Kubra Yildiz, Sefa Dedebek, Feyza Yıldırım Okay, Mehmet Ulvi Şimşek
Anomaly detection in the financial sector has a critical importance for financial markets, investors, and regulatory authorities. As financial environments change, real-time detection of anomalies becomes more difficult due to the increase in data speed and volume with increasing digitization. Recently, deep learning (DL) algorithms have been used as a promising approach to solving the anomaly detection problem. In this study, DL-based anomaly detection model in the financial sector is presented using various DL algorithms including Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), and one-dimensional Convolutional Neural Network (1dCNN). In addition, hyperparameter optimization is performed with the grid search method. These methods are applied to two popular financial data, Tesla's stock market and Ethereum cryptocurrency data sets. Accordingly, a comparative analysis is conducted on these data sets with DL algorithms. Experimental results clearly show that GRU algorithm has the highest prediction score in both data sets, while 1dCNN algorithm has the lowest prediction score. In addition, anomaly values are demonstrated graphically with GRU for both data sets.
Experimental data was obtained from Bitcoin public library made available at https://www.investing.com/crypto/bitcoin/historical-dat which has attributes: open price, high, low and closes as shown in Table 1.
Nur Azmina Mohamad Zamani, Jasy Liew Suet Yan, Ahmad Muhyiddin Yusof
The effect of sentiment from news sources on cryptocurrency prices has garnered significant interest from the financial community including researchers. However, current findings reporting the usefulness of sentiment features in cryptocurrency price prediction is still mixed. This paper aims to explore various feature combinations encompassing cryptocurrency historical prices and sentiment scores of the current day aggregated from English and Malay news sources to predict the closing price of the next day. Each news headline related to Bitcoin and Ethereum within the duration of one year is first manually annotated and then aggregated on a daily basis to generate the sentiment features. A Bi-GRU deep learning model is implemented to predict the cryptocurrency closing price of the next day given a combination of historical price and sentiment features of the current day. Our findings evidently show sentiment features when combined with the closing price of the current day contribute significantly to improve the model's performance. Our study is also the first attempt to examine the effect of Malay news sentiment on cryptocurrency prices and we have demonstrated that cryptocurrency price prediction models leveraging Malay news sentiment features for Bitcoin and Ethereum are able to yield performance that is at par with models using English news sentiment features.
Deep learning techniques can predict cognitive intelligence from large datasets involving complex computations with activation functions. However, the prediction output needs verification for trust and reliability. Moreover, these algorithms suffer from the model's provenance to keep track of model updates and developments. Blockchain smart contracts provide a trustable ledger with consensus-based decisions that assure integrity and verifiability. In addition, the immutability feature of blockchain also supports the provenance of data that can help deep learning algorithms. Nevertheless, smart contract languages cannot predict due to the absence of floating-point operations required by activation functions of neural networks. In this paper, we derive a novel method using the Taylor series expansion to compute the floating-point equivalent output for activation functions. We train the deep learning model off-chain using a standard Python programming language. Moreover, we store models and predict on-chain with blockchain smart contracts to produce a trusted forecast. Our experiment and analysis achieved an accuracy (99%) similar to popular Keras Python library models for the MNIST dataset. Furthermore, any blockchain platform can reproduce the activation function using our derived method. Last but not least, other deep learning algorithms can reuse the mathematical model to predict on-chain.
The machine learning algorithms can predict the events based on the trained models and datasets. However, a reliable prediction requires the model to be trusted and tamper-resistant. Blockchain technology provides trusted output with consensus-based transactions and an immutable distributed ledger. The machine learning algorithms can be trained on blockchain smart contracts to produce trusted models for reliable prediction. But most smart contracts in the blockchain do not support floating-point data type, limiting computations for classification, which can affect the prediction accuracy. In this work, we propose a novel method to produce floating-point equivalent probability estimation to classify labels on-chain with a Naive Bayes algorithm. We derive a mathematical model with Taylor series expansion to compute the ratio of the posterior probability of classes to classify labels using integers. Moreover, we implemented our solution in Ethereum blockchain smart-contract with the Solidity programming language, where we achieved a prediction accuracy comparable to the scikit-learn library in Python. Our derived method is platform-agnostic and can be supported in any blockchain network. Furthermore, machine learning and deep-learning algorithms can borrow the derived method.
Sang-Ha Sung, Jong‐Min Kim, Byung-Kwon Park, Sangjin Kim
Cryptocurrencies are highly volatile investment assets and are difficult to predict. In this study, various cryptocurrency data are used as features to predict the log-return price of major cryptocurrencies. The original contribution of this study is the selection of the most influential major features for each cryptocurrency using the volatility features of cryptocurrency, derived from the autoregressive conditional heteroskedasticity (ARCH) and generalized autoregressive conditional heteroskedasticity (GARCH) models, along with the closing price of the cryptocurrency. In addition, we sought to predict the log-return price of cryptocurrencies by implementing various types of time-series model. Based on the selected major features, the log-return price of cryptocurrency was predicted through the autoregressive integrated moving average (ARIMA) time-series prediction model and the artificial neural network-based time-series prediction model. As a result of log-return price prediction, the neural-network-based time-series prediction models showed superior predictive power compared to the traditional time-series prediction model.
Abstract Nowadays, the issue of fluctuations in the price of digital Bitcoin currency has a striking impact on the profit or loss of people, international relations, and trade. Accordingly, designing a model that can take into account the various significant factors for predicting the Bitcoin price with the highest accuracy is essential. Hence, the current paper presents several Bitcoin price prediction models based on Convolutional Neural Network (CNN) and Long-Short-Term Memory (LSTM) using market sentiment and multiple feature extraction. In the proposed models, several parameters, including Twitter data, news headlines, news content, Google Trends, Bitcoin-based stock, and finance, are employed based on deep learning to make a more accurate prediction. Besides, the proposed model analyzes the Valence Aware Dictionary and Sentiment Reasoner (VADER) sentiments to examine the latest news of the market and cryptocurrencies. According to the various inputs and analyses of this study, several effective feature selection methods, including mutual information regression, Linear Regression, correlation-based, and a combination of the feature selection models, are exploited to predict the price of Bitcoin. Finally, a careful comparison is made between the proposed models in terms of some performance criteria like Mean Square Error (MSE), Root Mean Square Error (RMSE), Mean Absolute Error (MAE), Median Absolute Error (MedAE), and coefficient of determination (R 2 ). The obtained results indicate that the proposed hybrid model based on sentiments analysis and combined feature selection with MSE value of 0.001 and R 2 value of 0.98 provides better estimations with more minor errors regarding Bitcoin price. This proposed model can also be employed as an individual assistant for more informed trading decisions associated with Bitcoin.
This study examines whether Islamic gold-backed cryptocurrencies (Onegram and X8X) provide any diversification benefits to the Islamic investors of Indonesia. We study the co-movements between return and volatility of cryptocurrencies and Indonesian Islamic equity indices during the pre-COVID-19 and COVID-19 periods. We employ Multivariate Generalized Autoregressive Conditional Heteroscedastic-Dynamic Conditional Correlation (M-GARCH-DCC) and Continuous Wavelet Transforms (CWT) for this study. We find that the COVID-19 crisis enhanced the spillover effect among the Islamic gold-backed cryptocurrencies and Islamic equities. We also provide evidence that Indonesian investors may invest in cryptocurrencies to minimize the equity sector risks during the pandemic. Our results bear significant implications for portfolio diversification strategies for Indonesian investors.
The traditional identification method of the digital bond trading contract system is single chain identification. Due to a large number of digital bond trading users, single chain identification will affect the performance of the system in the trading process. Therefore, we aim to propose a digital bond trading contract system based on blockchain technology and study the system architecture. In this paper, the architecture of the bond trading system is optimized by using the technology of blockchain, and the software architecture is developed by using the technology of blockchain. Through the performance test of the system, the advantages of the system architecture in practical application are verified. Firstly, the overall hardware architecture of the system is designed, and the internal structure of the controller in the control layer is redeployed. Combining the Mork tree and Patricia tree, the data structure based on blockchain and the block header data structure of the transaction contract system are optimized. By analyzing the operation process of the smart contract, the process of the smart contract consensus algorithm is optimized. The system performance test results show that the performance of the designed system in system transaction throughput, transaction delay, and system security is better than the traditional system, which verifies the effectiveness and reliability of the designed system.
The goal of this project is to develop a system that can predict the price of a cryptocurrency (Bitcoin) based on the sentiment of the input provided. This input will be supplied to the model using the Cryptopanic API, which will extract the latest news related to Bitcoin. These technological advancements can help us make accurate predictions thereby facilitating investments. We have tried to accomplish this by using a series of deep learning techniques and methodologies. Our decision to build this model using LSTM was based on the comparison of results between other algorithms like CNN (Convolutional Neural Network), GRU (Gated Recurrent Unit) and RNN (Recurrent Neural Network). Unlike technical analysis methods which are used for normal stock market prediction we have built a model which will be trained to classify news headlines based on the sentiment detected and give a predicted price. We believe that the use of LSTM to give accurate price prediction would be extremely useful for novice as well as professional Bitcoin traders. Also, it has been proven that public sentiments have been very influential in determining the price of Bitcoin and thus taking that into consideration would improve our understanding and prediction.
Muhammad Husaini Bin Mohd Sabri, Amgad Muneer, Shakirah Mohd Taib
Machine learning has become the backbone of bitcoin portfolio optimization in today's technological era. This research applies a deep neural network (DNN) model, Long Short-Term Memory (LSTM), to historical bitcoin prices and Sentiment Analysis to tweet data gathered from Twitter. The LSTM algorithm is used to train the model and forecast the future cryptocurrency price. Sentiment analysis, on the other hand, examines sentiment on Twitter to determine the relationship between sentiment and cryptocurrency price fluctuations. Sentiment analysis categorizes Twitter sentiment as positive or negative, and the fraction of positive and negative tweets is used to forecast bitcoin price fluctuations. The predicted price fluctuation data is then added to the LSTM predicted price to predict the new price for the next time frame. Finally, both models forecast future cryptocurrency prices and patterns, particularly Bitcoin.
Abstract This research makes the first attempt to design, optimize and use average true range (ATR)‐based trading systems for five popular cryptocurrencies. We used particle swarm optimization procedures to optimize systems with multiple objectives that are based on the ATR concept. Our aim was to determine the best configurations for each system that would maximize net profits, the profit factor, and the percentage of profitable trades. We demonstrate that the ATR‐based systems can predict the price trends of the examined cryptocurrencies. Our results also indicate that optimized Keltner Channel‐based systems improve the ability of the stand‐alone optimized ATR systems to forecast trends, net profits, and the profit factor. Finally, both systems perform better for long trades than for short trades.
Cryptocurrencies are becoming a well-known and commonly acknowledged kind of substitute trade money. Most monetary businesses now include cryptocurrency. Accordingly, cryptocurrency trading is widely regarded as the most of prevalent and capable types of lucrative investments. However, because this financial sector is already known for its extreme volatility and quick price changes, over brief periods of time. For such constantly changing nature of crypto trends and price, it has become a necessary part for traders and crypto enthusiast to get a detailed analysis before investing. Also, the construction of a precise and dependable forecasting model is regarded vital for portfolio management and optimization. In this paper we propose a web system, which will help to understand cryptocurrency in a more statistical way. Proposed system focuses mainly on four coins : Bitcoin, Ethereum, Dogecoin and Shiba Inu performing analysis and forecasting on all the four coins. System will also do statistical comparison between the coins. Analysis and comparison is carried out using python libraries and modules whereas LSTM and ARIMA are used for forecasting. Extensive research was conducted using real-time and historical information, on four key cryptocurrencies, two of which had the greatest market capitalization, notably Bitcoin and Ethereum, while the other, Dogecoin and Shiba Inu, that had a significant growth in market capitalization over the previous year. In comparison to old fully-connected deep neural networks, the suggested model may employ mixed crypto data more proficiently, minimizing overfitting and computing costs.
Cryptocurrencies have received much attention amongst investors and policymakers due to the innovative features and simplicity. However, prices of the cryptocurrencies are nonlinear and volatile, which creates challenges for the investors to forecast the cryptocurrency prices. The present study takes the price data of two important cryptocurrencies, i.e., Bitcoin and Ripple, for 2013 to 2020. The study presents the forecasting accuracy of statistical models such as random walk (RW) and autoregressive integrated moving average (ARIMA), and machine learning models such as artificial neural network (ANN) and ensemble model. The study develops the ensemble of RW, ARIMA, and ANN. The study compares the predictive power of all the models and demonstrates that the forecasting accuracy of the ensemble model is better than all the component models, i.e., RW, ARIMA, and ANN. The results of the study have several implications for investors, traders, and policymakers.
In the last decade, cryptocurrency trading has attracted the attention of private and professional traders and investors. To forecast the financial markets, algorithmic trading systems based on Artificial Intelligence (AI) models are becoming more and more established. However, they suffer from the lack of transparency, thus hindering domain experts from directly monitoring the fundamentals behind market movements. This is particularly critical for cryptocurrency investors, because the study of the main factors influencing cryptocurrency prices, including the characteristics of the blockchain infrastructure, is crucial for driving experts’ decisions. This paper proposes a new visual analytics tool to support domain experts in the explanation of AI-based cryptocurrency trading systems. To describe the rationale behind AI models, it exploits an established method, namely SHapley Additive exPlanations, which allows experts to identify the most discriminating features and provides them with an interactive and easy-to-use graphical interface. The simulations carried out on 21 cryptocurrencies over a 8-year period demonstrate the usability of the proposed tool.
This study aims to establish the model of the cryptocurrency price trend based on a financial theory using the Long Short-Term Memory (LSTM) networks model with multiple combinations between the window length and the predicting horizons. The Random Walk model is also applied with different parameter settings. The object of this study is the cryptocurrency and medical issues, primarily the Bitcoin and Ethereum and the COVID-19. Quantitative analysis is adopted as the method of this dissertation. The research tool is Python programming language, and the TensorFlow package is employed to model and analyze research topics. The results of this study show the limitations of the LSTM and Random Walk model for price prediction while demonstrating the different characteristics of both models with different parameter settings, providing a balance between the model's accuracy and the model's practicality.
With the prosperity of the financial market, more and more people are involved in securities trading. How to combine investment in bitcoin and gold to achieve the highest profit is one of the issues that market traders think about. To solve this problem, we build a model that predicts future prices in order to better help investors. We constructed an ARIMA model through differential stationarity processing, AEC, white noise test and other methods, and used the data of the current day and the previous day to predict the price of the next day. At the same time, we use the model to predict the average for the next N days. If it is predicted that the price of the asset will decrease in the future, it will be sold on the same day. If the price of the asset will increase in the future, and the income obtained is greater than the transaction cost and the expected income of the investor, the purchase will continue until the end. Last until the last day $323841.52. Finally, it can be verified that the accuracy of the ARIMA model is the highest by comparing other mainstream machine learning models. In other words, the ARIMA model is the best strategy for this problem.
Mustafa Shabbir Bhavanagarwala, K N Nagarjun, Tanzim Abbas Charolia, Mr Sonawane Vishal · 5 authors
In our project, the point is to anticipate long term esteem of the money related stocks of a company and crypto coins individually with fine precision. The future prices of stock and cryptocurrency are predicted by using the past available values. “Buy low, sell high" is a good saying but it is not a good choice for making speculations. Investment is best stock or crypto currency in awful time can have bad results, while investment in best stock or cryptocurrency at right time can have best benefits. Prediction for long term values is easy as compared to day-to-day basis as prices fluctuate a lot. So, our model predicts the price of stocks and cryptocurrencies, which helps the investors to invest in appropriate stocks and cryptocoins. The dataset used is taken from yahoo finance and twelve data using web scraping. The dataset retrieved is in raw format. It consists of collection of values of stock market data of various companies, and also data of various cryptocurrencies. First, raw data is converted into processed data, which is done using feature extraction. Then the dataset is splitted into training and test sets. We use the training dataset to train the model, and use test dataset to predict the future prices of stocks and cryptocurrencies. Now user can gain best knowledge about stock price trends of various companies and also cryptocurrency price trends, and can decide on for best investments in respective fields and gain best benefits.
Cryptocurrency is an intangible asset that has emerged due to the advancement in the fin-tech space and has led to many research opportunities. Cryptocurrency prices are subject to significant volatility, and predicting trends is very difficult. Around the world, there are hundreds of cryptocurrencies that are in daily use for purchasing goods and services or investing. Although Machine Learning models have shown satisfactory results in predicting cryptocurrency prices, it is necessary to consider the public sentiment regarding a cryptocurrency. Public opinion about crypto is most prevalent on Twitter. Therefore, we aim to classify Bitcoin investing better by combining Machine Learning Algorithm LSTM with Sentimental Analysis, which an investor can use to decide whether to buy a cryptocurrency or not.
As a popular cryptocurrency, Bitcoin has been an important investment tool in recent years. This study aims to analyze the factors that affect the Bitcoin price to help investors make better investment choices. Applying the simple linear regression model and Granger causality test to the data from January 2015 to December 2021, the research first examines the stationary of the data and then studies the relationships between Bitcoin price and other factors including Dow Jones Industrial Index, U.S. currency in circulation, U.S. disposable income. According to the result, all three factors have a positive effect on the price of Bitcoin and the Bitcoin price will in turn influence the Dow Jones Industrial Index and U.S. disposable income. This finding helps explain how certain economic indicators and Bitcoin prices interact. Since investment is always risky, investors must consider certain factors like the trend of DOW, M2, or PCI in advance to make a reasonable investment decision.