This study examines the causal relationship between cryptocurrencies and other major world economic assets, such as gold, stocks, oil, and bonds, using both Granger causality and correlation analyses. The study focuses on the period between 2018 and 2022, using a vector autoregressive model (VAR) to analyze data on cryptocurrencies and other major world economic assets, which collectively represent over 90% of the market during the observed period. Results show that correlation clearly identifies causal interdependency between cryptocurrencies and other major world economic assets and that the variation in cryptocurrencies increasingly explains other major world economic assets. The results reveal that there is Granger causality between the cryptocurrencies (Tether, USD Coin, and Binance USD) and the other major world economic assets (BOND, SP500, and GOLD). Additionally, the study finds evidence that market inefficiency in the cryptocurrency market increased between 2018 and 2022. The findings suggest that the properties of the cryptocurrency market are highly dynamic and that researchers should be hesitant to generalize the market properties observed during idiosyncratic periods. The relevant information is swiftly reflected in asset prices when investors are more interested in a news event, increasing volatility. Strong evidence suggests that volatility spill overs increase sharply at this time. The structure of these markets frequently changes, and a large number of cryptocurrencies appear and disappear every day.
The Bitcoin price was chosen as the research subject, and the observation period was set from January 2015 to September 2023. An ARIMA time series model was constructed to forecast the trading price. The results indicate that the optimal model for fitting the trading price is ARIMA (3, 2, 8). This model takes into account trends, seasonality, and other factors that may impact the price of Bitcoin. By analyzing the historical data, the model was able to accurately predict the short-term fluctuations in Bitcoin’s trading price. Based on this, short-term predictions were made for Bitcoin’s trading price in the next year. Recommendations were then provided by combining the forecast results with the economic development situation in the post-pandemic era. The recommendations suggest that Bitcoin has become a low-quality asset and is no longer suitable for diversifying one’s investment portfolio, but rather focus on the development of physical industries and adjust one’s investment portfolio in a timely manner.
Bitcoin is a digital currency created by a large number of calculations based on a specific algorithm. With the time development, more investors came into the market and the price of the bitcoin had been changing all the time. But bitcoin investors want to be able to predict price fluctuations because they don't want to lose their profits. This paper uses machine learning and artificial intelligence to make some reasonable predictions about Bitcoin price fluctuations.
The cryptocurrency and stock markets are dynamic environments that attract traders, seeking to enhance their investment returns. In cryptocurrency trading, there is a pullback in investors from trading due to recent market crashes, losses, and bankruptcies. For anticipating future market behavior, algorithmic trading has gained popularity due to its ability to provide consistent and accurate price and volatility predictions. Specifically, the bottom turning points of the market are where an investor can use to enter the market. Hence, identifying market turning points, particularly market bottoms, is vital in timing trading strategies for a maximum profit. This study introduces a novel and ground-breaking approach to market forecasting that focuses on identifying market bottoms, particularly in the domain of cryptocurrency trading. The study utilizes a Wasserstein Generative Adversarial Network (WGAN) with Gated Recurrent Unit (GRU) to identify future market trends effectively. A classifier is added into the model as a substantial contribution to forecast future market bottoms by utilizing hidden WGAN features. The research findings indicate that the combination of the price prediction and bottom classification models provides outperforming results in terms of prediction accuracy. In addition, the suitability of the proposed solution for locating stock market bottoms has been evaluated.
With the rapid development of the Internet, digital cryptocurrencies based on blockchain technology have been widely used globally. However, the huge volatility and high risk of cryptocurrency prices pose challenges for investors. To address this issue, predicting the prices of digital cryptocurrencies has become a research focus. However, most existing studies mainly focus on Bitcoin price prediction. This paper proposes a GRU (Gated Recurrent Unit) model-based method for predicting the price of Dogecoin, a popular emerging cryptocurrency. The choice of Dogecoin is motivated by its high price volatility and prediction difficulty as a relatively new cryptocurrency. The GRU model is a variant of the recurrent neural network (RNN) that has better prediction performance compared to the LSTM model. With this method, we can effectively predict the price of Dogecoin, reducing investment risks for investors and providing reference for policymakers in regulating the digital currency market.
Jongyeop Kim, Jongho Seol, Tasnim Akter Onisha, Yiming Ji
Hyperparameter configurations highly affected the accuracy of the deep learning model. This study focuses on finding an appropriate parameter set that can apply to the Long Short-Term Memory (LSTM) and GRU (Gated Recurrent Unit) for cryptocurrency price prediction. The 80% portion of the data set is composed of an Open-high-low-close (OHLC), movement in the price over time, considered a training data set to predict the remaining 20% of OHLC. Our method classified several appropriate hyperparameter sets, leading to a high accuracy in terms of root mean square error (RMSE) on varying conditions, including number of layers, epochs, and batch size.
Jutur Manogna, Gogineni Sravan Chowdary, Gogineni Meghana, Priyanka C. Nair
Bitcoin is a decentralized digital currency that has received a lot of interest in recent years because of its unique qualities and possibilities as an alternative investment. Trading in Bitcoin might be difficult owing to the extreme volatility of its price, which is impacted by a variety of variables such as market sentiment and regulatory changes. To solve this issue, researchers and traders have been investigating the use of social media data, namely Twitter data, to forecast Bitcoin price changes using sentiment analysis. The work focuses on leveraging real-time Twitter data and bitcoin prices from the Twitter and Coin Market API respectively to predict short-term Bitcoin price movements. Real-time tweets are collected, sentiment is extracted using sentiment analysis methods, and then combined with relevant pricing data. A predictive framework is developed to forecast the next hour's Bitcoin price by employing univariate and multivariate time series forecasting and generating deep learning models such as LSTM, BIGRU, RNN and LSTM+GRU. Multivariate time series forecasting model based on BIGRU has performed well among the deep learning models used, attaining a 130.529 RMSE score. The primary aim of this study is to provide valuable insights to traders, as short-term market sentiment plays a crucial role in their trading strategies.
Abstract: The goal of this project is to use machine learning to forecast cryptocurrency values. As a result of their high levels of volatility, cryptocurrencies are notoriously difficult to anticipate in terms of value. SARIMA (Seasonal Auto Regressive Integrated Moving Average) algorithm that we suggest using to capture the intricate dynamics of the bitcoin market. Our machine learning models will be trained using the gathered data, and they will then be utilised to forecast future cryptocurrency values. The project's final product is anticipated to be a useful tool for cryptocurrency traders, analysts, and investors, giving them a more precise way to make data-drive investment decisions.
Md. Nafis Tahmid Akhand, Md. Ahsan Habib, Kazi Md. Rokibul Alam
Cryptocurrency has emerged as a popular investment option due to its decentralized nature and potential for high returns. However, the cryptocurrency market is characterized by high volatility and price fluctuations, making it difficult for investors and traders to make informed decisions. This paper aims to address this challenge by performing a time series analysis of cryptocurrency prices to estimate values in real-time. To achieve this goal, historical price data for three cryptocurrencies—Bitcoin, Ethereum, and Litecoin are gathered and preprocessed. Following this, a range of time series techniques are used to analyze the patterns and trends in the data. This study focuses on hourly and daily data of the cryptocurrencies and employs three hybrid models such as CNN-LSTM (CLT), CNN-GRU (CGR), and CNN-BiLSTM (CBL) to forecast upcoming prices. Among three models, the CLT technique outperforms other models with RMSE of 235.97, MAE of 135.42, and MAPE of 0.47% on hourly Bitcoin price prediction. The experimental results demonstrate the effectiveness of the proposed methods in predicting the prices of cryptocurrencies in real-time.
In recent times, the cryptocurrency market has emerged as one of the fastest-growing financial markets worldwide. It is, however, known for its high volatility and illiquidity compared to traditional markets such as equities, foreign exchange, and commodities. This inherent risk creates uncertainty among investors. The aim of this research is to forecast the level of risk in the cryptocurrency market. To assist cryptocurrency investors in navigating these challenges, we propose an approach that involves calculating the risk factor based on existing parameters. We employed various machine learning algorithms, including CNN, LSTM, BiLSTM, and GRU, to predict the risk factor in twenty elements of the cryptocurrency market. Through extensive experimentation, we developed a new model that outperformed existing models, achieving the highest Root Mean Square Error (RMSE) value of 1.3229 and the lowest RMSE value of 0.0089. Furthermore, we tested the generalization ability of our proposed model on a new dataset, different from the one used for training. Even with this new dataset, our model displayed robust performance. In contrast, the other existing models achieved higher RMSE values, with the highest being 14.5092 and the lowest 0.02769. By adopting our approach, investors can trade more confidently in complex and challenging financial assets such as Bitcoin, Ethereum, and Dogecoin. Our proposed model demonstrates superior performance and generalization capabilities, providing valuable insights for participants in the cryptocurrency market.
B Nataraj, K R Prabha, V Swetha, S Sukitha · 5 authors
Bitcoin, a decentralized digital currency, operates without the involvement of traditional financial institutions. Bitcoin transactions are conducted directly between parties, without the use of intermediaries, thanks to blockchain technology. Wallets, public keys, and private keys are necessary for the secure transactions of this cryptocurrency. Transaction can take place using Bitcoin simply going via centralized exchanges, in contrast with numerous other cryptocurrencies. This paper explores the unique features of Bitcoin and its advantages over other crypto assets. Notably, the decentralized and distributed ledger of blockchain technology ensures the secure storage of verified bitcoin transactions by network nodes. This distinguishes Bitcoin from assets that rely on centralized exchanges for verification. Recognizing the growing acceptance of Bitcoin in various transactions, including those conducted by small businesses, this research focuses on the need for accurate early prediction of Bitcoin prices. The study proposes leveraging machine learning algorithms, specifically Random Forest and Deep Learning (Long Short Term Memory), to predict the open and close values of Bitcoin. This predictive analysis aims to assist investors in making informed decisions and optimizing their Bitcoin investments. Accurate early forecast of Bitcoin prices is a need, as this research highlights, given the increasing use of Bitcoin in a variety of activities, including small company transactions. In order to forecast the open and closing prices of Bitcoin, the study suggests using machine learning techniques, notably Random Forest and Deep Learning (Long Short Term Memory). By using predictive analysis, investors may maximize their Bitcoin investments and make well-informed judgements. The study emphasizes the importance of accurate price predictions for Bitcoin investors and introduces machine learning algorithms as effective tools for achieving this goal. A comparative analysis of Random Forest and Long Short Term Memory algorithms will be conducted to evaluate their accuracy in predicting Bitcoin prices. The research aims to provide investors with valuable insights into optimizing their investment strategies based on reliable early predictions of Bitcoin values. This idea is reliable using Machine Learning which provides effective results. By Comparing the accuracy between two algorithms the prediction of bitcoin will be implemented. This can be done using Machine Learning Algorithm (Random Forest) and Deep Learning Algorithm (Long Short Term Memory). This prediction will help the bitcoin investors to identify the open and close value of the bitcoin so that the investors can invest their bitcoin in an efficient manner and get benefited.
Sulalitha Bowala, A. Thavaneswaran, Ruppa K. Thulasiram, Md Erfanul Hoque · 5 authors
Trading volume is an important variable to successfully capture market risks along with asset price/returns. Recently, there has been a growing interest in deep learning methods to forecast the trading volume of stocks using historical volatility as a feature. Unlike the existing work, a novel datadriven log volatility forecast is proposed in this paper as an extra feature to improve trading volume forecasts. Recently, neural networks for volatility and neural nets for electricity demand forecasting, constructed with nnetar function, have shown to be superior. The novelty of this paper is to demonstrate the neural network based on the nnetar function from the forecast package in R for trading volume forecast shows superiority over the other neural network.
Mehedi Hasan, Md. Tahmid Rahman, Kazi Ahnaf Alavee, Abu Hasnayen Zillanee · 6 authors
In the fast-paced realm of global financial markets, characterized by rapid trading of both stocks and cryptocurren-cies, it has become essential to grasp the influence of sentiment on market dynamics. With more than 630,000 publicly traded companies worldwide and major stock exchanges like the NYSE handling a substantial portion of global equity transactions, the inherent volatility of the stock market is well-established. Over the past decade, various factors have contributed to the consistent fluctuations in stock prices. One key factor is the influence of investor reviews sourced from diverse news outlets and social media platforms such as Twitter. Understanding how these reviews can be collected and effectively summarized is crucial. This paper centers on the intricate field of market sentiment analysis and its profound impact on user sentiment, subsequently affecting price fluctuations in both stocks and cryptocurrencies. In this study, we present a comprehensive exploration of the development and evaluation of an automated sentiment analysis system tailored for summarizing web-based news related to stocks and cryptocurrencies.We have implemented BERT (Bidirectional Encoder Representations from Transformers) in combination with NLTK for text summarization, a highly accurate model with a performance level of 95.84%, as part of our proposed approach.
This research employs a Selective Neural Network Ensemble driven by Genetic Algorithms, employing an Artificial Neural Network ensemble methodology. The ensemble incorporates base model of any neural network is the multi-layered perceptron. Aim to explore the correlation between Bitcoin's features and its subsequent day's price movement. Leveraging approximately 200 cryptocurrency attributes over a two-year span, the ensemble predicts the direction of Bitcoin's price the following day, aiming to assess its practicality and relevance in real-world scenarios. In a comparative the ensemble-based trading strategy is evaluated using back-testing analysis over a 50-day period versus a “prior day trend“ trading approach. The former approach demonstrated noteworthy results, offering insights into its potential effectiveness. The best data range for training a Bitcoin price prediction model is determined by applying financial terms and methods, such as Simple Moving Average and Exponential Moving Average, as described in the article. A Linear Regression Model addresses the problem of choosing the appropriate dataset for improved forecasting results, achieving a high 97% prediction accuracy by adhering to the model's recommended data piece.
Valeriia Baklanova, Aleksei Kurkin, Тамара Теплова
Purpose The primary objective of this research is to provide a precise interpretation of the constructed machine learning model and produce definitive summaries that can evaluate the influence of investor sentiment on the overall sales of non-fungible token (NFT) assets. To achieve this objective, the NFT hype index was constructed as well as several approaches of XAI were employed to interpret Black Box models and assess the magnitude and direction of the impact of the features used. Design/methodology/approach The research paper involved the construction of a sentiment index termed the NFT hype index, which aims to measure the influence of market actors within the NFT industry. This index was created by analyzing written content posted by 62 high-profile individuals and opinion leaders on the social media platform Twitter. The authors collected posts from the Twitter accounts that were afterward classified by tonality with a help of natural language processing model VADER. Then the machine learning methods and XAI approaches (feature importance, permutation importance and SHAP) were applied to explain the obtained results. Findings The built index was subjected to rigorous analysis using the gradient boosting regressor model and explainable AI techniques, which confirmed its significant explanatory power. Remarkably, the NFT hype index exhibited a higher degree of predictive accuracy compared to the well-known sentiment indices. Practical implications The NFT hype index, constructed from Twitter textual data, functions as an innovative, sentiment-based indicator for investment decision-making in the NFT market. It offers investors unique insights into the market sentiment that can be used alongside conventional financial analysis techniques to enhance risk management, portfolio optimization and overall investment outcomes within the rapidly evolving NFT ecosystem. Thus, the index plays a crucial role in facilitating well-informed, data-driven investment decisions and ensuring a competitive edge in the digital assets market. Originality/value The authors developed a novel index of investor interest for NFT assets (NFT hype index) based on text messages posted by market influencers and compared it to conventional sentiment indices in terms of their explanatory power. With the application of explainable AI, it was shown that sentiment indices may perform as significant predictors for NFT sales and that the NFT hype index works best among all sentiment indices considered.
Non-Fungible Tokens (NFTs) have revolutionized various industries and aspects of the digital world in several ways. Built on blockchain technology, NFTs provide a secure and transparent way to establish ownership and provenance of unique digital or physical items. This has wide-ranging implications, from art and collectibles to virtual real estate and digital goods. While NFTs offer many benefits, however, they also raise concerns, including environmental impacts due to energy-intensive blockchain networks, copyright and plagiarism issues, and speculative bubbles in the NFT market. In this work, we collected 200,000 tweets about NFTs and employed state-of-the-art neurosymbolic AI tools to better understand what are the online conversation drivers and sentiments around NFTs and, hence, gain insights about what makes them valuable.
Wulan Septya Zulmawati, Nonong Amalita, Syafriandi Syafriandi, Admi Salma

 Cryptocurrency provides the most return compared to other investment instruments, causing many novice traders to be attracted to crypto as a tool to make significant profits in the short term. One of the most widely used cryptocurrencies is Bitcoin. Trading is closely related to technical analysis. Various techniques in technical analysis cause beginner traders to have difficulties choosing the right technique. Machine learning methods can be an alternative to overcoming the barriers of beginner traders in the crypto market with predictive methods. One method of machine learning for prediction is Support Vector Regression (SVR). Using the Grid Search algorithm shows that this method has a good predictive accuracy value of 99,25% and MAPE 8,70%.
The autoregressive integrated moving average (ARIMA) model is a widely used technique for capturing past dependencies and trends in order to generate future predictions. This study presents a comparative analysis of the ARIMA model’s forecasting capabilities as applied to gold and Bitcoin prices. The methodology employed consists of obtaining historical price data, implementing machine learning techniques, fitting the ARIMA model, then validating its predictive ability using multiple error metrics. Our results indicated that the optimal ARIMA parameters for Bitcoin and gold are different, which emphasizes their different price behaviors. Additionally, the study examined implications for policy, including issues such as prices for CPUs and GPUs, the role of market dynamics, as well as the possibility for price manipulation, which is of special relevance for cryptocurrencies that exist outside the mainstream. The study also suggests potential directions for future research, such as applying advanced machine-learning techniques and adopting cross-validation. This research offers important insights regarding Bitcoin and gold price dynamics and demonstrates the applicability of the ARIMA model for financial forecasting while demonstrating the necessity of further investigation into more refined predictive models.
This study aims to analyze the effect of Bitcoin price spillover volatility on Altcoin prices (Ethereum, Tether, Binance Coin) and the price of the S&P 500 Index. The data used is weekly data with a research period from January 2018 to December 2022. The analysis used in this study is the Exponential Generalized Autoregressive Conditional Heteroscedasticity (EGARCH) model. The results show a volatility spillover effect between Bitcoin and Binance Coin with more positive shocks than adverse shocks in Bitcoin price volatility on Binance Coin price. Meanwhile, the spillover volatility between Bitcoin and Ethereum, Tether, and the S&P 500 Index cannot be known because the price data is homoscedastic, so it cannot be continued with EGARCH modelling because the data needs to meet the modelling requirements.
With the advent of the Web3.0 era, virtual assets have gained prominence in individuals’ asset portfolios, making Non-Fungible Tokens (NFTs) increasingly significant within the financial trading landscape. To address the issue of multicollinearity in regression analysis, this paper employs Principal Component Analysis (PCA) to perform dimensionality reduction on five correlated foundational sectors. Moreover, to enhance the accuracy and reliability of predictive outcomes, the study combines the Long Short-Term Memory (LSTM) model with the Autoregressive Moving Average-Generalized Autoregressive Conditional Heteroskedasticity (ARMA-GARCH) model. Through the application of these methods and practical implementation, the study forecasts the NFT index of the Hong Kong stock market for the next 30 days. This forecasting of return volatility contributes vital insights for investment decision-making. The research complements and offers application recommendations in financial innovation, deepening, and regulation. By devising novel products and tools to meet investor demands, providing risk management and investment opportunities, the model’s predictive outcomes can be utilized in regulatory and risk management strategies within the national financial trading market. This study provides regulatory guidance, policy formulation insights, and envisions further refinements of the research methodology by integrating information shock effects.
Cryptocurrencies created by Nakamoto in 2009 have gained significant interest due to their potential for high returns. However, the cryptocurrency market's unpredictability makes it challenging to forecast prices accurately. To tackle this issue, a deep learning model has been developed that utilizes Long Short-Term Memory (LSTM) neural networks and Convolutional Neural Networks (CNNs) to predict cryptocurrency prices. LSTMs, a type of recurrent neural network, are well-suited for analyzing time series data and have been successful in various prediction applications. Additionally, CNNs, primarily used for image analysis tasks, can be employed to extract relevant patterns and characteristics from input data in Bitcoin price prediction applications. This study contributes to the existing related works on cryptocurrency price prediction by exploring various predictive models and techniques, which involve a machine learning model, deep learning model, time series analysis, and as well as a hybrid model that combines deep learning methods to predict cryptocurrency prices as well as enhance the accuracy and reliability of the price predictions. To ensure accurate predictions in this study, a trustworthy dataset from investing.com was sought. The dataset, sourced from investing.com, consists of 1826 time series data samples. The dataset covers the time frame from January 1, 2018, to December 31, 2022, providing data for a period of 5 years. Subsequently, pre-processing was conducted on the dataset to guarantee the quality of the input. As a result of absent values and concerns regarding the dataset's obsolescence, an alternative dataset was sourced to avoid these issues. The performance of the LSTM and CNN models was evaluated using root mean squared error (RMSE), mean squared error (MSE), mean absolute error (MAE) and R-squared (R2). It was observed that they outperformed each other to a certain degree in short-term forecasts compared to long-term predictions, where the R2Â values for LSTM range from 0.973 to 0.986, while for CNNs, they range from 0.972 to 0.988 for 1 day, 3 days and 7 days windows length. Nevertheless, the LSTM model demonstrated the most favorable performance with the lowest error rate. The RMSE values for the LSTM model ranged from 1203.97 to 1645.36, whereas the RMSE values for the CNNs model ranged from 1107.77 to 1670.93. As a result, the LSTM model exhibited a lower error rate in RMSE and achieved the highest accuracy in R2Â compared to the CNNs model. Considering these comparative outcomes, the LSTM model can be deemed as the most suitable model for this specific case
This study employed variable moving average (VMA) trading rules and heatmap visualization because the flexibility advantage of the VMA technique and the presentation of numerous outcomes using the heatmap visualization technique may not have been thoroughly considered in prior financial research. We not only employ multiple VMA trading rules in trading crypto futures but also present our overall results through heatmap visualization, which will aid investors in selecting an appropriate VMA trading rule, thereby likely generating profits after screening the results generated from various VMA trading rules. Unexpectedly, we demonstrate in this study that our results may impress Ethereum futures traders by disclosing a heatmap matrix that displays multiple geometric average returns (GARs) exceeding 40%, in accordance with various VMA trading rules. Thus, we argue that this study extracted the diverse trading performance of various VMA trading rules, utilized a big data analytics technique for knowledge extraction to observe and evaluate numerous results via heatmap visualization, and then employed this knowledge for investments, thereby contributing to the extant literature. Consequently, this study may cast light on the significance of decision making via big data analytics.