The emerging cryptocurrency market has lately received great attention for asset allocation due to its decentralization uniqueness. However, its volatility and brand new trading mode have made it challenging to devising an acceptable automatically-generating strategy. This study proposes a framework for automatic high-frequency bitcoin transactions based on a deep reinforcement learning algorithm-proximal policy optimization (PPO). The framework creatively regards the transaction process as actions, returns as awards and prices as states to align with the idea of reinforcement learning. It compares advanced machine learning-based models for static price predictions including support vector machine (SVM), multi-layer perceptron (MLP), long short-term memory (LSTM), temporal convolutional network (TCN), and Transformer by applying them to the real-time bitcoin price and the experimental results demonstrate that LSTM outperforms. Then an automatically-generating transaction strategy is constructed building on PPO with LSTM as the basis to construct the policy. Extensive empirical studies validate that the proposed method performs superiorly to various common trading strategy benchmarks for a single financial product. The approach is able to trade bitcoins in a simulated environment with synchronous data and obtains a 31.67% more return than that of the best benchmark, improving the benchmark by 12.75%. The proposed framework can earn excess returns through both the period of volatility and surge, which opens the door to research on building a single cryptocurrency trading strategy based on deep learning. Visualizations of trading the process show how the model handles high-frequency transactions to provide inspiration and demonstrate that it can be expanded to other financial products.
We analyze the predictability of the bitcoin market across prediction horizons ranging from 1 to 60 min. In doing so, we test various machine learning models and find that, while all models outperform a random classifier, recurrent neural networks and gradient boosting classifiers are especially well-suited for the examined prediction tasks. We use a comprehensive feature set, including technical, blockchain-based, sentiment-/interest-based, and asset-based features. Our results show that technical features remain most relevant for most methods, followed by selected blockchain-based and sentiment-/interest-based features. Additionally, we find that predictability increases for longer prediction horizons. Although a quantile-based long-short trading strategy generates monthly returns of up to 39% before transaction costs, it leads to negative returns after taking transaction costs into account due to the particularly short holding periods.
S. Venkata Lakshmi, N.G. Swadthi, V.Suba Shree, M. Swethamura
Bitcoin, the ruler of cryptocurrency plays an important role in blockchain technology. Every transaction stored in blockchain costs certain amount of bitcoins. The price fluctuation of bitcoins are very unstable and doesn't depend on any business or marketing strategies which builds both interest and fear in the minds of traders. By predicting the bitcoin price it is possible that users can analyze and invest in bitcoin which improves the utilization of digital money. So it is necessary to develop a prediction model with high prediction rate. This work focuses on implementing various machine learning models to predict the price of bitcoin. These machine learning models are evaluated using error percentage from which the best model for predicting bitcoin value is suggested. From the results it is evident that, among the state of the art ML algorithms, LSTM is found to be the best.
Sumit Biswas, Mohandas V. Pawar, Sachin L. Badole, Nachiket Galande · 5 authors
This rise in cryptocurrencies' value has contributed to the decentralization of authority, lowering control amongst countries. The wide price range of digital currencies highlights the need for reliable preparation for predicting the currency's price. A new model is a situation in which this paper presents a new way of forecasting digital value for money by considering several variables, such as stock market capitalization, volume, distribution, and high-end delivery. To include training results, active LSTM networks, and an overview of long-term organizations are considered. The proposed technique is used for the benchmark data sets. The results indicate the efficiency of the forecasting of digital currency.
Purpose It is crucial to find a better portfolio optimization strategy, considering the cryptocurrencies' asymmetric volatilities. Hence, this research aimed to present dynamic optimization on minimum variance (MVP), equal risk contribution (ERC) and most diversified portfolio (MDP). Design/methodology/approach This study applied dynamic covariances from multivariate GARCH(1,1) with Student’s- t -distribution. This research also constructed static optimization from the conventional MVP, ERC and MDP as comparison. Moreover, the optimization involved transaction cost and out-of-sample analysis from the rolling windows method. The sample consisted of ten significant cryptocurrencies. Findings Dynamic optimization enhanced risk-adjusted return. Moreover, dynamic MDP and ERC could win the naïve strategy (1/N) under various estimation windows, and forecast lengths when the transaction cost ranging from 10 bps to 50 bps. The researcher also used another researcher's sample as a robustness test. Findings showed that dynamic optimization (MDP and ERC) outperformed the benchmark. Practical implications Sophisticated investors may use the dynamic ERC and MDP to optimize cryptocurrencies portfolio. Originality/value To the best of the author’s knowledge, this is the first paper that studies the dynamic optimization on MVP, ERC and MDP using DCC and ADCC-GARCH with multivariate- t- distribution and rolling windows method.
Recently, the deep learning architecture has been used with an increasing rate for forecasting in financial markets. In this paper, the LSTM model is used to forecast the daily closing price direction of the BTC/USD. Both model accuracy and the profit or loss of the trades made based on the proposed model are analyzed. In addition, the effects of the MACD indicator and the input matrix dimension on forecasting accuracy are evaluated. The potential risks and actual risks encountered by the trader who trades based on the proposed model were also analyzed. The obtained results indicate that the optimization of the LSTM parameters using the Bayesian optimization model has enhanced the model’s accuracy. The results obtained from analyzing the drawdown and reward/risk resulting from the trades made based on the model show that the model enables the trader to trade with peace of mind due to the low level of actual risks and potential risks.
Paolo De Angelis, Roberto De Marchis, Mario Marino, Antonio Luciano Martire · 5 authors
Abstract In this paper, we come up with an original trading strategy on Bitcoins. The methodology we propose is profit-oriented , and it is based on buying or selling the so-called Contracts for Difference, so that the investor’s gain, assessed at a given future time t , is obtained as the difference between the predicted Bitcoin price and an apt threshold. Starting from some empirical findings, and passing through the specification of a suitable theoretical model for the Bitcoin price process, we are able to provide possible investment scenarios, thanks to the use of a Recurrent Neural Network with a Long Short-Term Memory for predicting purposes.
This paper consists of cryptocurrency prediction and analysis using different algorithms, the major cryptocurrency took into account for analysis and prediction are Bitcoin (BTC), Ethereum (ETH), Chainlink (LINK), Bitcoin Cash (BTC), XRP (XRP). Nowadays, investing in cryptocurrency has become a major deal, with huge cash flow and billions of industries which has taken over the small industry that was over the past. With this investment, it is important to understand the high & low of a particular cryptocurrency and what output will be generated with such decisions. Prediction of cryptocurrencies is tangible and requires lots of understanding regarding the flow of money on daily basis. The machine learning industry has advanced to a great extent and it would further do, this advancement has led us to a bigger problem-solving technique, that is prediction of data or analysis of trend which can be in any format. The format in this paper is a time series analysis of the daily high-low-close of digital currency. The algorithms used for such analysis is LSTM (Long Short-Term Memory) which is part of Deep Learning and further Fbprophet which is an Auto Machine Learning for prediction is used. The metric used for the analysis of the algorithm is MAE (Mean Absolute Error). The programming language used is Python, which solves the majority of use cases.
Στόχος της παρούσας μελέτης ήταν η διερεύνηση της συμπεριφοράς των τιμών πέντε κρυπτονομισμάτων BTC, LTC, ETH, XMR και XRP, των διακυμάνσεων, των πιθανών μέγιστων τιμών, των ελάχιστων τιμών και εάν υπάρχει σύνδεση, συνεργασία στη συμπεριφορά των κρυπτονομισμάτων. Για τον λόγο αυτό, οι ημερήσιες τιμές των πέντε κρυπτονομισμάτων από το 2013 έως το 2020 ανακτήθηκαν από την ιστοσελίδα www.coinmarketcap.com. Αρχικά, πραγματοποιήθηκε ανάλυση συσχέτισης με τη χρήση κυλιόμενου παραθύρου 100 ημερών κάθε ζεύγους κρυπτονομισμάτων, BTC - LTC, BTC - ETH, BTC - XMR, BTC - XRP, LTC - ETH, LTC - XMR, LTC - XRP, ETH - XMR, ETH - XRP και XMR – XRP. Επίσης, πραγματοποιήθηκε μια ανάλυση συνολοκλήρωσης με τη χρήση της δοκιμής Johansen. Η ανάλυση κυλιόμενης συσχέτισης κατέληξε στο συμπέρασμα ότι και τα πέντε κρυπτονομίσματα πριν από το έτος 2017 παρουσίασαν ένα ασταθές μοτίβο. Εν αντιθέσει, μετά το 2017, το επίπεδο συσχέτισης ήταν υψηλότερο από 0,6 και για τα πέντε κρυπτονομίσματα το οποίο αποτελεί ένδειξη σταθερού και παρόμοιου μοτίβου μεταξύ των κρυπτονομισμάτων. Τέλος, η ανάλυση δοκιμής Johansen/συνολοκλήρωσης κατέληξε στο συμπέρασμα ότι υπήρξε μια εξίσωση συνολοκλήρωσης για την περίοδο 2017 έως το 2020. Αυτό το αποτέλεσμα ήταν σύμφωνο με το αποτέλεσμα της ανάλυσης κυλιόμενου παραθύρου.
Bitcoin, the most important of all cryptocurrencies, is a currency that is not included in the central monetary system and has a digital format. With the rapid increase in the value of Bitcoin in recent years, it has attracted the attention of investors. Bitcoin, as an alternative to traditional investment tools, has sparked a lot of controversies. In this study, the Bitcoin price relationship between stock market indices and the BRICS countries belonging to Turkey is intended to be detected. In this direction of Bitcoin 01.01.2013-31.12.2019 period to test the relationship between Turkey and the BRICS countries, stock index monthly data are used. After determining the basic statistical properties of the series, cointegration and causality test was performed to determine the financial relationship. ADF (Augmented Dickey-Fuller) unit root tests and stationarity analysis are applied, and then the existence of long-term relationships between stock exchanges is explained with the Johansen cointegration test. The Vector Error Correction Model (VECM) was used to analyze whether the long-term relationship is in equilibrium. Also, short-term relationships were determined by Granger causality analysis. The analysis performed between variables were identified as a result of a long-term relationship, Russia (MOEX) and Turkey (BIST100) was found to be the cause of the stock market index of Bitcoin. It has been determined that Bitcoin is the cause of China (SHANGAI) exchange. In these exchanges, it has been observed that the change in Bitcoin prices in the short term affects investment decisions.
Mayukh Samaddar, Rishiraj Saha Roy, Sayantani De, Raja Karmakar
Machine learning is growing rapidly and has made many theoretical breakthroughs which find its application in many fields. Bitcoin is a very secure, decentralized, peer to peer currency with no third-party involvement. The price prediction of Bitcoin in the following years is a difficult task. The objective is to take a dig in the prediction of the future prices, dealing with real world data. A comparative study of the results produced by different machine learning models, along with graphs for epoch vs price, error and accuracy for each model using both linear and non-linear functions is done. We are using both neural network algorithms, such as artificial neural network (ANN), recurrent neural network (RNN) and convolutional neural network (CNN), as well as some famous supervised learning algorithms such as Random Forest (RF) and k-nearest neighbors (k-NN), to form the analysis. The time price prediction graphs and the epoch loss accuracy graphs are used for the analysis of each algorithm working on the same data and produces different results. Finally, the best suited algorithm are used for the prediction of future Bitcoin price.
Marco Ortu, Nicola Uras, Claudio Conversano, Giuseppe Destefanis · 5 authors
This work aims to analyse the predictability of price movements of\ncryptocurrencies on both hourly and daily data observed from January 2017 to\nJanuary 2021, using deep learning algorithms. For our experiments, we used\nthree sets of features: technical, trading and social media indicators,\nconsidering a restricted model of only technical indicators and an unrestricted\nmodel with technical, trading and social media indicators. We verified whether\nthe consideration of trading and social media indicators, along with the\nclassic technical variables (such as price's returns), leads to a significative\nimprovement in the prediction of cryptocurrencies price's changes. We conducted\nthe study on the two highest cryptocurrencies in volume and value (at the time\nof the study): Bitcoin and Ethereum. We implemented four different machine\nlearning algorithms typically used in time-series classification problems:\nMulti Layers Perceptron (MLP), Convolutional Neural Network (CNN), Long Short\nTerm Memory (LSTM) neural network and Attention Long Short Term Memory (ALSTM).\nWe devised the experiments using the advanced bootstrap technique to consider\nthe variance problem on test samples, which allowed us to evaluate a more\nreliable estimate of the model's performance. Furthermore, the Grid Search\ntechnique was used to find the best hyperparameters values for each implemented\nalgorithm. The study shows that, based on the hourly frequency results, the\nunrestricted model outperforms the restricted one. The addition of the trading\nindicators to the classic technical indicators improves the accuracy of Bitcoin\nand Ethereum price's changes prediction, with an increase of accuracy from a\nrange of 51-55% for the restricted model, to 67-84% for the unrestricted model.\n
The present study aims to establish the model of the cryptocurrency price trend based on financial theory using the LSTM model with multiple combinations between the window length and the predicting horizons, the random walk model is also applied with different parameter settings.
The recent surge in Bitcoin price performance has attracted significant attention from both the market and academic researchers. This paper constitutes the first principled attempt to determine market risk own-funds requirements for Bitcoin. To this end, we examine price microstructure of the USD per bitcoin, and compare to other financial variables, as a proxy toward classifying Bitcoin into the appropriate risk-class. Using the outcomes of this analysis, we classify and quantify the entailed risk from a market risk minimum capital requirements perspective. To perform the prescribed analysis, we introduce a novel methodological paradigm, which adopts bleeding-edge concepts from the field of Data Science and Machine Learning.
Iftikhar Ahmad, Muhammad Ovais Ahmad, Mohammed A. Alqarni, Abdulwahab Ali Almazroi · 5 authors
Cryptocurrencies such as Bitcoin (BTC) have seen a surge in value in the recent past and appeared as a useful investment opportunity for traders. However, their short term profitability using algorithmic trading strategies remains unanswered. In this work, we focus on the short term profitability of BTC against the euro and the yen for an eight-year period using seven trading algorithms over trading periods of length 15 and 30 days. We use the classical buy and hold (BH) as a benchmark strategy. Rather surprisingly, we found that on average, the yen is more profitable than BTC and the euro; however the answer also depends on the choice of algorithm. Reservation price algorithms result in 7.5% and 10% of average returns over 15 and 30 days respectively which is the highest for all the algorithms for the three assets. For BTC, all algorithms outperform the BH strategy. We also analyze the effect of transaction fee on the profitability of algorithms for BTC and observe that for trading period of length 15 no trading strategy is profitable for BTC. For trading period of length 30, only two strategies are profitable.
This study investigates the volatility of daily Bitcoin returns and multifractal properties of the Bitcoin market by employing the rolling window method and examines relationships between the volatility asymmetry and market efficiency. Whilst we find an inverted asymmetry in the volatility of Bitcoin, its magnitude changes over time, and recently, it has become small. This asymmetric pattern of volatility also exists in higher frequency returns. Other measurements, such as kurtosis, skewness, average, serial correlation, and multifractal degree, also change over time. Thus, we argue that properties of the Bitcoin market are mostly time dependent. We examine efficiency-related measures: the Hurst exponent, multifractal degree, and kurtosis. We find that when these measures represent that the market is more efficient, the volatility asymmetry weakens. For the recent Bitcoin market, both efficiency-related measures and the volatility asymmetry prove that the market becomes more efficient.
Since its founding in 2008, Bitcoin (financial code: BTC) has emerged as a digital currency in market cap and continues to attract investors and policymakers' attention. In recent years, BTC has high price volatility, a substantial increase in 2016, followed by a significant decline in 2018. Unlike stock markets, BTC is open for 24x7 dan has no closing period. It means everyone can trade it for any time. However, this flexibility carries investment risk. This research attempts to forecast BTC's price by considering the blockchain's information to minimize the risk. We employ Long-Short Term Memory (LSTM), the artificial Recurrent Neural Network (RNN) architecture. Its model can avoid long-term problems. The data used is BTC's price and blockchain information data from August 4, 2018, to January 21, 2020. The model with 20 neurons and 500 epochs has the smallest MSE value. Then a prediction has an accuracy rate of 91.07%.
P Nithyakani, Rijo Jackson Tom, Piyush Gupta, A. Shanthini · 6 authors
Machine Learning and Artificial Intelligence based money exchanging have p ulled in enthusiasm in the recent years with the introduction of Bitcoins. The cost of Bitcoins has increased in a large scale and it is fairly difficult to predict the future cost per Bitcoin. In this study, we utilize a machine learning and deep learning model to analyze the digital currency market to predict the cost of Bitcoin per day. We dissect everyday information for 1,691 cryptographic forms of money for the period between November 2017 and April 2019. The study shows that straightforward exchanging procedures assisted by best in class AI algorithms have met the standard benchmarks. Our outcomes also show that non-inconsequential, basic algorithmic instruments can help in envision of momentary development of the cryptographic money. The proposed system uses a Bi- directional LSTM for forecasting the bitcoin prices. The proposed model was able to trace the test dataset with Mean Absolute Percentage Error of 13%. The model is helpful for the user to take decision on investing in Bitcoins.
Ioannis E. Livieris, Niki Kiriakidou, Stavros Stavroyiannis, Panagiotis Pintelas
Nowadays, cryptocurrencies are established and widely recognized as an alternative exchange currency method. They have infiltrated most financial transactions and as a result cryptocurrency trade is generally considered one of the most popular and promising types of profitable investments. Nevertheless, this constantly increasing financial market is characterized by significant volatility and strong price fluctuations over a short-time period therefore, the development of an accurate and reliable forecasting model is considered essential for portfolio management and optimization. In this research, we propose a multiple-input deep neural network model for the prediction of cryptocurrency price and movement. The proposed forecasting model utilizes as inputs different cryptocurrency data and handles them independently in order to exploit useful information from each cryptocurrency separately. An extensive empirical study was performed using three consecutive years of cryptocurrency data from three cryptocurrencies with the highest market capitalization i.e., Bitcoin (BTC), Etherium (ETH), and Ripple (XRP). The detailed experimental analysis revealed that the proposed model has the ability to efficiently exploit mixed cryptocurrency data, reduces overfitting and decreases the computational cost in comparison with traditional fully-connected deep neural networks.