Bitcoin Data Research Articles

Graph-Based LSTM for Anti-money Laundering: Experimenting Temporal Graph Convolutional Network with Bitcoin Data

Elliptic data—one of the largest Bitcoin transaction graphs—has admitted promising results in many studies using classical supervised learning and graph convolutional network models for anti-money laundering. Despite the promising results provided by these studies, only few have considered the temporal information of this dataset, wherein the results were not very satisfactory. Moreover, there is very sparse existing literature that applies active learning to this type of blockchain dataset. In this paper, we develop a classification model that combines long-short-term memory with GCN—referred to as temporal-GCN—that classifies the illicit transactions of Elliptic data using its transaction’s features only. Subsequently, we present an active learning framework applied to the large-scale Bitcoin transaction graph dataset, unlike previous studies on this dataset. Uncertainties for active learning are obtained using Monte-Carlo dropout (MC-dropout) and Monte-Carlo based adversarial attack (MC-AA) which are Bayesian approximations. Active learning frameworks with these methods are compared using various acquisition functions that appeared in the literature. To the best of our knowledge, MC-AA method is the first time to be examined in the context of active learning. Our main finding is that temporal-GCN model has attained significant success in comparison to the previous studies with the same experimental settings on the same dataset. Moreover, we evaluate the performance of the provided acquisition functions using MC-AA and MC-dropout and compare the result against the baseline random sampling model.

ESS: An Efficient Storage Scheme for Improving the Scalability of Bitcoin Network

With the continuous development of blockchain technology, Bitcoin as the first cryptocurrency has drawn massive attention from various sectors. Bitcoin on-chain data storage is over 338GB as of September 1, 2021. According to the exponential growth trend of block data, the size of a Bitcoin full node will exceed 500GB in two years. The huge storage problem makes it difficult for general users to store complete Bitcoin data conveniently, which weakens the decentralization capability of Bitcoin network. In this work, we propose an efficient storage scheme (ESS) based on the distribution characteristics of the unspent transaction outputs in Bitcoin network. ESS sets a UTXO-weight for each block. According to UTXO-weight, it dynamically prunes the blocks which have lower query frequency, improving the scalability of the Bitcoin network. When a new block is generated, ESS enables joining nodes to verify most of the transaction inputs instantly. Only a small amount of transaction outputs of older blocks need to be retrieved from the full node for payment verification. Experimental results demonstrated that ESS can reduce the size of a normal node in current Bitcoin network by about 82.14% at a low communication cost.

PERAMALAN HARGA BITCOIN DENGAN METODE SMOOTH TRANSITION AUTOREGRESSIVE (STAR)

One of the spectacular advances in technology in the economic field is the cryptocurrency it created. The fluctuating price of Bitcoin, is widely used as a means of making profit. The time series forecasting method that can be used for the case of nonlinear time series data such as Bitcoin data is the smooth transition autoregressive (STAR) model. STAR is an extension of the autoregressive model for nonlinear time data. The purpose of this study is to obtain the results of forecasting Bitcoin price data for the next 2 two months using the STAR method. The data used in this study is Bitcoin daily price data from September 2017 to April 2021. To estimate the STAR model, several things that must be determined are the autoregressive model, transition variables, and transition functions. If the STAR model has been estimated, forecasting will be carried out for the next 2 months, which results in the forecast for the highest Bitcoin price falling on June 30, 2021 and the lowest Bitcoin price falling on May 1, 2021.

Combined trading strategy of bitcoin and gold

Market traders buy and sell volatile assets frequently, with a goal to maximize their total return. We have been asked to develop a model that uses only the past stream of daily prices to date to determine each day if the trader should buy, hold, or sell their assets in their portfolio. The assets that can be traded are Bitcoin and gold. We will start with $1000 on 9/11/2016 and try to maximize the total return until 9/10/2021. We will start from forecasting prices and developing trading strategies. In terms of price prediction, we use MSE and Trend_Acc as indicators, and use XGBoost, a representative strong learning algorithm in traditional machine learning, and LSTM, which is good at time series prediction in deep learning, to fit and forecast the data respectively. At first glance, the curve fitting MSE are satisfactory , but, a closer look reveals that the model either firmly remembers the data of the training set, resulting in a lack of generalization ability for unknown data (XGBoost), or tends to take the previous day's results as the predicted results, resulting in a significant lag in the prediction curve (LSTM), all of which are reflected in the models' poor performance in predicting whether prices will rise or fall in the future. In our view, since a large number and complexity of factors affecting prices, it is unrealistic to predict future prices accurately from past prices alone, unless we can get rid of the limitation of the problem, use additional data to assist the prediction, or use all the data as a training set for fitting, we cannot achieve good results in the prediction, but such behavior is inconsistent with our original intention. We established restricted trading model based on composite index judgment. The model not only applies the traditional economic Relative Strength Index and Stochastics Oscillator Index, but also introduces the K-Lipschitz limitation in deep learning into the model. The model dynamically adjusts each transaction strategy according to the changes of working capital and total assets, purchase cost, selling profit and other factors, and the total income of the model is $132433.1. In the horizontal comparison, the profit of our model is more than 39.6%-283.6% than that of the traditional moving average strategy and RSI-STC strategy, and 51.3% higher than that of the random walk model using Montmarlowe algorithm. In addition, we collected the transaction data of bitcoin and gold from 2012-01-01 to 2022-02-21, and applied the model to the historical data, and received good returs. For example, from 2012-1-1 to 2022-2-21, the return was $8418074.9. It is proved that the model has high generalization performance and strong stability. To test the sensitivity of the model to transaction costs, we also analyze the changes in trading strategies that should occur when fees rise. The analysis shows that the traders' single trading volume decreases first and then increases with the increase of the commission fee. The results of sensitivity analysis show that the return change is less than 3%, which proves the robustness of the model. Finally, we made a memo to summarize our work.

Bayesian nonlinear expectation for time series modelling and its application to Bitcoin

This paper proposes a two-stage approach to parametric nonlinear time series modelling in discrete time with the objective of incorporating uncertainty or misspecification in the conditional mean and volatility. At the first stage, a reference or approximating time series model is specified and estimated. At the second stage, Bayesian nonlinear expectations are introduced to incorporate model uncertainty or misspecification in prediction via specifying a family of alternative models. The Bayesian nonlinear expectations for prediction are constructed from closed-form Bayesian credible intervals evaluated using conjugate priors and residuals of the estimated approximating model. Using real Bitcoin data including some periods of Covid 19, applications of the proposed method to forecasting and risk evaluation of Bitcoin are discussed via three major parametric nonlinear time series models, namely the self-exciting threshold autoregressive model, the generalized autoregressive conditional heteroscedasticity model and the stochastic volatility model.

FROM DISCRETE TO CONTINUOUS: GARCH VOLATILITY MODELING OF THE BITCOIN

Volatility is an important concept for identifying and predicting the risk of financial products. The aim of the study is to determine the most appropriate discrete model for the volatility of Bitcoin returns using the discrete-time GARCH model and its extensions and compare it with the Lévy driven continuous-time GARCH model. For this purpose, the volatility of Bitcoin returns is modeled using daily data of Bitcoin / United States Dolar exchange rate. By comparing discrete-time models according to information criteria and likelihood values, the All-GARCH model with Johnson's-SU innovations is found to be the most adequate model. The persistence of the volatility and half-life of the volatility of the returns are calculated according to the estimation of the discrete model. This discrete model has been compared with the continuous model in which the Lévy increments are derived from the compound Poisson process using various error measurements. As a conclusion, it is found that the continuous-time GARCH model shows a better performance to predict the volatility.

Real-Time Big Data Architecture for Processing Cryptocurrency and Social Media Data: A Clustering Approach Based on k-Means

Cryptocurrencies have recently emerged as financial assets that allow their users to execute transactions in a decentralized manner. Their popularity has led to the generation of huge amounts of data, specifically on social media networks such as Twitter. In this study, we propose an iterative kappa architecture that collects, processes, and temporarily stores data regarding transactions and tweets of two of the major cryptocurrencies according to their market capitalization: Bitcoin (BTC) and Ethereum (ETH). We applied a k-means clustering approach to group data according to their principal characteristics. Data are categorized into three groups: BTC typical data, ETH typical data, BTC and ETH atypical data. Findings show that activity on Twitter correlates to activity regarding the transactions of cryptocurrencies. It was also found that around 14% of data relate to extraordinary behaviors regarding cryptocurrencies. These data contain higher transaction volumes of both cryptocurrencies, and about 9.5% more social media publications in comparison with the rest of the data. The main advantages of the proposed architecture are its flexibility and its ability to relate data from various datasets.

Jointly forecasting the value-at-risk and expected shortfall of Bitcoin with a regime-switching CAViaR model

We explore the dynamic tail risk in the Bitcoin market by jointly estimating value-at-risk (VaR) and expected shortfall (ES) using the conditional autoregressive value-at-risk (CAViaR) model. To enable more accurate measurement, we construct a Markov regime-switching (MS) model in which the time-varying transition probability is driven by the information contained in asset price bubbles. This is motivated by prior evidence that bubbles are a key indicator of the economic cycle and contain important information on systemic risk. Using daily Bitcoin data from 2013 to 2021, the results provide strong evidence of a form of regime change in Bitcoin’s VaR and ES. Furthermore, the bubble index has a significant impact on tail risk and improves the model’s ability to estimate and predict VaR and ES.

Attacking Bitcoin anonymity: generative adversarial networks for improving Bitcoin entity classification

Classification of Bitcoin entities is an important task to help Law Enforcement Agencies reduce anonymity in the Bitcoin blockchain network and to detect classes more tied to illegal activities. However, this task is strongly conditioned by a severe class imbalance in Bitcoin datasets. Existing approaches for addressing the class imbalance problem can be improved considering generative adversarial networks (GANs) that can boost data diversity. However, GANs are mainly applied in computer vision and natural language processing tasks, but not in Bitcoin entity behaviour classification where they may be useful for learning and generating synthetic behaviours. Therefore, in this work, we present a novel approach to address the class imbalance in Bitcoin entity classification by applying GANs. In particular, three GAN architectures were implemented and compared in order to find the most suitable architecture for generating Bitcoin entity behaviours. More specifically, GANs were used to address the Bitcoin imbalance problem by generating synthetic data of the less represented classes before training the final entity classifier. The results were used to evaluate the capabilities of the different GAN architectures in terms of training time, performance, repeatability, and computational costs. Finally, the results achieved by the proposed GAN-based resampling were compared with those obtained using five well-known data-level preprocessing techniques. Models trained with data resampled with our GAN-based approach achieved the highest accuracy improvements and were among the best in terms of precision, recall and f1-score. Together with Random Oversampling (ROS), GANs proved to be strong contenders in addressing Bitcoin class imbalance and consequently in reducing Bitcoin entity anonymity (overall and per-class classification performance). To the best of our knowledge, this is the first work to explore the advantages and limitations of GANs in generating specific Bitcoin data and “attacking” Bitcoin anonymity. The proposed methods ultimately demonstrate that in Bitcoin applications, GANs are indeed able to learn the data distribution and generate new samples starting from a very limited class representation, which leads to better detection of classes related to illegal activities.

Modeling and Forecasting the Realized Volatility of Bitcoin using Realized HAR-GARCH-type Models with Jumps and Inverse Leverage Effect

Using the high-frequency data of Bitcoin, this study aims to model the time-varying volatility identified in the residuals of the heterogeneous autoregressive (HAR) model of realized volatility using the symmetric, asymmetric and long-memory generalized autoregressive conditional heteroscedastic models (GARCH) models. We further extended these models by incorporating jumps and continuous components in the realized volatility estimators and investigating the impact of the inverse leverage effect. The Diebold Mariano and model confidence set test confirm that the forecasting performance of HAR-type models can be effectively improved by these innovations. The long memory HAR-GARCH model with jumps and continuous components provided better forecasting accuracy for Bitcoin volatility as compared to other realized volatility models. The findings of this study may benefit individual investors and risk managers who wish to minimize risks and diversify their portfolios to maximize profits in Bitcoin’s investment.

Is There Any Witching in the Cryptocurrency Market?

This paper explores price effects caused by the expiration of derivatives in the cryptocurrency market. Applying different statistical tests (ANOVA, Mann–Whitney, and t-tests) and econometric methods (the modified cumulative abnormal return approach, regression analysis with dummy variables, and the trading simulation approach) to daily and weekly Bitcoin data over the period 2018–2021, the following hypotheses are tested: (H1) Expiration days create patterns in price behavior in the cryptocurrency market; and (H2) Price patterns can be exploited to generate abnormal profits from trading. The results suggest that expiration effects are only nominally present in the cryptocurrency market. There are differences in returns between expiration-related periods and average returns, but these differences are statistically insignificant. The only case in which an anomaly was detected was related to abnormally high returns during the week of expiration: returns during such weeks were positive in 65% of cases, and were on average 5 times higher than during usual weeks. Trading strategies based on this fact were able to generate results different from those of random trading, with a Sharpe ratio above 1. This is evidence in favor of the existence of a real price anomaly, which contradicts the efficient market hypothesis, and this could be implemented in the practice of traders and investors by creating trading strategies based on detected price effects or special technical analysis indicators to generate trading signals. For academics, these results might provide an opportunity to improve time series forecasting analysis in the case of Bitcoin.

Bitcoin Ransomware Detection Employing Rule-Based Algorithms


 Cryptocurrencies have completely altered the digital transaction process all over the globe. Almost a decade after Satoshi Nakamoto generated the first Bitcoin block; many cryptocurrencies have been established. The Ransomware attack is a type of cybercrime and a class of malware that encrypts the files and prevents users from accessing their data or systems and demands payment for decrypting and retrieving access to their files. Ransomware data classification using present data mining and machine learning methods is difficult because predictions aren't always correct. We aim to build two models that effectively address these challenges and can diagnose and classify Ransomware attacks accurately, then compare the performance of the models. In this paper, we investigated the use of Rule-Based algorithms for mining Bitcoin Ransomware Data to classify Ransomware attacks in Bitcoin transactions. Employing Rule-Based techniques in detecting Bitcoin data is beneficial because the algorithms effectively classify non-linear datasets. The analysis was done on a Bitcoin dataset for 61,004 addresses selected from 29 Ransomware families and contained ten descriptive and decision attributes. Both Rule-Based algorithms were illustrated and compared on the dataset employing 10-fold cross-validation. Experimental results show that classification under partial decision tree (PART) algorithm performed better in different metrics than the Decision Table algorithm. It provides an accuracy of 96.01%, a recall of 96%, a precision of 95.9%, and an F-Measure of 95.6%. Experimental results propose that it is beneficial to further investigate the application of PART to predictive modelling tasks in Ransomware studies.

A Diluted Bitcoin-Dollar-Gold Mean Prediction Scheme Based on Periodic Prediction Method

This paper introduces a diluted prediction method for bitcoin and gold based on cycle prediction. This method does not need to quantify the external parameters like robot learning and neural network autoregressive model, but mainly uses ARIMA to feedback the parameter values into risk coefficients under the condition of obtaining the optimal solution circularly, and the price prediction of a single period in the future is carried out with a fixed number of samples, thus realizing the high-precision prediction of bitcoin and gold prices. In the application simulation, the real data of bitcoin and gold from 2016 to 2021 are selected. After 1000 times of Monte Carlo simulations, 919 times of the yield is more than 3 times, 157 times of the yield is more than 8 times, and the minimum yield is about 2 times. At the same time, this paper puts forward an investment strategy for this prediction method, which realizes a very safe profit with a final return rate of 6.2 times under the condition of making full use of the prediction risk coefficient. The prediction method and investment scheme bring a brand-new high-precision prediction method and targeted investment strategy with high safety coefficient to all the investors, which has great economic value.

Time series prediction using machine learning: a case of Bitcoin returns

PurposeThe purpose of this study is to compare five data-driven-based ML techniques to predict the time series data of Bitcoin returns, namely, alternating model tree, random forest (RF), multiple linear regression, multi-layer perceptron regression and M5 Tree algorithms.Design/methodology/approachThe data used to forecast time series data of Bitcoin returns ranges from 8 July 2010 to 30 Aug 2020. This study used several predictors to predict bitcoin returns including economic policy uncertainty, equity market volatility index, S&P returns, USD/EURO exchange rates, oil and gold prices, volatilities and returns. Five statistical indexes, namely, correlation coefficient, mean absolute error, root mean square error, relative absolute error and root relative squared error are determined. The results of these metrices are used to develop colour intensity ranking.FindingsAmong the machine learning (ML) techniques used in this study, RF models has shown superior predictive ability for estimating the Bitcoin returns.Originality/valueThis study is first of its kind to use and compare ML models in the prediction of Bitcoins. More studies can be carried out by using further cryptocurrencies and other ML data-driven models in future.

Examining Heterogeneity Structured on a Large Data Volume with Minimal Incompleteness

While Big Data analytics can provide a variety of benefits, processing heterogeneous data comes with its own set of limitations. A transaction pattern must be studied independently while working with Bitcoin data, this study examines twitter data related to Bitcoin and investigate communications pattern on bitcoin transactional tweet. Using the hashtags #Bitcoin or #BTC on Twitter, a vast amount of data was gathered, which was mined to uncover a pattern that everyone either (speculators, teaches, or the stakeholders) uses on Twitter to discuss Bitcoin transactions. This aim is to determine the direction of Bitcoin transaction tweets based on historical data. As a result, this research proposes using Big Data analytics to track Bitcoin transaction communications in tweets in order to discover a pattern. Hadoop platform MapReduce was used. The finding indicate that In the map step of the procedure, Hadoop's tokenize the dataset and parse them to the mapper where thirteen patterns were established and reduced to three patterns using the attributes previously stored data in the Hadoop context, one of which is the Emoji data that was left out in previous research discussions, but the text is only one piece of the puzzle on bitcoin transaction interaction, and the key part of it is “No certainty, only possibilities” in Bitcoin transactions

Using A Feed Forward Neural Network Algorithm to Predict Prices of Multiple Cryptocurrencies

This paper initially presents a nontechnical overview of cryptocurrency, its history, and the technicalities of its usage as a means of exchange. Bitcoin’s working methodology and mathematical baseline is further presented in more depth. For the remaining majority of the paper, recent cryptocurrency price data of Bitcoin, Ethereum, Tether, Dogecoin, and Binance coin was used to train a machine learning model of Feed Forward Neural Networks to predict future prices for each of the datasets. Further and in conclusion, the results are discussed, and the efficiency and accuracy of these models are evaluated.

Herding intensity and volatility in cryptocurrency markets during the COVID-19

This paper investigates whether herding is present before and during the COVID-19 pandemic, analyzing intraday data of Bitcoin and eight altcoins. The herding intensity measure of Patterson and Sharma (2006) is calculated for the first time for cryptocurrency markets. Furthermore, we employed a novel Granger causality methodology with a Fourier approximation to determine the relationship between herding and volatility, considering the structural breaks. Our results indicate a significant herding behavior, concentrating during the COVID-19 outbreak. The causality test results show that herding has a significant effect on market volatility. Our results do not support the efficient market hypothesis.

The economic dependency of bitcoin security

ABSTRACT We studied the extent to which bitcoin blockchain security permanently depends on the underlying distribution of cryptocurrency market outcomes using daily blockchain and bitcoin data for 2014–2019 and employing the autoregressive-distributed lag (ARDL) approach. We tested three equilibrium hypotheses: (i) sensitivity of the bitcoin blockchain to mining reward, (ii) security outcomes of the bitcoin blockchain and the proof-of-work cost, and (iii) the speed of adjustment of the bitcoin blockchain security to deviations from the equilibrium path. Our results suggest that bitcoin price and mining rewards were intrinsically linked to bitcoin security outcomes.The bitcoin blockchain security’s dependency on mining costs was geographically differenced – it was more significant for the global mining leader China than for other world regions. Bitcoin blockchain security tended to revert relatively fast to its equilibrium security level after the input or output of price shocks.

An investigation of cryptocurrency data: the market that never sleeps

Since the creation of Bitcoin, the adequacy of data in the cryptocurrency market has not been widely analysed by scholars. Indeed, the research conducted by Alexander and Dakos (2020) is the only one that has focused on the properties and differences of several data sources, underlining inconsistencies in the time series of prices. In our paper, we contribute to this strand of the literature by examining one of the main features of digital currencies: the cryptocurrency market never sleeps. Given that cryptocurrencies trade on a 24/7 basis, specialised crypto companies offer two kinds of prices (close and weighted prices) to proxy Bitcoin daily prices. However, scholars and practitioners have not considered this issue in their analyses. We show that these prices are statistically different, which affects the financial decisions of investors and the most relevant fields in the cryptocurrency market (efficiency, risk management and volatility forecasting). Therefore, our paper demonstrates that the data processing used by specialised crypto firms is a relevant issue that changes the underlying mechanism of Bitcoin data, affecting the results of investors and scholars.

Application of Error Correction model with monetary function factors in the return of bitcoin

Bitcoin is a kind of virtual currency based on block chain technology. In many scenarios, it shows the function of currency to a certain extent. However, many believe that the value of Bitcoin is hollow, speculative and unable to undertake monetary functions, which prevents it to circulate as a real currency. This paper assumes that Bitcoin can undertake monetary functions, which can verify the relationship between monetary functions and Bitcoin earnings. According to the availability of the data and the previous research results, the monthly value of the indexes to measure the monetary function from 2010 to 2020 are selected to form a regression equation with the monthly return data of Bitcoin in the corresponding period. ADF test shows that the explained variable is a first-order simple integer, and the explanatory variables have both first-order simple integer and stationary sequence, and there is a cointegration relationship between them. The result of error correction model shows that the relationship between Bitcoin rate of return and the relevant indicators to measure monetary function is not significant, which rejects the original assumption that Bitcoin can assume monetary function, indicating that Bitcoin does not have the ability and potential to assume monetary function. This may be due to the speculative nature of Bitcoin and the lack of powerful credit support. Therefore, investment in Bitcoin needs to be cautiously thought, but the blockchain technology used in Bitcoin needs to be vigorously developed, which could lead to a technological revolution; in addition, it is momentous for countries to develop digital currencies of central bank to meet the challenge.