Prabakaran Raghavendran, Tharmalingam Gunasekar, M. L. Suresh, Sumaiya Banu S. S · 5 authors
This research investigates the growing demand from investors, traders, and government entities for accurate forecasting of ethereum prices. As one of the pioneering crypto currencies, ethereum has gained traction, in part, due to its promise of freedom from centralized government control. Our methodology employs cutting-edge artificial neural networks (ANN) to forecast ethereum prices. The key advantage lies in the adaptability of these networks to capture the dynamic and often unpredictable patterns inherent in crypto currency markets. To enhance forecast accuracy and timeliness, we explore various lag configurations within intervals, specific time demonstrating the model's effectiveness through outcomes. Our the predictions resulting undergo thorough validation, with a focus on assessing the root mean square error as a critical performance metric. Consistently, the outcomes from our proposed artificial neural network model outperform traditional prediction methods, providing valuable insights for individuals, industries, and governmental bodies navigating the complex landscape of crypto currencies
In the past years, the widespread diffusion of Artificial Intelligence (AI) in the finance domain transformed different services, with particular attention to the stock market. Although different AI-based approaches have been proposed for stock forecasting, they are focused on news content or sentiment without considering fundamental features and vice versa. In turn, other approaches rely on handmade rules or ones based on technical indicators for providing advice without considering contextual information that can strongly affect the stock market. In this paper, we propose an Advisor Neural Network framework using Long Short-Term Memory (LSTM)-based Informative Stock Analysis for Daily investment Advice. Specifically, the forecasting unit relies on a LSTM-based model, which combines technical indicators, contextual information, and financial data for stock forecasting. Successively, the advice unit provides next-day advice based on predicted information in conjunction with the proposed Heuristic Stocks Selection algorithm. This framework has been evaluated on the Stock and Cryptocurrencies markets, considering a subset of 417 stocks and 67 cryptocurrencies over three years, respectively. We compared the proposed framework with several state-of-the-art approaches, showing how it outperforms the baseline in both markets. Furthermore, we achieved a financial gain greater than 41%, despite the downward trend of the NASDAQ market in the quarter under review, and we obtained a 39.38% return on investment for the Cryptocurrencies market.
Abstract Although the 2022 cryptocurrency market crash prompted despair among investors, the rallying cry, “wagmi” (We’re all gonna make it.) emerged among cryptocurrency enthusiasts in the aftermath. Did cryptocurrency enthusiasts respond to this crash differently compared to traditional investors? Using natural language processing techniques applied to Twitter data, this study employed a difference-in-differences method to determine whether the cryptocurrency market crash had a differential effect on investor sentiment toward cryptocurrency enthusiasts relative to more traditional investors. The results indicate that the crash affected investor sentiment among cryptocurrency enthusiastic investors differently from traditional investors. In particular, cryptocurrency enthusiasts’ tweets became more neutral and, surprisingly, less negative. This result appears to be primarily driven by a deliberate, collectivist effort to promote positivity within the cryptocurrency community (“wagmi”). Considering the more nuanced emotional content of tweets, it appears that cryptocurrency enthusiasts expressed less joy and surprise in the aftermath of the cryptocurrency crash than traditional investors. Moreover, cryptocurrency enthusiasts tweeted more frequently after the cryptocurrency crash, with a relative increase in tweet frequency of approximately one tweet per day. An analysis of the specific textual content of tweets provides evidence of herding behavior among cryptocurrency enthusiasts.
In this study, we investigate how adding Bitcoin can influence the investment portfolios. For this purpose, we consider a portfolio including Bitcoin and the five major sector indices of the Tehran Stock Exchange (TSE). At first, the asset returns are predicted through an estimation model based on higher moments. In the second step, the properties of Bitcoin in the face of other assets in a portfolio are studied by the asymmetric dynamic conditional correlation (ADCC) model. Then, the optimal weights in the portfolios are estimated. Accordingly, we used four portfolio optimization models with different objective functions, including a hybrid function of the higher moments, predicted risk from the ADCC model, and maximizing Sharpe and Sortino ratios. The out-of-sample results showed the relative efficiency of the proposed model in predicting the asset returns in Tehran Stock Exchange. In addition, the results of the ADCC model showed that Bitcoin plays a risk-hedging role for the pharmaceutical and banking sectors in TSE. We also know Bitcoin as a safe haven for the banking, petrochemical, metals, automobile, and pharmaceutical sectors. The results of portfolio selection also prove the effectiveness of adding Bitcoin with a maximum weight of 10% in the investment portfolios.
ABSTRACT Bitcoin, being one of the most triumphant cryptocurrencies, is gaining increasing popularity online and is being used in a variety of transactions. Recently, research on Bitcoin price predictions is receiving more attention, and researchers have investigated the various state‐of‐the‐art machine learning (ML) and deep learning (DL) models to predict Bitcoin price. However, despite these models providing promising predictions, they consistently exhibit uncertainty, which cannot be adequately quantified by classical ML models alone. Motivated by the enormous success of applying Bayesian approaches in several disciplines of ML and DL, this study aims to use Bayesian methods alongside Long Short‐Term Memory (LSTM) to predict the closing Bitcoin price and consequently measure the uncertainty of the prediction model. Specifically, we adopted the Monte Carlo dropout (MC‐Dropout) method with the Bayesian LSTM model to quantify the epistemic uncertainty of the model's predictions and provided confidence intervals for the predicted outputs. Experimental results showed that the proposed model is efficient and outperforms other state‐of‐the‐art models in terms of root mean square error (RMSE), mean absolute error (MAE) and R 2 . Thus, we believe that these models may assist the investors and traders in making critical decisions based on short‐term predictions of Bitcoin price. This study illustrates the potential benefits of utilizing Bayesian DL approaches in time series analysis to improve data prediction accuracy and reliability.
Domícián Máté, Hassan Raza, Ishtiaq Ahmad, Sándor J. Kovács
Cryptocurrencies are quickly becoming a key tool in investment decisions. The volatile nature of bitcoin prices has spurred the demand for robust predictive models. The primary objective of this study is to evaluate and compare the effectiveness of different machine learning models with the combination of technical indicators in predicting bitcoin prices. The study used 27 critical technical indicators to evaluate four machine learning techniques, namely Artificial Neural Network (ANN), a Hybrid Convolutional Neural Network and Long Short-Term Memory (CNN-LSTM), Support Vector Machine (SVM), and Random Forest. The results showed that ANN and SVM achieve a significant prediction accuracy of 81% and 82%, respectively, which is higher than the results of traditional models such as standard ARIMA. In practical applications, these methods often improve prediction accuracy by 20-30% over traditional models. The novelty of the analysis lies in the use of temporal and spatial trends via momentum, ROC, and %K features, making for a holistic approach to cryptocurrency market forecasting. This study underscores the critical importance of specific technical indicators and the imperative role of data mining in revolutionizing cryptocurrency market navigation. The research results highlight opportunities to improve investment strategies and risk management policies in the bitcoin market using machine learning models, making the latter valuable to investors and financial experts.
Juan Guillermo Lazo Lazo, Diego Alejandro Ruíz Cárdenas, Sebastián Esquives Bravo
El mercado de las criptomonedas es conocido por su constante evolución, alta volatilidad, incertidumbre y frecuentes fluctuaciones de precios. Esto expone a los inversores a riesgos considerables, pero también ofrece grandes oportunidades de beneficios. Ante este escenario, los inversores buscan estrategias que maximicen las ganancias, minimicen los riesgos y reduzcan los costes operativos. La complejidad de estas decisiones hace muy atractivo el uso de técnicas de aprendizaje automático. Estas técnicas exploran grandes volúmenes de datos para desarrollar modelos predictivos, basados en la identificación de patrones, que pueden ayudar en la toma de decisiones. Este estudio propone un modelo de estrategia de inversión basado en inteligencia computacional y datos financieros. El modelo propuesto pretende realizar inversiones de tres días de duración, con el objetivo de maximizar los beneficios y mitigar los riesgos asociados a la volatilidad de los precios, especialmente durante los periodos de caídas bruscas y rápidas que son habituales en el mercado de criptodivisas. Para ello, se utilizaron redes neuronales artificiales y datos históricos de precios; se llevó a cabo el preprocesamiento de datos y el análisis de índices estadísticos. Los resultados obtenidos fueron prometedores, demostrando la capacidad de la estrategia propuesta para generar beneficios significativos durante el periodo de prueba. Los beneficios fueron mayores y con menor exposición al riesgo en comparación con los resultados obtenidos por la estrategia de mercado de comprar y mantener, lo que pone de relieve el potencial de los enfoques basados en el aprendizaje automático para optimizar las inversiones en el mercado de criptomonedas.
Currently, there are relatively few studies on Bitcoin price prediction, and accurate prediction of Bitcoin price is the focus of economic policy makers as well as investors. In this paper, we combine the methods of traditional time series model (ARIMA) and deep learning model (LSTM) to analyze the price prediction of bitcoin historical data. The empirical results show that the single LSTM model has the best prediction effect in the three prediction intervals of short-term, medium-term, and long-term. The ARIMA-LSTM hybrid model has a little improvement compared to the ARIMA model, which is chosen in this paper under the assumption that the relationship between the linear and nonlinear parts is additive, which will make the prediction effect worse. This problem will be avoided if the nonlinear combination is used for modeling, making the prediction better.
An Internet of Things (IoT)-enabled blockchain helps to ensure quick and efficient immutable transactions. Low-power IoT integration with the Bitcoin network has created new opportunities and difficulties for blockchain transactions. Utilising data gathered from IoT-enabled devices, this study investigates the application of ML regression models to analyse and forecast Bitcoin transaction patterns. Several ML regression algorithms, including Lasso Regression, Gradient Boosting, Extreme Boosting, Extra Tree, and Random Forest Regression, are employed to build predictive models. These models are trained using historical Bitcoin transaction data to capture intricate relationships between various transaction parameters. To ensure model robustness and generalisation, cross-validation techniques and hyperparameter tuning are also applied. The empirical results show that the Bitcoin cost prediction of blockchain transactions in terms of time series. Additionally, it highlights the possibility of fusing block- chain analytics with IoT data streams, illuminating how new technologies might work together to enhance financial institutions.
Ebenezer Fiifi Emire Atta Mills, Yuexin Liao, Zihui Deng
Due to the recent fluctuations in cryptocurrency prices, Ethereum has gained recognition as an investment asset. Given its volatile nature, there is a significant demand for accurate predictions to guide investment choices. This paper examines the most influential features of the daily price trends of Ethereum using a novel approach that combines the Random Forest classifier and the ReliefF method. Integrating the Adaptive Neuro-Fuzzy Inference System (ANFIS) and Short-Time Fourier Transform (STFT) resulted in high accuracy and performance metrics for Ethereum price trend predictions. This method stands out from prior research, primarily based on time series analysis, by enhancing pattern recognition across time and frequency domains. This adaptability leads to better prediction capabilities with accuracy reaching 76.56% in a highly chaotic market such as cryptocurrency. The STFT's ability to reveal cyclical trends in Ethereum's price provides valuable insights for the ANFIS model, leading to more precise predictions and addressing a notable gap in cryptocurrency research. Hence, compared to models in literature such as Gradient Boosting, Long Short-Term Memory, Random Forest, and Extreme Gradient Boosting, the proposed model adapts to complex data patterns and captures intricate non-linear relationships, making it well-suited for cryptocurrency prediction.
Financial markets exhibit complex, nonlinear dynamic behavior that is often challenging to interpret using traditional linear analysis methods. Risk assessment and, accordingly, portfolio management are the most important tasks for investors in order to avoid big losses. Gold trading is considered a safe haven, but it shows slow growth. In contrast, bitcoin or cryptocurrency trading is being proliferated. Many established financial institutions, like PayPal and Fidelity, are making it easier than ever for individuals to buy and sell Bitcoin. While Bitcoin trading can be potentially lucrative, it’s crucial to understand the significant risks involved before diving The volatile nature of Bitcoin can be emotionally stressful, leading to anxiety and poor decision-making. High-risk activity like stock trading requires a strong understanding of the market, particularly in high-risk assets like Bitcoin and gold, which demands a deep comprehension of market dynamics and the ability to tolerate volatility. Through this study, we delve into the application of machine learning algorithms to comprehend correlations, causality, and predict volatility in stock trading. Specifically, we focus on analyzing the trading behavior of cryptocurrency (bitcoin) vs. physical assets (gold) in terms of their volatility and comparing them in correlation with the stock index of crude-oil, S&P 500, and the dollar. This structured study aims to shed light on the effectiveness of machine learning algorithms in understanding and predicting volatility in stock trading, with a specific focus on Bitcoin and gold as high-risk assets.
Decentralized Finance (DeFi) is increasingly studied and adopted for its potential to provide accessible and transparent financial services. Analyzing how investors use DeFi is important for reaching a better understanding of their usage and for regulation purposes. However, analyzing DeFi transactions is challenging due to often incomplete or inaccurate labeled data. This paper presents a method to extract ego network motifs from the token transfer network, capturing the transfer of tokens between users and smart contracts. Our results demonstrate that smart contract methods performing specific DeFi operations can be efficiently identified by analyzing these motifs while providing insights into account activities.
This paper delves into the intricate relationship between liquidity indicators and the price dynamics of Bitcoin, a prominent cryptocurrency.Liquidity, a fundamental aspect of financial markets, profoundly influences market stability and efficiency.Leveraging statistical analysis and AI modeling techniques, my study explores various liquidity metrics-including trading volume, bid-ask spread, volatility, number of transactions, and bid and ask sums as separate indicators-to assess their impact on the price of Bitcoin.The findings offer valuable insights into the factors driving Bitcoin price movements and shed light on the role of liquidity in cryptocurrency markets.Through correlation analysis as well as three different machine learning models -random forests, XGBoost, and linear regression -, my study evaluates the significance of individual liquidity factors and their relationships with Bitcoin prices.The best performing model was the random forest regressor and XGBoost where I identified that the volatility was the feature that was the most informative of the model's performance.My research contributes to advancing our understanding of liquidity and price discovery in cryptocurrency markets and underscores the need for future studies to explore alternative factors and mechanisms shaping cryptocurrency prices.By embracing the findings and continuously refining analytical approaches, researchers can navigate the evolving landscape of cryptocurrency trading, ultimately enhancing market efficiency and informing regulatory decisions.
Quang Phung Duy, Oanh Nguyen Thi, Phuong Hao Le Thi, Hai Duong Pham Hoang · 6 authors
Purpose The goal of the study is to offer important insights into the dynamics of the cryptocurrency market by analyzing pricing data for Bitcoin. Using quantitative analytic methods, the study makes use of a Generalized Autoregressive Conditional Heteroskedasticity (GARCH) model and an Autoregressive Integrated Moving Average (ARIMA). The study looks at how predictable Bitcoin price swings and market volatility will be between 2021 and 2023. Design/methodology/approach The data used in this study are the daily closing prices of Bitcoin from Jan 17th, 2021 to Dec 17th, 2023, which corresponds to a total of 1065 observations. The estimation process is run using 3 years of data (2021–2023), while the remaining (Jan 1st 2024 to Jan 17th 2024) is used for forecasting. The ARIMA-GARCH method is a robust framework for forecasting time series data with non-seasonal components. The model was selected based on the Akaike Information Criteria corrected (AICc) minimum values and maximum log-likelihood. Model adequacy was checked using plots of residuals and the Ljung–Box test. Findings Using the Box–Jenkins method, various AR and MA lags were tested to determine the most optimal lags. ARIMA (12,1,12) is the most appropriate model obtained from the various models using AIC. As financial time series, such as Bitcoin returns, can be volatile, an attempt is made to model this volatility using GARCH (1,1). Originality/value The study used partially processed secondary data to fit for time series analysis using the ARIMA (12,1,12)-GARCH(1,1) model and hence reliable and conclusive results.
David Umoru, Beauty Igbinovia, Isah Aisha Shaibu, Muhammed Adamu Obomeghie
The study examined the volatility of Bitcoin prices and volatility of exchange rates of oil-producing countries. The study used ARIMA, GARCH estimators for analysis. The study found ARCH effects in the data (heterskedasticity test; p<.05). The GARCH results laid credence to a confirmation of adjustments in the Bitcoin market having significant volatility influence on local currencies. Persistent volatility and volatility clustering found in some of the sampled countries denote increased risk and uncertainty in foreign exchange markets that stimulates increased borrowing costs and reduced liquidity. The actual and forecast values based on the ARIMA method match with an Out-of-Sample period plotted for forecast (27/12/2022 to 27/12/2024) except for Nigeria. The ARIMA models for UAE and Kuwait stand out with excellent fit and prediction accuracy. The poor ARIMA model for Nigeria was ascribed to the hyper-inflation in the economy and extremely volatile money market. In line with the efficient market hypothesis, significant interactions are pegged on available information being already reflected in the current value of the currencies. In effect, past currency rates and Bitcoin trading prices are useful predictors of future prices having factored in the relevant information that could influence currency's value. In addition, future values of local currencies can be forecasted from past values at a significant level of accuracy. Countries should ensure adequate regulation of the foreign exchange markets so as to curtail the wave of volatility risks on returns associated with Bitcoin trading and exchange rates.
Abstract Cryptocurrency, particularly Bitcoin, has become a major topic in the financial and digital trading sectors due to its ability to facilitate direct transactions without intermediaries and the transparency offered by blockchain technology. However, the high volatility of Bitcoin prices necessitates accurate prediction methods to support better investment decisions. This research aims to compare the accuracy of Linear Regression and Long Short-Term Memory (LSTM) methods in predicting Bitcoin prices using historical data from Yahoo Finance. The research process begins with the collection of historical Bitcoin price data from September 17, 2014, to July 15, 2024, followed by data processing that includes cleaning and splitting the dataset into training and test data. Linear Regression and LSTM models are applied to the training data and tested to evaluate their performance in price prediction. The research findings show that the LSTM model significantly outperforms the Linear Regression model in terms of prediction accuracy. The LSTM model records much lower Mean Squared Error (MSE) and Root Mean Squared Error (RMSE), as well as perfect R² scores on both datasets, demonstrating its high precision in prediction. In contrast, the Linear Regression model shows higher errors and lower explanatory power of data variability. These findings indicate that LSTM is more effective in capturing temporal patterns and Bitcoin price fluctuations, offering better accuracy and potentially being more suitable for future cryptocurrency price analysis, providing better guidance for investors in this highly dynamic market.
Stock price prediction is currently a research focus in the financial field, especially in blockchain research. The central focus of this research is to forecast Bitcoin's closing price through the integration of deep learning techniques, specifically employing Long Short-Term Memory (LSTM). This study takes into account that Bitcoin is a mainstream virtual currency, and predicting its future price can help investors make better judgments in trading. The goal of this exploration is to identify the most favorable parameter combinations and function prediction applications, ultimately obtaining the most accurate prediction results. The research process includes dataset selection, data processing, model construction, and training. Then adjust and improve the parameters used in the model, and record the process. Finally, test the model and output the test results. And model testing and result output. At the end of the experiment, the effects of different optimizers and parameters on the training results were compared, and the optimal combination was found. The model's predictive accuracy was evaluated through the examination of test data. This study can provide valuable references for researchers and firms.
Poorva Nayyar, K. K. Bhardwaj, Saiyam Gupta, Ravi Prakash Chaturvedi · 6 authors
The development of financial technology has given rise to a new kind of asset called cryptocurrency, which has presented a significant potential for study. Forecasting cryptocurrency prices is challenging because of their dynamism and unpredictability. Each of the three recurrent neural network, or RNN, algorithms proposed in this paper may be used to predict the prices of three distinct cryptocurrency types: Bitcoin (BTC), Litecoin (LTC), and Ethereum (ETH). The algorithms generate accurate forecasts based on the average absolute percentage error (MAPE). For both cryptocurrency variations, the gated recurrent unit (GRU) fared better in terms of prediction than the long short-term memory (LSTM) and bidirectional LSTM (bi-LSTM) models, according to the models' results. It is therefore regarded as the best algorithm.
Abstract This paper applies deep learning models to predict Bitcoin price directions and the subsequent profitability of trading strategies based on these predictions. The study compares the performance of the convolutional neural network–long short-term memory (CNN–LSTM), long- and short-term time-series network, temporal convolutional network, and ARIMA (benchmark) models for predicting Bitcoin prices using on-chain data. Feature-selection methods—i.e., Boruta, genetic algorithm, and light gradient boosting machine—are applied to address the curse of dimensionality that could result from a large feature set. Results indicate that combining Boruta feature selection with the CNN–LSTM model consistently outperforms other combinations, achieving an accuracy of 82.44%. Three trading strategies and three investment positions are examined through backtesting. The long-and-short buy-and-sell investment approach generated an extraordinary annual return of 6654% when informed by higher-accuracy price-direction predictions. This study provides evidence of the potential profitability of predictive models in Bitcoin trading.
<p>In this study, we use Euler-Maruyama method to simulate Bitcoin prices dynamics. We investigate a year-long movement of Bitcoin prices. Daily closing prices were collected over a period starting from May 27th, 2023 and ending on May 27th, 2024. This data provides a comprehensive picture of how Bitcoin behaved on a daily basis throughout that specific year. The Euler-Maruyama method is used as numerical method for solving stochastic differential equations (SDEs). The method involves discretizing the SDE into an iterative process to obtain a simulated price trajectory. The drift term was estimated by the average daily return of Bitcoin prices over the study period. The volatility term was estimated by the standard deviation of daily Bitcoin returns. A Monte Carlo simulation was performed to generate a range of possible price trajectories using the Euler-Maruyama method. The result shows that the Euler-Maruyama method was able to capture the short-term trend of Bitcoin prices effectively. However, the method can also be used to capture the long-term trend of Bitcoin prices.</p>