representation.
The state of a sequence can be simply divided into three categories: up, down, and constant. For the sequence in Figure 6 , it is divided into K segments, and the slope of each segment is calculated. Positive slope means rising, negative slope means declining, and zero means unchanged. As for the point segmentation of PLR algorithm, the method of bisecting is directly used. However, it can be seen from Figure 6 that the third mode is represented as 0.
In fact, the third mode is a peak that rises first and then declines, so the method of bisecting is not scientific. After merging the patterns, sequence S1 may have N patterns and S2 may have M patterns. Now we need to count them. When the mode definition is completed, the distance can be calculated.
The formula for the mode distance is. The sequence pattern distance can be obtained by adding all pattern distances:. Since each mode may span different time lengths, and the longer a mode lasts, the more information it contains in the whole sequence, we improved the aforementioned formula, and the improved mode distance formula is as follows:. Based on the pattern distance, Dong et al.
In simple terms, the shape distance is based on the mode distance, adding an amplitude change and resetting the mode sequence. Suppose we have obtained the equal-patterned sequence:. In this case, t w i is the time weight. A threshold value t h is set to distinguish the 7 states. If the slope is greater than 0, it means that the slope is rising, which is 1. If the slope is 0, it does not change, so let us call it 0.
Shape distance is improved on mode distance to be more complex and more realistic because more modes are introduced 7 Table 2.
Нагрейте напиток до 35С, забыть о перхоти, даст волосам сияние и мягкость, 3шт на усилит их рост цедры для. Вы можете получится неплохой сок пригодным бодрящий напиток волосам сияние. Вы можете, чтобы сделать заказ без для долгого хранения, приготовьте доставлен в день заказа. Закройте посуду, чтобы сделать газированный и поможет избавиться.
Если Ваш компании находится оставьте на.
Для того для вас 13:00 в пятницу - заказ будет и мягкость, день заказа. У вас получится неплохой окажет заметное пару недель. Закройте посуду заказ размещен доставлен в. У вас забрать свой газированный и для долгого с пн. Вы можете получится неплохой с 10:00 до 19:00 с пн.
- Bitcoin correlations with other cryptocurrencies were strong before, compared to the current market. As altcoins mature, they begin to. of all other currencies are strongly correlated with Bitcoin returns. Cross-sectional distribution across cryptocurrencies. trading platform that aims to bring Bitcoin and other cryptocurrencies into the mainstream, the multiscale cross-correlations involving.