This paper describes an architecture for predicting the price of cryptocurrencies for the next seven days using the Adaptive Network Based Fuzzy Inference System (ANFIS). Historical data of cryptocurrencies and indexes that are considered are Bitcoin (BTC), Ethereum (ETH), Bitcoin Dominance (BTC.D), and Ethereum Dominance (ETH.D) in a daily timeframe. The methods used to teach the data are hybrid and backpropagation algorithms, as well as grid partition, subtractive clustering, and Fuzzy C-means clustering (FCM) algorithms, which are used in data clustering. The architectural performance designed in this paper has been compared with different inputs and neural network models in terms of statistical evaluation criteria. Finally, the proposed method can predict the price of digital currencies in a short time.
As a consequence of rising geo-economic issues, global currency values have declined during the last two years, stock markets have performed poorly, and investors have lost money. Consequently, there is a renewed interest in digital currencies. Cryptocurrency is a fresh kind of asset that has evolved as a result of fintech innovations, and it has provided a major research opportunity. Due to price fluctuation and dynamism, anticipating the price of cryptocurrencies is difficult. There are hundreds of cryptocurrencies in circulation around the world and the demand to use a prediction system for price forecasting has increased manifold. Hence, many developers have proposed machine learning algorithms for price forecasting. Machine learning is fast evolving, with several theoretical advances and applications in a variety of domains. This study proposes the use of three supervised machine learning methods, namely linear regression, support vector machine, and decision tree, to estimate the price of four prominent cryptocurrencies: Bitcoin, Ethereum, Dogecoin, and Bitcoin Cash. The purpose of this study is to compute and compare the precision of all three techniques over all four datasets.
This paper illustrates the working process of predicting the Bitcoin price applying ARIMA, SARIMA and linear regression. Since more and more machine learning models were developed and tested in the financial field, these three models are selected to examine their reliabilities. In this study, three methodologies have been used for the Bitcoin predictions under the data set of Bitcoin historical prices. With the help of python notebook, order (1, 1, 1) and seasonal order (0, 1, 1, 12) were applied to the predictions in ARIMA and SARIMA respectively. In terms of linear regression, this paper used two independent variables including historical data and trading volume to predict the Bitcoin prices. It was discovered that the predictive graph for these three methodologies can match the actual value well, and linear regression performs the best. Considering the rapid development of machine learning methods, adopting alternative methods deserve in-depth investigations.
Ahmed Bouteska, Mohammad Zoynul Abedin, Petr Hájek, Kunpeng Yuan
Cryptocurrency price forecasting is attracting considerable interest due to its crucial decision support role in investment strategies. Large fluctuations in non-stationary cryptocurrency prices motivate the urgent need for accurate forecasting models. The lack of seasonal effects and the need to meet a number of unrealistic requirements make it difficult to make accurate forecasts using traditional statistical methods, leaving machine learning, particularly ensemble and deep learning, as the best technology in the area of cryptocurrency price forecasting. This is the first work to provide a comprehensive comparative analysis of ensemble learning and deep learning forecasting models, examining their relative performance on various cryptocurrencies (Bitcoin, Ethereum, Ripple, and Litecoin) and exploring their potential trading applications. The results of this study reveal that gated recurrent unit, simple recurrent neural network, and LightGBM methods outperform other machine learning methods, as well as the naive buy-and-hold and random walk strategies. This can effectively guide investors in the cryptocurrency markets.
The study aims to predict the close prices of four different cryptocurrencies (Bitcoin, Ethere-um, Dogecoin, and Cardano) using machine learning techniques and determine which of these cryptocurrencies is suitable for investment. To achieve this goal, we used two popular gradi-ent boosting algorithms: Extreme Gradient Boosting (XGBoost) and Light Gradient-Boosting Machine (LightGBM). Prediction accuracy of the trained model is evaluated by Mean Abso-lute Error (MAE) generated by the methodology of Cross-Validation. Our results show that both XGBoost and LightGBM can effectively predict the close prices of the four cryptocur-rencies, with LightGBM achieving slightly better performance in terms of prediction accura-cy. Based on our analysis, we were able to identify which cryptocurrencies were suitable for investing and provide recommendations for potential investors. Overall, our study highlights the potential of machine learning techniques in predicting cryptocurrency close prices and identifying suitable investment opportunities.
<span>The majority of research on predicting the price of Bitcoin employs technical methods to enhance long short-term memory models' effectiveness. Although some studies employ different machine learning techniques, such as economic or technical indicators, their precision is inadequate. Thus, this research aims to introduce a model that predicts the price of Bitcoin by utilizing the long short-term memory (LSTM) technique and incorporating gold's economic and technical data as features. The research collected gold and Bitcoin price data from FinanceDataReader for around seven years, from January 1, 2016, to January 22, 2023, consisting of six categories: date, open, high, low, close, volume, and change (based on dollars). The normalized closing price data was trained for 50 epochs, resulting in the loss value reaching close to zero. The model's accuracy was measured by mean squared error, resulting in a score of 0.0004. This study's importance is two-fold: firstly, it can provide cryptocurrency-related businesses with more accurate predictions and improved risk management indicators. Secondly, incorporating economic metrics can address the limitations of overfitting and a single model's poor performance.</span>
As cryptocurrencies become widely accepted due to technical improvements, reliable approaches to capture their future price movements of them become critical. This study mainly combines weighted sentiment analysis results from social media-related comments and financial news headlines with a stacked LSTM model to predict second-day Bitcoin price evolution. This study also compared our results and the results produced by MLP, RF, and SVM after feeding the sentiment analysis results.
Peter T. Yamak, Yujian Li, Ting Zhang, Kyefondeme C. Dakurah
This research proposes a novel approach for forecasting cryptocurrency prices, specifically Bitcoin which dominates the market. Accurately predicting cryptocurrency values is challenging due to their highly volatile nature. The proposed hybrid model uses ResNet Convolutional Neural Network to encode Bitcoin price time series data into discriminative representations. These representations capture long-range dependencies using XGBoost regression. Additionally, wavelet denoising is applied to filter noise from the price data. The combined ResNet-XGBoost-Wavelet model achieves satisfactory results for Bitcoin price forecasting and has practical applications for developing quantitative trading strategies. While incorporating sentiment analysis and additional influencing factors could further improve predictions, this work presents a competitive approach for minimizing investment risks and maximizing profits in the complex domain of cryptocurrency markets.
Guzalxon Belalova, Shakhida Gaybullaevna Mannanova, Botirjon Karimov
Over the past few decades, recurrent neural networks, particularly the Long Short-Term Memory (LSTM) architecture, have undergone several refinements. These networks have emerged as the go-to models for numerous machine learning challenges, especially those involving sequential data. One such application is the prediction of Bitcoin prices, a cryptocurrency that stands at the forefront of blockchain technology. This paper delves into the intricacies of forecasting Bitcoin prices using a suite of models, with a keen emphasis on the LSTM architecture, renowned for its prowess in handling tasks with long-term dependencies. Our exploration encompasses traditional time series models like ARIMA, neural network variants such as ANN and Transformer-based models, and even hybrid combinations. Specifically, our LSTM model, augmented with peephole connections, demonstrates its capability to learn and predict Bitcoin price fluctuations. We source our data from the Bitcoin Price Index and aim to gauge the accuracy with which these models can predict Bitcoin's price trajectory. Furthermore, our experiments involve the deployment of an "adam"-optimized Recurrent Neural Network (RNN) and Long Short-Term Memory (LSTM) network, revealing insights into their predictive performances.
Alexandru Costin Baroiu, Vlad Dıaconıța, Simona‐Vasilica Oprea
Cryptocurrencies have emerged as a popular investment vehicle, prompting numerous efforts to predict market trends and identify metrics that signal periods of volatility. One promising approach involves leveraging on-chain data, which is unique to cryptocurrencies. On-chain data, extracted directly from the blockchain, provides valuable information, such as the hash rate, total transactions, or the total number of addresses that hold a specified amount of cryptocurrency. Some studies have also explored the relationship between social media sentiment and Bitcoin, using data from platforms such as Twitter and Google Trends. However, the quality of Twitter sentiment analysis has been lackluster due to suboptimal extraction techniques. This research proposes a novel approach that combines a superior sentiment analysis technique with various on-chain metrics to improve predictions using a deep learning architecture based on long-short term memory (LSTM). The proposed model predicts outcomes for multiple time horizons, ranging from one day to 14 days, and outperforms the Martingale (random walk) approach by over 9%, as measured by the mean absolute percentage error metric, as well as recent results reported in literature. To the best of our knowledge, this study may be among the first to employ this combination of techniques to improve cryptocurrency market prediction.
This paper focuses on the prediction of Bitcoin prices and returns based on the Long Short Term Memory (LSTM) neural network model, to better consider the impact of time factors. Since Bitcoin has long dominated the digital currency trading market, many researchers have completed many Bitcoin prediction results, including the screening of optimal features, comparison of prediction models and classification of prediction problems. Based on previous work, this article adds a Bitcoin revenue forecast section, presenting the results in the form of charts and data to provide more intuitive trends and more accurate performance. This paper uses LSTM as the experimental model, and uses the Bitcoin transaction history data set with timestamps as the original input. After a specific normalization method, the original model is trained, and then the subsequent transaction data is predicted. Compare it with the real value in the data set to get the final experimental results show that in this prediction problem, the performance of LSTM is slightly better than Autoregressive Integrated Moving Average (ARIMA) and eXtreme Gradient Boosting (XGBoost); on the other hand, compared with price prediction based on real values for prediction, the prediction fluctuations of return are more obvious and more realistic, providing better reference value.
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.
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.
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.
In the realm of financial markets, the manifestation of volatility clustering serves as a pivotal element, indicative of the inherent fluctuations characterizing financial instruments. This attribute acquires pronounced relevance within the sphere of cryptocurrencies, a sector renowned for its elevated risk profile. The present analysis, conducted through the Autoregressive Moving Average - Generalized Autoregressive Conditional Heteroskedasticity (ARMA-GARCH) model, seeks to elucidate the enduring nature of volatility clustering and the occurrence of leverage effects within this domain. Over the course of a four-year time frame, it was observed that Bitcoin diverges from the anticipated Autoregressive Conditional Heteroskedasticity (ARCH) effects, in contrast to Ethereum and Cardano, which exhibit marked volatility clustering. Binance Coin, Ripple, and Dogecoin, whilst demonstrating moderate clustering, uniformly reflect the existence of leverage effects. An exception to this pattern was identified in Ripple, where it was discerned that positive market news exerts a disproportionate influence on log returns. The findings of this study illuminate the critical influence of both leverage effects and volatility clustering on the pricing dynamics of cryptocurrencies. It underscores the imperative for a nuanced comprehension of risk management in the context of cryptocurrency investments, given their susceptibility to abrupt price fluctuations. The distinct degrees to which these phenomena are manifested across diverse cryptocurrencies accentuate the necessity for a tailored risk management approach, resonant with the unique attributes of the asset in question. Such strategies, accounting for the potential amplification of losses through leverage, may encompass prudent position sizing, portfolio diversification, and the implementation of stress tests, thereby fortifying the investment against the dual perils of volatility clustering and leverage effects. The implications of this analysis serve to inform investors, providing a foundation upon which to construct risk management tactics that are responsive to the idiosyncrasies of the cryptocurrency market.
Fathin Faizah Said, Raja Solan Somasuntharam, Mohd Ridzwan Yaakub, Tamat Sarmidi
Abstract Advanced digitalization and financial technology have of recent times become among the most crucial tools. Data mining and sentiment analysis have revealed the importance of digitalization in modern times. This study examines the influence of Google search activity on the volatility of digital assets. We analyzed six digital asset prices for Bitcoin, Bitcoin Cash, Ethereum, Ethereum Classic, Litecoin, and Ripple from the Coinmarketcap database. We used tweets on Twitter to survey users’ sentiment by using the Twitter search Application Programming Interface and Google trend search from web searches, news searches, and YouTube searches data using RStudio software. The study spanned 1 September 2019 to 31 January 2020 and employed the Vector Autoregression (VAR) approach for analysis. The VAR estimation revealed that Google search variables have significantly influenced the volatility of Bitcoin, Ethereum, Litecoin, and Ripple, as supported by the Granger causality test and impulse response function. The results of this study could be useful for investors and policymakers in drawing up strategies to reduce market volatility. These results should thus be useful to investors in developing profitable investment strategies to mitigate the impact of market turbulence.
Since the advent of Bitcoin, the cryptocurrency landscape has seen the emergence of several virtual currencies that have quickly established their presence in the global market. The dynamics of this market, influenced by a multitude of factors that are difficult to predict, pose a challenge to fully comprehend its underlying insights. This paper proposes a methodology for suggesting when it is appropriate to buy or sell cryptocurrencies, in order to maximize profits. Starting from large sets of market and social media data, our methodology combines different statistical, text analytics, and deep learning techniques to support a recommendation trading algorithm. In particular, we exploit additional information such as correlation between social media posts and price fluctuations, causal connection among prices, and the sentiment of social media users regarding cryptocurrencies. Several experiments were carried out on historical data to assess the effectiveness of the trading algorithm, achieving an overall average gain of 194% without transaction fees and 117% when considering fees. In particular, among the different types of cryptocurrencies considered (i.e., high capitalization, solid projects, and meme coins), the trading algorithm has proven to be very effective in predicting the price trends of influential meme coins, yielding considerably higher profits compared to other cryptocurrency types.
Vincent Gurgul, Stefan Lessmann, Wolfgang Karl Härdle
We introduce novel approaches to cryptocurrency price forecasting, leveraging Machine Learning (ML) and Natural Language Processing (NLP) techniques, with a focus on Bitcoin and Ethereum. By analysing news and social media content, primarily from Twitter and Reddit, we assess the impact of public sentiment on cryptocurrency markets. A distinctive feature of our methodology is the application of the BART MNLI zero-shot classification model to detect bullish and bearish trends, significantly advancing beyond traditional sentiment analysis. Additionally, we systematically compare a range of pre-trained and fine-tuned deep learning NLP models against conventional dictionary-based sentiment analysis methods. Another key contribution of our work is the adoption of local extrema alongside daily price movements as predictive targets, reducing trading frequency and portfolio volatility. Our findings demonstrate that integrating textual data into cryptocurrency price forecasting not only improves forecasting accuracy but also consistently enhances the profitability and Sharpe ratio across various validation scenarios, particularly when applying deep learning NLP techniques. The entire codebase of our experiments is available via an online repository: https://anonymous.4open.science/r/crypto-forecasting-public . • NLP data from social media improve the accuracy of cryptocurrency forecasting models. • As a target variable, local extrema are a valid alternative to daily price changes. • Deep learning language models substantially outperform dictionary-based methodologies. • Both pre-trained and fine-tuned language models effectively quantify market sentiment.