Over the past few years, with the advent of blockchain technology, there has been a massive increase in the usage of Cryptocurrencies. However, Cryptocurrencies are not seen as an investment opportunity due to the market's erratic behavior and high price volatility. Most of the solutions reported in the literature for price forecasting of Cryptocurrencies may not be applicable for real-time price prediction due to their deterministic nature. Motivated by the aforementioned issues, we propose a stochastic neural network model for Cryptocurrency price prediction. The proposed approach is based on the random walk theory, which is widely used in financial markets for modeling stock prices. The proposed model induces layer-wise randomness into the observed feature activations of neural networks to simulate market volatility. Moreover, a technique to learn the pattern of the reaction of the market is also included in the prediction model. We trained the Multi-Layer Perceptron (MLP) and Long Short-Term Memory (LSTM) models for Bitcoin, Ethereum, and Litecoin. The results show that the proposed model is superior in comparison to the deterministic models.
Bitcoin is considered to be most valuable and expensive currency in the world. Besides being first decentralized digital currency, its value has also experienced a steep increase, from around 1 dollar in 2010 to around 18000 in 2017. In recent years, it has attracted considerable attention in a diverse set of fields, including economics, finance and computer science. In economics, the primary focus has always been on studying how it affects the market, determining reasons behinds its price fluctuations, and predicting its future prices. In computer science, the focus is on its vulnerabilities, scalability, and other techno-cryptoeconomic issues. Firstly, we are going to collect the historical data of Bitcoin prices over the years 2013 to 2019 and do prediction for the year 2020. We have aimed to justify the usefulness of traditional Autoregressive Integrative Moving Average (ARIMA) model for predicting bitcoin prices. We have predicted the closing price of bitcoin for first seven days of January 2020. Further, we have created web services using ASP.NET to make the predictions on bitcoin price online and lastly, we have plotted the results in a responsive chart using Highcharts.
Günümüzde kripto para birimlerinin önemi gittikçe artmaktadır. Kripto para birimleri sanal oyun platformlarında kullanılırken, şu an pek çok kurum ve kuruluş tarafından ödeme aracı olarak kullanılmaktadır. Güvenlik risklerine karşı blockchain (Blok Zinciri) adı verilen algoritması ile üretimi sağlanmaktadır. Kripto para fiyatlarının doğru olarak tahmin edilmesi yatırımcı ve karar vericiler açısından büyük önem taşımaktadır. Bu çalışma kapsamında en çok kullanılan dört kripto para birimine (Bitcoin, Ethereum, Ripple, Litecoin) ait fiyat değerleri tahmin edilmiştir. Çoklu kırılma testinden yararlanılarak her seriye ait kırılmalar belirlenerek analiz genişletilmiştir. Ele alınan sanal para değerlerini doğru bir şekilde tahmin etmek amacıyla hem klasik zaman serisi modellerinden hem de üç farklı tür yapay sinir ağı modelinden faydalanılmıştır. Ayrıca elde edilen tahminler üzerinde basit birleştirilme teknikleri uygulanmıştır. Rassal yürüyüşün egemen olduğu bu seriler arasından, özellikle işlem hacmi ve bilinilirliği en fazla olan Bitcoin sanal parasında rassal yürüyüş modelinden daha iyi sonuçlar elde edildiği gözlemlenmiştir.
In this study, the relationship between the popularity of cryptocurrencies and their price, return and trading volumes are examined through time series analysis. The popularity variable is determined according the frequency of cryptocurrencies being searched on the internet. Stationarity of series is examined by Vogelsang and Perron (1998) structural breaks ADF unit root test. According to the test results, all series are found to be stationary at level values. VAR analyses and impulse-response functions are performed to reveal dynamic interaction between the series. According to impulse - response test results, returns of BITCOIN decreased against a decreasing shock in the number searches on the internet and its price and trading volume followed a fluctuating course. In order to see the causality relationship between variables the Granger causality test is conducted. Regression analyses are performed using ordinary least squares (OLS) method through three different equations. According to the result of the regression analysis, an increase in the number of internet searches for cryptocurrencies was found to positively affect prices, returns and trading volumes of all cryptocurrencies. The highest impact on prices and trading volume is observed in BITCOIN, while the highest effect on returns is observed in LITECOIN. According to the findings, popularity can be considered an important determinant for price, returns and trading volumes of cryptocurrencies.
Bu çalışma, Bitcoin’in fiyat davranışını otoregresif birim kökü olan iki rejimli bir TAR modeli kullanarak araştırmaktadır. Çalışmada, durağan dışılığı ve doğrusal olmamayı eş zamanlı olarak sınayan Caner ve Hansen (2001) tarafından geliştirilen yöntem kullanılmıştır. Bu amaçla, 16.07.2010 – 27.11.2018 dönemi için (3.056 adet günlük gözlem) Bitcoin kapanış fiyatlarına ait veri seti oluşturularak Bitcoin fiyatlarının etkin olup olmadığı incelenmiştir. Elde edilen bulgular, Bitcoin fiyatlarının tüm dönem dikkate alındığında zayıf formda etkin piyasalar hipotezini desteklemektedir. Ancak rejimler arası geçiş dikkate alındığında Bitcoin fiyat serisinde iki rejim olduğu sonucuna ulaşılmıştır. Birinci rejimde zayıf forma etkin piyasalar hipotezinin geçerli olduğu, ancak ikinci rejimde geçerli olmadığı tespit edilmiştir.
The aim of this study is to develop forecasting cryptocurrency price movement using moving average. The cryptocurreny that selected in this study is Bitcoin Cash. The observation periods involved in this
Ángeles López Cabarcos, Ada M. Pérez-Pico, Juan Piñeiro Chousa, Aleksandar Šević
Bitcoin is the cryptocurrency with the largest market capitalization, and many studies have examined its role in financial markets. In this manuscript, we contribute to the extant body of knowledge by analyzing the Bitcoin behavior and the effect that investor sentiment, S&P 500 returns, and VIX returns have on Bitcoin volatility using GARCH and EGARCH models. The results suggest that Bitcoin volatility is more unstable in speculative periods. In stable periods, S&P 500 returns, VIX returns, and sentiment influence Bitcoin volatility.
İnternet kullanımındaki hızlı gelişmeler ile birlikte, insan hayatına fiziksel olarak dahil olan para da dijitalleşmeye başlamıştır. Bu tür dijitalleşmiş para birimlerine genel olarak kripto para denilmektedir. Hali hazırda, Bitcoin, kripto para birimleri arasında en yüksek işlem hacmine sahiptir. İlk Bitcoin 2009 yılında piyasaya sürüldü. Fakat son birkaç yılda ciddi derecede ilgi çekmeye başladı. Bu ilginin temel nedenlerinden birisi, Bitcoin'in değerinde önemli artışların olmasıdır. Söz konusu değer artışları bağlamında, Bitcoin piyasasında spekülatif balonların varlığının araştırılması önem arz etmektedir. Bu bağlamda, çalışmanın amacı 2015-2018 dönemi boyunca Bitcoin piyasasında spekülatif balonların varlığını araştırmaktır. Amaç doğrultusunda, spekülatif balonların tespiti için Phillips vd. (2015) tarafından geliştirilen Genelleştirilmiş Eküs ADF testi kullanılmıştır. Elde edilen bulgular, Bitcoin piyasasında çok sayıda baloncuk olduğunu göstermektedir.
Cryptocurrency market is very volatile, trading prices for some tokens can experience a sudden spike up or downturn in a matter of minutes. As a result, traders are facing difficulty following with all the trading price movements unless they are monitoring them manually. Hence, we propose a real-time alert system for monitoring those trading prices, sending notifications to users if any target prices match or an anomaly occurs. We adopt a streaming platform as the backbone of our system. It can handle thousands of messages per second with low latency rate at an average of 19 seconds on our testing environment. Long-Short-Term-Memory (LSTM) model is used as an anomaly detector. We compare the impact of five different data normalisation approaches with LSTM model on Bitcoin price dataset. The result shows that decimal scaling produces only Mean Absolute Percentage Error (MAPE) of 8.4 per cent prediction error rate on daily price data, which is the best performance achieved compared to other observed methods. However, with one-minute price dataset, our model produces higher prediction error making it impractical to distinguish between normal and anomaly points of price movement.
Current research shows that stock market price, collected as a type of time-series data, could be forecasted by machine learning. The pricing data of cryptocurrency could also be used to conduct time-series prediction by leveraging different models, such as Long Short-Term Memory, Bayesian regression, GLM/Random Forest. This paper compares some of the machine learning methods used in predicting the price of cryptocurrencies by illustrating the nature of cryptocurrency, data availability, model used, results associated, and challenges.
Stock exchanges around the world are exploring the best possible solution that can improve trading efficiency, lower the risks and tighten secu- rity levels. The working and functioning of a stock exchange involves very hectic and cumbersome pro- cedures which are time consuming, cost inefficient and can be prone to numerous risks. Machine learning and Blockchain are most popular upcoming technologies. In this paper we present a novel secure and de- centralized intelligent stock market prediction model. We present a blockchain based solution for stock exchange model that uses machine learning accessible smart contracts. The machine learning model makes a prediction on the future of the stock market providing an intelligent solution for secure stock market.
Temiloluwa I. Adegboruwa, Steve A. Adeshina, Moussa Mahamat Boukar
Bitcoin is the first digital currency that uses decentralization to solve the issue of trust in performing the functions of a digital currency successfully. This digital currency has shown extraordinary growth and intermittent plunge in value and market capitalization over time. This makes it important to understand what determines the volatility of bitcoin and to what extent they are predictable. Long Short Term Memory Neural Networks (LSTM-NN) have recently grown popular for time series prediction systems but there has been no consensus on methods to model time series inputs for LSTMs, this paper proposes the need for this problem to be solved by conducting an experimental research on the efficacy of an LSTM-NN given the form of its time-series input features.
Bitcoin is a peer-to-peer (P2P) electronic currency that allows online payments around the world without the management of a third party. Many studies have been conducted on the performance prediction of the stock market; in particular, the Autoregressive Integrated Moving Average model (ARIMA) is one of the linear regressive models widely used in the time series. Nevertheless, as artificial intelligence has become a heated research topic in modern days, A number of studies have also shown that the Back-propagation Neural Network (BPNN) is very effective in prediction. Hence, this paper compares the ARIMA model with the BPNN model in the prediction of Bitcoin price.
Stephen Wooley, Andrew N. Edmonds, Arunkumar Bagavathi, Siddharth Krishnan
Explosive growth in the value of cryptocurrencies like Bitcoin and Ethereum in recent years has attracted the attention of many speculators. Unlike traditional currencies, cryptocurrencies are not backed by any government agencies resulting in prices being strongly influenced by public opinion. Understanding the relationship between cryptocurrency prices and the public sentiment can lead to improved predictions of price movement. In this paper, we give an exploratory analysis of a network of 24 Reddit communities related to Bitcoin, Ethereum, or other cryptocurrencies to analyze Bitcoin and Ethereum price movements. We engineer a set of 112 time series features from submissions and comments made on the selected subreddits, run Granger causality tests on engineered time series against cryptocurrency price movements, and use these time series to forecast the cryptocurrency price movements using classification models. Results from these models support the Granger causality test results showing that with only lagged price values and lagged values from a single Reddit data derived feature, the direction of Bitcoin and Ethereum price movements can be predicted with 74.2% and 73.1% accuracy respectively.
Bitcoin alongside other cryptocurrencies became one of the largest trends recently, due to its redefinition of the concept of money, and its price fluctuation. Especially on the social media, people keep discussing Bitcoin topics, consulting, and advising about cryptocurrency trading. This paper explores the relationship between Twitter feed on Bitcoin and sentiment analysis of it, comparing and evaluating different data mining classifiers and deep learning methods that might help in better sentiment classification of Bitcoin tweets, the study uses different language modeling approaches, such as tweet embedding and N-Gram modeling. We also evaluate the quality of automated sentiment classification in comparison to manually assigned sentiment labeling. The results show that the manual approach gives significantly better results in some datasets, and superior performance of MLP, WiSARD and decision tree methods. On the other hand, R-Auto Tweets Sentiment (RATS) gives more stable performance overall datasets. using time-series, we found partial correlation between Bitcoin price fluctuation and sentiment class accuracy fluctuations using different machine learning algorithms.
The world has more than 5000 digital-currencies, bitcoin is one of it, which has more than 5.8 million dynamic client and approximately more than 111 exchanges throughout the world. So, the aim for this paper is to do the near prediction of the price of Bitcoin in USD. Precious details are taken from the price index of Bitcoin. A Bayesian recurrent hierarchical (RNN) neural network and a long-term memory (LSTM) network can accomplish this function. The total identification accuracy of 52% and an 8% RMSE is obtained by the LSTM. In contrast to the profound training systems, the common ARIMA method for the prediction of time series. This model have not much efficient as deep learning model can be performed. The deep learning methods were predicted to outperform the poorly performing ARIMA prediction. So here we used Gated Recurrent Network model (GRU) to forecasting Bitcoin price Eventually, all deep learning models have a GPU and CPU that beat the GPU implemented by 94.70 percent for their GPU training time.
Cryptocurrencies, such as Bitcoin, are becoming increasingly popular, having been widely used as an exchange medium in areas such as financial transaction and asset transfer verification. However, there has been a lack of solutions that can support real-time price prediction to cope with high currency volatility, handle massive heterogeneous data volumes, including social media sentiments, while supporting fault tolerance and persistence in real time, and provide real-time adaptation of learning algorithms to cope with new price and sentiment data. In this paper we introduce KryptoOracle, a novel real-time and adaptive cryptocurrency price prediction platform based on Twitter sentiments. The integrative and modular platform is based on (i) a Spark-based architecture which handles the large volume of incoming data in a persistent and fault tolerant way; (ii) an approach that supports sentiment analysis which can respond to large amounts of natural language processing queries in real time; and (iii) a predictive method grounded on online learning in which a model adapts its weights to cope with new prices and sentiments. Besides providing an architectural design, the paper also describes the KryptoOracle platform implementation and experimental evaluation. Overall, the proposed platform can help accelerate decision-making, uncover new opportunities and provide more timely insights based on the available and ever-larger financial data volume and variety.
This study aims to evaluate the effect of adding bitcoin in a diversified portfolio comprising traditional assets (bonds, European, Asian and international stock market indices) and alternative assets (gold and commodities) from an European investor point of view. Monthly data cover the period from August 2010 to March 2016. This period is divided into two sub-periods during the euro zone debt crisis and after the crisis. To do this, we will, first of all, apply the genetic algorithms method to optimize two types of portfolio with and without bitcoin for both subperiods. Next, we will compare the two optimal portfolios using the stochastic dominance approach during the two sub-periods. Genetic algorithms show that the weighting of bitcoin during the crisis is greater than that after the crisis, which proves that bitcoin has a safe haven value during unstable periods. The results of stochastic dominance show that during and after the crisis, the portfolio including bitcoin dominates the one without bitcoin according to the 2nd and 3rd order. This shows that risk-averse investors prefer to include bitcoin in their portfolios to maximize their expected utility.
Bitcoin’s value is highly dependent on the communities that use it. This network effect is true for all new technologies. Today’s online communities are so large in population that both the Facebook user and Youtuber populations have surpassed the Chinese population. We take a big data approach using millions of samples of posts from Twitter, Telegram, and Reddit to study how and if social media platforms, the epitome of online communities, affect Bitcoin’s price and volume as well as the price and volume of fifteen other top cryptocurrencies. We work in collaboration with Solume, a data centered fin-tech startup, as well as with Sentistrength, an opinion mining tool developed by researchers in the UK, to classify the sentiment of the millions of posts we study. We collected millions of posts related to 16 cryptocurrencies from November 2017 through August 2018 on an hourly basis and explore social media volume sentiment effect on these cryptocurrencies. Findings confirm that volumes of exchanged posts may predict the fluctuations of Bitcoin’s price but mainly, they predict volume. We also find that Reddit and Telegram posts have greater impact on Bitcoin volume than Twitter. Results indicate that information about the use of social media platforms can assist in tracking real world behavior and may even predict real financial market trends.
Over the last few years, cryptocurrencies have turned into a worldwide wonder known to many people. Bitcoin and Ethereum are widely used at present in global financial markets and so far, have received a significant value improvement and market capitalization with Ethereum being more impervious to downgrading than the other cryptocurrencies. The main aim of this study is to ascertain the prediction accuracy of both Bitcoin and Ethereum currencies using historical data (blockchain data with cryptocurrency data). This is the first study of this kind that we are aware of that predicts cryptocurrency prices using a Deep learning algorithm and blockchain information (Bitcoin and Ethereum). To accomplish this task, the blockchain data (2015–2018) of both Bitcoin and Ethereum currencies were collected to enhance the security and the prediction rate. Further, the market price of Bitcoin and Ethereum currencies were retrieved online. The effectiveness of the prediction accuracy was investigated using Deep learning approach with crypto currency data and blockchain data. The results of data analysis showed that when the blockchain data were used together with Bitcoin and Ethereum prices, the prediction performance is high for both currencies. In addition, the comparison between Bitcoin and Ethereum revealed that Ethereum currency has the highest percentage of the prediction accuracy and the lowest error rate. Moreover, descriptive analysis was undertaken for blockchain attributes such as difficulty, hash rate, number of transactions, average block size and miner’s revenue. The blockchain data directly influenced the prediction accuracy of both currencies (i.e., Bitcoin and Ethereum). The Deep learning prediction approach was found to be very effective for analysing blockchain and cryptocurrency data set. The price prediction of the cryptocurrency market price is vital as Bitcoin and Ethereum are effective in the present financial market.
With data accumulated at a rapid phase through multiple channels, algorithmic trading becomes critical in stock markets and crypto markets. In algorithmic trading, an innovative approach to integrating machine learning can provide data-driven solutions to help people invest with minimal risk and maximum returns. This study explores various machine learning techniques to model the nonlinear relationship between bitcoin prices and social sentiment data and predict the price values with some lead time. Also, the cryptocurrency market is very volatile and lacking strict governing bodies and regulators across regions making it more complex and challenging to predict the prices. Through the analysis, it is found that the sentiment data model is superior in capturing the nonlinear relationship compared to the conventional methods of technical indicators and decision trees, while the neural network models are robust and offer better accuracy in predicting bitcoin price.
David Zhao, Alessandro Rinaldo, Christopher Brookins
Few assets in financial history have been as notoriously volatile as cryptocurrencies. While the long term outlook for this asset class remains unclear, we are successful in making short term price predictions for several major crypto assets. Using historical data from July 2015 to November 2019, we develop a large number of technical indicators to capture patterns in the cryptocurrency market. We then test various classification methods to forecast short-term future price movements based on these indicators. On both PPV and NPV metrics, our classifiers do well in identifying up and down market moves over the next 1 hour. Beyond evaluating classification accuracy, we also develop a strategy for translating 1-hour-ahead class predictions into trading decisions, along with a backtester that simulates trading in a realistic environment. We find that support vector machines yield the most profitable trading strategies, which outperform the market on average for Bitcoin, Ethereum and Litecoin over the past 22 months, since January 2018.