Umair Khalid, Syeda Asnia Arif, Muhammad Umair Khan
Purpose: The research aims to analyze the log-returns of Bitcoin exchange rates against the US Dollar and Chinese Yuan by applying parametric distributions for understanding behavior and suggesting a best-fitted distribution.
 Design/Methodology/Approach: Methodology involves the volatility risk analysis using the GARCH model for analyzing the behavior of Bitcoin Exchange rates of USD and CNY.
 Findings: The results showed that the Weibull distribution gives the best fit to both of the currencies’ exchange rates
 Implications/Originality/Value: The exchange rates of Bitcoin analyzed in this study in midst of myriad other cryptocurrencies using parametric distributions thereby encouraging the application of nonparametric and semiparametric distributions in similar scenarios. The application of this study would enable not only individual investors but also institutional investors and venture capital firms to stay informed of alternating trends and movements through distributions for predicting future returns.
Jacques Fleischer, Gregor von Laszewski, Carlos Theran, Yohn Jairo Parra Bautista
Digitization is changing our world, creating innovative finance channels and emerging technology such as cryptocurrencies, which are applications of blockchain technology. However, cryptocurrency price volatility is one of this technology’s main trade-offs. In this paper, we explore a time series analysis using deep learning to study the volatility and to understand this behavior. We apply a long short-term memory model to learn the patterns within cryptocurrency close prices and to predict future prices. The proposed model learns from the close values. The performance of this model is evaluated using the root-mean-squared error and by comparing it to an ARIMA model.
Objective: The purpose of this paper is to demonstrate the effectiveness of the nonparametric GARCH model for the prediction of future Bitcoin prices. Methodology: The parametric GARCH models to characterize the volatility of Bitcoin returns are widely used in the empirical literature. Alternatively, we consider a non-parametric approach to model and forecast the volatility of Bitcoin returns. Results: We show that the volatility forecast of the nonparametric GARCH model yields superior performance compared to an extended class of parametric GARCH models. Originality / relevance: The improved accuracy of forecasting the volatility of Bitcoin returns based on the nonparametric GARCH model suggests that this method offers an attractive and viable alternative to commonly used GARCH parametric models.
It has been emphasized in many studies that the developments in the crypto money markets have a serious impact on the world stock markets. Due to these effects, the fluctuations in the world stock markets have increased, and it has become necessary for investors to follow these markets more closely and determine their strategies according to these developments. In this study, it was examined whether the developments in the crypto money market have an effect on Borsa Istanbul (BIST) indices. For this purpose, data of the three most popular cryptocurrencies Bitcoin, Ethereum and Ripple were used, and their spillover effects on BIST100, BIST30 and banking (XBANK) indices were investigated. Oil prices (WTI) and fear index (VIX) variables were also used as control variables in the study. The findings obtained from the analyses in our study carried out for the period 01/01/2014-31/12/2021 showed that there is a positive spillover effect from the crypto money markets to the indices we examined. While oil prices were found to be statistically significant in all models among the control variables, different results were obtained on the effect of the fear index. The findings show that it is imperative for stock market investors to closely monitor the developments in the crypto money market in addition to track various economic variables, in their investment decisions.
Due to its growing popularity and commercial acceptance, cryptocurrency is playing an increasingly essential role in altering the financial system. While many people are investing in cryptocurrency, the dynamic characteristics and predictability of cryptocurrency are still largely unknown, putting investments at risk. In this paper, we attempt to anticipate the Bitcoin price by taking into account a variety of factors that influence its value with the highest possible accuracy using (LSTM) Recurrent Neural Network. The data we use in this work includes updated daily records of many aspects of Bitcoin pricing over a five-year period. Since the cryptocurrency (Bitcoin) data is so volatile, we implement an effective pre-processing of the data in order to have a better prediction result. With this solution, we gain accuracy of 95.7% and RMSE of 0.05. Furthermore, we compare this work with other existing methods based on performance and accuracy. This comparison demonstrates that utilizing LSTM with adequate hyperparameter tweaking is one of the most efficient ways for cryptocurrency price prediction.
Muhammad Nazrin Farhan Nasarudin, Ahmad Ihsan Mohd Yassin, Megat Syahirul Amin Megat Ali, Mohd Khairil Adzhar Mahmood · 6 authors
Bitcoin is a decentralized digital currency that enables people to exchange value without requiring a third-party intermediary. Due to its many advantages, it has received much interest from institutional and individual investors. Despite its meteoric increase, the price of Bitcoin extremely volatile asset class as it purely relies on supply and demand. This presents an interesting opportunity to create a forecasting model. However, many research papers in this area does not analyse the residuals as part of the forecasting resulting in potentially biased models. In this paper, we demonstrate System Identification (SI) residual analysis techniques to the analysis of our forecasting model. The Multi-Layer Perceptron (MLP) Nonlinear Autoregressive with Exogeneous Inputs (NARX) uses historical price data and several technical indicators to predict the future price movements of Bitcoin. The Particle Swarm Optimization (PSO) algorithm was used to find optimal parameters for the model. The model was able to predict one day ahead price in the prediction test. The model has successfully captured the dynamics of the data through the tests performed on residuals. It is also proving the randomness of residuals, albeit some minor violations.
This study examines the weak form of the efficient market hypothesis for Bitcoin using a feedforward neural network. Due to the increasing popularity of cryptocurrencies in recent years, the question has arisen, as to whether market inefficiencies could be exploited in Bitcoin. Several studies we refer to here discuss this topic in the context of Bitcoin using either statistical tests or machine learning methods, mostly relying exclusively on data from Bitcoin itself. Results regarding market efficiency vary from study to study. In this study, however, the focus is on applying various asset-related input features in a neural network. The aim is to investigate whether the prediction accuracy improves when adding equity stock indices (S&P 500, Russell 2000), currencies (EURUSD), 10 Year US Treasury Note Yield as well as Gold&Silver producers index (XAU), in addition to using Bitcoin returns as input feature. As expected, the results show that more features lead to higher training performance from 54.6% prediction accuracy with one feature to 61% with six features. On the test set, we observe that with our neural network methodology, adding additional asset classes, no increase in prediction accuracy is achieved. One feature set is able to partially outperform a buy-and-hold strategy, but the performance drops again as soon as another feature is added. This leads us to the partial conclusion that weak market inefficiencies for Bitcoin cannot be detected using neural networks and the given asset classes as input. Therefore, based on this study, we find evidence that the Bitcoin market is efficient in the sense of the efficient market hypothesis during the sample period. We encourage further research in this area, as much depends on the sample period chosen, the input features, the model architecture, and the hyperparameters.
T Sanjana, V Mahalakshmi, M. Preethi Darshini, N Hemanth. · 5 authors
Abstract: Cryptocurrencies are now deeply rooted and widely used as a form of money. Almost all financial instruments are affected, but Bitcoin trading is seen as one of the most recognizable potentials. Since this ever-growing short-term financial market is characterized by high volatility and significant fluctuations in value in a short period of time, the promotion of more accurate and reliable valuation models is seen as an important improvement feature in any firm. Wallet. This study describes a correlated deep learning model that can digitally predict the value and evolution of money. Use of CNN, CNN-LSTM and ARIMA algorithms. Almost any precise feature can be predicted. Keywords: Deep learning; CNN, CNN- LSTM; ARIMA Algorithm; Forecasting style;
R Marriammal, Reni Hena Helen R, M Rubika, T Sowbhagya
Bitcoin, the king of cryptocurrencies, is central to blockchain technology. A fixed amount of bitcoins is required for each transaction stored in the blockchain. The price of bitcoins fluctuates wildly and is unaffected by any company or marketing techniques, creating both curiosity and terror in the minds of traders. It is possible for consumers to study and invest in bitcoin by anticipating the bitcoin price, which promotes the use of digital money. As a result, a high-prediction-rate prediction model is required. The goal of this project is to employ a variety of machine learning models to predict the price of bitcoin. The best model for predicting bitcoin value is given based on the error percentage of these machine learning algorithms.
In this research, we provided an answer to a very important trading question, what is the optimal number of technical tools in order to achieve the best trading results for both swing trade that uses daily bars and intraday trade that uses minutes bars? We designed Machine Learning (ML) systems that can trade four major cryptocurrencies: Bitcoin, Ethereum, BNB, and Solana. We found that more indicators do not necessarily mean better trading performance. Swing traders that use daily bars should trade Bitcoin and Solana using Ichimoku Cloud (IC) plus Moving Average Convergence Divergence (MACD), Ethereum with IC plus Chaikin Money Flow (CMF), and BNB with IC alone. With regard to intraday trading, we documented that different cryptocurrencies should be trading using different time frames. These results emphasize that the optimal number of indicators that are used to trade daily bars is one or, at maximum, two. The Multi-Layer (MUL) system that consists of all three examined technical indicators failed to improve the trading results for both days (swing) and intraday trades. The main implication of this study for traders is that more indicators does not necessarily improve trades performances.
Bitcoin is a type of Internet currency that is both a digital asset and a payment method. It enables for anonymous payment from one person to another, making it a popular payment mechanism for online illegal activity. Due to its recent price increase, Bitcoin has gotten a lot of attention from the media and the general public. The goal of this research is to discover the Bitcoin price's predictable price direction. Machine learning models are likely to provide us with the information we require to understand the future of cryptocurrency. It won't tell us what will happen in the future, but it might show us the overall trend and direction in which prices are likely to move. The proposed methodology aims to create a machine learning model that uses data to learn about the patterns in the dataset and then uses a machine learning algorithm to forecast the bitcoin price.
As a new type of electronic currency, bitcoin is more and more recognized and sought after by people, but its price fluctuation is more intense, the market has certain risks, and the price is difficult to be accurately predicted. The main purpose of this study is to use a deep learning integration method (SDAE-B) to predict the price of bitcoin. This method combines two technologies: one is an advanced deep neural network model, which is called stacking denoising autoencoders (SDAE). The SDAE method is used to simulate the nonlinear complex relationship between the bitcoin price and its influencing factors. The other is a powerful integration method called bootstrap aggregation (Bagging), which generates multiple datasets for training a set of basic models (SDAES). In the empirical study, this study compares the price sequence of bitcoin and selects the block size, hash rate, mining difficulty, number of transactions, market capitalization, Baidu and Google search volume, gold price, dollar index, and relevant major events as exogenous variables uses SDAE-B method to compare the price of bitcoin for prediction and uses the traditional machine learning method LSSVM and BP to compare the price of bitcoin for prediction. The prediction results are as follows: the MAPE of the SDAE-B prediction price is 0.016, the RMSE is 131.643, and the DA is 0.817. Compared with the other two methods, it has higher accuracy and lower error, and can well track the randomness and nonlinear characteristics of bitcoin price.
In recent years, Bitcoin cryptocurrency has become a growing trend in the world. For this reason, researchers from many fields are examining various artificial intelligence models to predict Bitcoin rates. In particular, Deep Learning algorithms have been shown to outperform traditional models in predicting cryptocurrency rates. However, very few studies have examined the effect of parameters used in deep learning algorithms on the algorithm. Optimization and loss functions are very important, which affect the algorithm's ability to make a successful prediction. In this study, Long-Short Term Memory, a deep learning algorithm, is used to predict daily Bitcoin prices and the effect of optimization/loss functions on the accuracy rate is evaluated. Experimental results showed that the Long-Short Term Memory model made the best predictions as a result of working with the Adam optimization function and the Mean Square Error loss function.
Portfolio optimization is one of the most complex problems in the financial field, and technical analysis is a popular tool to find an optimal solution that maximizes the yields. This paper establishes a portfolio optimization model consisting of a weighted unidirectional dual-layer LSTM model and an SMA-slope strategy. The weighted unidirectional dual-layer LSTM model is developed to predict the daily prices of gold/Bitcoin, which addresses the traditional problem of prediction lag. Based on the predicted prices and comparison of two representative investment strategies, simple moving average (SMA) and Bollinger bands (BB), this paper adopts a new investment strategy, SMA-slope strategy, which introduces the concept of k-slope to measure the daily ups and downs of gold/Bitcoin. As two typical financial products, gold and Bitcoin are opposite in terms of their characteristics, which may represent many existing financial products in investors’ portfolios. With a principle of $1000, this paper conducts a five-year simulation of gold and Bitcoin trading from 11 September 2016 to 10 September 2021. To compensate for the SMA and BB that may miss buying and selling points, 4 different parameters’ values in the k-slope are obtained through particle swarm optimization simulation. Also, the simulation results imply that the proposed portfolio optimization model contributes to helping investors make investment decisions with high profitability.
The financial risk of investing in Bitcoin is increasing, and everyone partic-ipating in the transaction is aware of it. The rise and fall of bitcoin’s value is difficult to predict, and the system is fraught with uncertainty. As a result, this study proposed to use the «Deep learning» technique for predicting fi-nancial risk associated with bitcoin investment, that is linked to its «weighted price» on the bitcoin market’s volatility. The dataset used included Bitcoin historical data, which was acquired «at one-minute intervals» from selected exchanges of January 2012 through December 2020. The deep learning lin-ear-SVM-based technique was used to obtain an advantage in handling the high-dimensional challenges related with bitcoin-based transaction transac-tions large data volume. Four variables («High», «Low», «Close», and «Volume (BTC)».) are conceptualized to predict weighted price, in order to indi-cate if there is a propensity of financial risk over the effect of their interaction. The results of the experimental investigation show that the fi-nancial risk associated with bitcoin investing is accurately predicted. This has helped to discover engagements and disengagements with doubts linked with bitcoin investment transactions, resulting in increased confidence and trust in the system as well as the elimination of financial risk. Our model had a significantly greater prediction accuracy, demonstrating the utility of deep learning systems in detecting financial problems related to digital currency.
Abstract In this study, we predict Bitcoin price trends using the back propagation neural network (BPNN), autoregressive integrated moving average (ARIMA), and generalized autoregressive conditional heteroscedasticity (GARCH) models. Based on principal component analysis (PCA), we extract two new input components for BPNN from Bitcoin’s three-day closing prices, MA5, MA20, daily trading volume, Ether price, and Ripple price. The training set covers the period between September 1, 2015 and March 31, 2020, and the forecasting set covers the period between April 1, 2020 and June 30, 2020. Empirical results reveal (1) the predictive ability of BPNN over that of the ARIMA models; (2) BPNN with two hidden layers is able to predict price trends more precisely than that with only one hidden layer; (3) in terms of time series models, the ARIMA-GARCH family of models demonstrates better predictive performance than ARIMA models; and (4) among the ARIMAGARCH family of models, the ARIMA-EGARCH model is proven to produce the best predictive results on price, and the ARIMA-GARCH model predicts more accurately than the ARIMA-GJR-GARCH model. Specifically, our findings provide a reference on Bitcoin for market participants. JEL classification numbers: C32, C45, C53, G17. Keywords: Bitcoin, Back propagation neural network, Autoregressive integrated moving average, Generalized autoregressive conditional heteroscedasticity, Principal component analysis.
Oluwatobi A. Adekunle, Adedeji Daniel Gbadebo, Joseph Akande
Bitcoin price exhibits patterns predictable on its historical pasts. We adopt ARIMA(auto), ARIMA(fix)models and the Holt-Winters filter (HWF) with trend plus additive seasonal HWF (𝛾[0,1]), and no seasonality HWF (𝛾[False]) to forecast the price of Bitcoin under three datasets–Actual (observed), Polynomial (fitted) and STL-Trend (fitted). We apply daily time-series from 1/09/2014–28/12/2020,and establish 18 models to forecast the price of Bitcoin. The results show that HWF (𝛾[0,1]) with lower limit fitted on STL-Trend provides the best prediction on the first training-sample, while ARIMA(fix) fitted on actual data outperform in the second training-set with the smallest Mean-Absolute-Error (MAE). The training-set forecast performance of the ARIMA(fix) for the actual function provides better performance with the least MAE. The HWF is appropriate for prediction of the daily Bitcoin price with the generalise STL-Trend function, but ARIMA(fix) is more accurate for the actual series.
In recent years, popularity and use of cryptocurrencies has been rising along with their prices and Ethereum is the second most famous cryptocurrency after Bitcoin. Cryptocurrencies are based on blockchain, which is a distributed and empowered technology that has the power to transform any banking systems. It has become an attractive investment for traders as well as individuals looking to invest. The price of Ethereum varies and is controlled by different factors, such as the crypto market in which it is sold, supply and demand. Ethereum is so valuable because it could be used as cash, we could also pay a portion or part of Ethereum to someone in exchange and it is easily guaranteed by the blockchain. Unlike stocks, Ethereum price is much more variable, as it has a trading time of 24-hours a day without any close time. The paper compares the results of three different models, namely Recurrent Neural Networks (RNNs), Long Short-Term Memory (LSTMs) and Bi-directional Long Short-Term Memory (Bi-LSTMs). The dataset consists of the closing price for the last 2000 days that is used to predict both short-term (30 days) and long-term (90 days) Ethereum prices. These prices are being fetched from an API which is in JSON format and are updated every day.
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.
In this paper, we give a simple but very general definition of 'price stability' for a class of markets. This class of markets includes the popular constant function market makers (CFMMs) such as Uniswap, Curve, and Balancer, used extensively in decentralized finance (DeFi), which now have daily trading volumes in the billions of dollars. We show that our definition of price stability is deeply connected to the curvature of the trading function used in the CFMM, making the folk intuition that "flatter CFMMs are more price stable" more concrete. We also show that this definition gives sufficient conditions for the profitability of liquidity providers, and, similar to the classical market microstructure literature, gives bounds on the edge of informed traders and bounds on the losses of liquidity providers. We also show how these bounds help explain some of the behaviors observed in decentralized finance in the second half of 2020, including the rise of 'yield farming ' and 'vampire attacks.'
Bitcoin, with its ever-growing popularity, has demonstrated extreme price volatility since its origin. This volatility, together with its decentralised nature, make Bitcoin highly subjective to speculative trading as compared to more traditional assets. In this paper, we propose a multimodal model for predicting extreme price fluctuations. This model takes as input a variety of correlated assets, technical indicators, as well as Twitter content. In an in-depth study, we explore whether social media discussions from the general public on Bitcoin have predictive power for extreme price movements. A dataset of 5,000 tweets per day containing the keyword `Bitcoin' was collected from 2015 to 2021. This dataset, called PreBit, is made available online. In our hybrid model, we use sentence-level FinBERT embeddings, pretrained on financial lexicons, so as to capture the full contents of the tweets and feed it to the model in an understandable way. By combining these embeddings with a Convolutional Neural Network, we built a predictive model for significant market movements. The final multimodal ensemble model includes this NLP model together with a model based on candlestick data, technical indicators and correlated asset prices. In an ablation study, we explore the contribution of the individual modalities. Finally, we propose and backtest a trading strategy based on the predictions of our models with varying prediction threshold and show that it can used to build a profitable trading strategy with a reduced risk over a `hold' or moving average strategy.
The fame of cryptocurrencies soars in 2017 because of a few consecutive months of the exponential development of their market capitalization. Even though machine learning has been fruitful in anticipating stock market costs through a large group of various time series models, its application in foreseeing cryptocurrency costs has been very prohibitive. The reason behind this is clear as the costs of cryptocurrencies rely upon a ton of factors like technological progress, internal competition, pressure on the markets to deliver, economic problems, security issues, political factors and so on Their high volatility prompts the incredible capability of high benefit if savvy designing systems are taken. Sadly, because of their absence of lists, cryptocurrencies are somewhat capricious contrasted with traditional financial predictions like stock market predictions. The proposed paper describes how Cryptocurrency works, its use, legal prospect, security and what is the technology behind it
Virtual currencies or cryptocurrencies are based on Blockchain technology, also known as distributed ledger technology. As of March 2022, there are already over 10k virtual coins, their number being continuously growing since 2013. This paper aims to extract the public sentiment expressed towards the cryptocurrency market and Blockchain technology, two topics widely debated in the last decade. Our research was based on the use of Twitter data, collected with the help of an API in the RStudio environment.