This study aims to assess the extent to which blockchain technology (BCT) may constitute an alternative to the conventional stock trading system and emphasize the changing roles of the key parties. It is expected that BCT would enhance the performance of the process across the three stages (i.e. trading, clearing and settlement within the stock exchange environment). A thorough literature review is conducted to understand the BCT performance modeling techniques and approaches (empirical and analytical) and to examine the theoretical potentials and capabilities of BCT in the financial markets. The case study and simulation methods are used to evaluate the impact of BCT implementation in optimizing the process of trading, clearing and settlement in Abu Dhabi Securities Exchange (ADX) stock-trading activities. This paper presents a simulation analysis comparing a blockchain system with a traditional trading system in the context of stock market. The simulation procedures involve modeling processes over different durations and transaction volumes, using metrics such as process time and cycle time to evaluate performance. The performance index combines these metrics with weights to ensure accurate and consistent measurements. Simulation results reveal that the blockchain system significantly outperforms the current trading system, especially at higher transaction volumes, highlighting its scalability and efficiency. A threshold of 30,000 transactions is identified as the point where blockchain’s benefits become apparent. The analysis shows that blockchain significantly elevate the process efficiency. It reduces both cycle time and process time across varying transaction volumes, maintaining consistency and reliability. Additionally, a simple simulation using the Hyperledger Fabric platform demonstrates the practical implementation of a permissioned blockchain for clearing transactions, emphasizing the system's capability to manage high transaction volumes efficiently and securely. The use of blockchain network for handling seamless transactions using pre-defined smart contracts significantly improves the performance of the stock trading processes, specifically in the clearing phase. Interestingly, the BCT system drops the need for a “third party” (i.e. stock custodian) across the three stages. At the end of the paper, we propose a thereat mitigating model for stock trading with a new blockchain system.
The COVID-19 pandemic has had a significant impact on various aspects of life, including financial markets. The cryptocurrency market, already renowned for its volatility, experienced a surge in activity and significant changes in investor behavior during this period. This research aims to analyze various economic behavioral anomalies that emerged in cryptocurrency transactions during the COVID-19 pandemic. This research uses a literature study method to identify several dominant behavioral anomalies, such as FOMO (Fear of Missing Out), Herding Behavior, Noise Trading, Overconfidence, and Anchoring Bias. These behavioral anomalies trigger extreme market volatility, asset bubbles, and financial losses for investors. This research highlights the importance of investor education, market regulation, and technology development to minimize the impact of behavioral anomalies and protect investors in the cryptocurrency market.
The price of cryptocurrency can be affected by several factors these years, such as technology, social media, COVID-19, etc. One of the examples of these factors is Elon Mask’s tweets about cryptocurrency, which help to increase cryptocurrency prices. With the spread of the epidemic, people are restricted from meeting in person. Therefore, more and more people are active on online social media sites such as Twitter. This research wants to determine if tweets related to cryptocurrency (Bitcoin, one of the most popular cryptocurrencies nowadays) affect price. By taking 5 machine learning models and the Granger causality test, the correlation and causation relationship between sentiment analysis and bitcoin price can be determined.
One of the financial assets in currency exchange is now cryptocurrency. The public is drawn to cryptocurrency trading because it is considered a lucrative form of investing. For cryptocurrency investors to maximize their earnings, accurate price forecasting is crucial. As price forecasting involves time series analysis, a hybrid deep learning model is suggested to project cryptocurrency prices in the future. Long Short-Term Memory and Gated Recurrent Unit (LSTM-GRU) networks are integrated into the hybrid model. Three cryptocurrency datasets are evaluated using the suggested hybrid model: Ethereum, Ripple, and Bitcoin. According to experimental results, the suggested LSTM-GRU model may provide the lowest MSE and RMSE values on the Bitcoin dataset (0.0611 and 0.2472), the Ethereum dataset (0.0369 and 0.19222), and the Ripple dataset (0.0006 and 0.0247).
Given that cryptocurrencies are now involved in nearly every financial transaction due to their widespread acceptance as an alternative method of payment and currency exchange, researchers and economists have increased opportunities to analyze cryptocurrency prices. Over time, predicting the daily closing price of Ethereum has been challenging for investors, traders, and investment banks because of its significant price volatility. The daily closing price of cryptocurrency is crucial for trading or investing in Ethereum. This report aims to conduct a comparative analysis of the predictive performance of deep machine learning algorithms within a stacking ensemble modeling framework, utilizing daily historical price data of Ethereum from Coindesk, tweets from Twitter spanning from August 1, 2022, to August 8, 2022, and five additional covariates (closing price lag1, closing price lag2, noltrend, daytype, and month) derived from Ethereum's closing price. Seven models are employed to forecast the daily closing price of Ethereum: recurrent neural network, ensemble stacked recurrent neural network, gradient boosting machine, generalized linear model, distributed random forest, deep neural networks, and a stacked ensemble of gradient boosting machine, generalized linear model, distributed random forest, and deep neural networks. The primary evaluation metric is the mean absolute error (MAE). Based on MAE, the RNN forecasts outperform the other models in this study, achieving an MAE of 0.0309.
Decarbonization and digitalization play critical roles in the drive towards a more sustainable energy sector. Energy provenance, or ability to track emissions and meet standards, has become increasingly important. To address these challenges, the European Union (EU) has implemented the European Energy Certificate System (EECS), but the existing certificate trading process remains cumbersome, bureaucratic, and prone to errors. As an alternative solution, this paper presents a decentralized Guarantees of Origin (GO) trading platform based on Ethereum, which integrates marketplace and transfer routes for energy certificates. Utilizing smart contracts for a double auction, certificates are traded as quasi-unique Semi-Fungible Token (SFT), offering a more transparent and traceable process. This is implemented alongside a power simulation representing the production side of the certificate market. Thus, two models are deployed in a co-simulation with the goal of showcasing a proof-of-concept for such a decentralized marketplace.
As of December 2021, the cryptocurrency market had a market value of over US$270 billion, and over 5,700 types of cryptocurrencies were circulating among 23,000 online exchanges. Reinforcement learning (RL) has been used to identify the optimal trading strategy. However, most RL-based optimal trading strategies adopted in the cryptocurrency market focus on trading one type of cryptocurrency, whereas most traders in the cryptocurrency market often trade multiple cryptocurrencies. Therefore, the present study proposes a method based on deep Q-learning for identifying the optimal trading strategy for multiple cryptocurrencies. The proposed method uses the same training data to train multiple agents repeatedly so that each agent has accumulated learning experiences to improve its prediction of the future market trend and to determine the optimal action. The empirical results obtained with the proposed method are described in the following text. For Ethereum, VeChain, and Ripple, which were considered to have an uptrend, a horizontal trend, and a downtrend, respectively, the annualized rates of return were 725.48%, −14.95%, and − 3.70%, respectively. Regardless of the cryptocurrency market trend, a higher annualized rate of return was achieved when using the proposed method than when using the buy-and-hold strategy.
Gerasimos Vonitsanos, Andreas Kanavos, Foteini Grivokostopoulou, Spyros Sioutas
Statistical models, enhanced by deep learning techniques, have become pivotal in various predictive tasks, including financial forecasting. This paper addresses the challenge of predicting cryptocurrency prices, utilizing a dataset comprising various cryptocurrencies treated as time series. We employ several deep neural network architectures-including Multilayer Perceptrons (MLP), Recurrent Neural Networks (RNN), Long Short-Term Memory networks (LSTM), and Bidirectional LSTM networks-to forecast future prices. Our methodology involves a detailed analysis of cryptocurrency time series to inform the design of these networks. The performance of each model is rigorously compared, highlighting their predictive capabilities in the context of cryptocurrency markets. This study not only contributes to the empirical literature by applying advanced neural networks to high-volume financial data but also provides a comparative analysis that may guide future applications of deep learning in economic forecasting.
Melike Bildirici, Yasemen Uçan, Ramazan Tekercioğlu
This paper introduces hybrid models designed to analyze daily and weekly bitcoin return spanning the periods from 18 July 2010 to 28 December 2023 for daily data, and from 18 July 2010 to 24 December 2023 for weekly data. Firstly, the fractal and chaotic structure of the selected variables was explored. Asymmetric Cantor set, Boundary of the Dragon curve, Julia set z2 −1, Boundary of the Lévy C curve, von Koch curve, and Brownian function (Wiener process) tests were applied. The R/S and Mandelbrot–Wallis tests confirmed long-term dependence and fractionality. The largest Lyapunov test, the Rosenstein, Collins and DeLuca, and Kantz methods of Lyapunov exponents, and the HCT and Shannon entropy tests tracked by the Kolmogorov–Sinai (KS) complexity test determined the evidence of chaos, entropy, and complexity. The BDS test of independence test approved nonlinearity, and the TeraesvirtaNW and WhiteNW tests, the Tsay test for nonlinearity, the LR test for threshold nonlinearity, and White’s test and Engle test confirmed nonlinearity and heteroskedasticity, in addition to fractionality and chaos. In the second stage, the standard ARFIMA method was applied, and its results were compared to the LieNLS and LieOLS methods. The results showed that, under conditions of chaos, entropy, and complexity, the ARFIMA method did not yield successful results. Both baseline models, LieNLS and LieOLS, are enhanced by integrating them with deep learning methods. The models, LieLSTMOLS and LieLSTMNLS, leverage manifold-based approaches, opting for matrix representations over traditional differential operator representations of Lie algebras were employed. The parameters and coefficients obtained from LieNLS and LieOLS, and the LieLSTMOLS and LieLSTMNLS methods were compared. And the forecasting capabilities of these hybrid models, particularly LieLSTMOLS and LieLSTMNLS, were compared with those of the main models. The in-sample and out-of-sample analyses demonstrated that the LieLSTMOLS and LieLSTMNLS methods outperform the others in terms of MAE and RMSE, thereby offering a more reliable means of assessing the selected data. Our study underscores the importance of employing the LieLSTM method for analyzing the dynamics of bitcoin. Our findings have significant implications for investors, traders, and policymakers.
The study explores diverse AI methodologies employed in the cryptocurrency domain, focusing on their applications in key areas such as price prediction, sentiment analysis, market trend analysis, volatility prediction, trading strategy optimization, fraud detection, and portfolio management. Various machine learning models, including regression, neural networks, and reinforcement learning, are investigated for their effectiveness in predicting cryptocurrency prices and optimizing trading strategies. The integration of Natural Language Processing (NLP) techniques is discussed in the context of sentiment analysis, where AI algorithms analyze vast amounts of textual data from social media, news articles, and online forums to gauge market sentiment and its potential impact on cryptocurrency prices. Additionally, the paper examines the role of AI in identifying patterns, trends, and anomalies in market data, facilitating effective decision-making for traders and investors. However, the paper emphasizes the need for caution, acknowledging the inherent uncertainties and risks associated with cryptocurrency investments. It concludes by highlighting the potential for continued advancements in AI applications, contributing to a deeper understanding of cryptocurrency market dynamics and aiding in more informed decision-making in this rapidly evolving financial landscape.
Cryptocurrency is a form of decentralized digital currency. Ethereum is the second-largest cryptocurrency by market capitalization and the largest altcoin. Cryptocurrencies including Ethereum are highly volatile. Hence, shortterm directional forecasts in the cryptocurrency market have become a widely discussing topic. Candlestick charts are useful visualizations of the open, high, low and close prices which can identify patterns and gauge the near-term direction of prices. This research explores the effectiveness of forecasting hourly Ethereum closing price direction based on candlestick charts within a short time horizon. The proposed forecasting algorithm incorporates clustering methods such as fuzzy K-means, K-means and partition around medoids clustering to cluster candlestick chart properties namely upper shadow length, body length and lower shadow length. Classification methods such as random forest, support vector machine and K-nearest neighbour were used to forecast closing price direction using 16 different predictor variable sets including open, high, low and close prices, candlestick chart price direction, USL, BL and LSL. The accuracy for all considered cases was around 50%. Clustering improved the accuracy slightly and including the CPD with the predictor variable sets under consideration can increase the accuracy slightly. However, this approach is performing better in predicting the Down cases to the total number of actual Down cases because there is a higher sensitivity of 81.20% based on the SVM with Open, High, Low and Close at t in the clustering ignored method.
Neural networks (NNs) are well established and widely used in time series forecasting due to their frequent dominance over other linear and nonlinear models. Thus, this paper does not question their appropriateness in forecasting cryptocurrency prices; rather, it compares the most commonly used NNs, i.e. feedforward neural networks (FFNNs), long short-term memory (LSTM) and convolutional neural networks (CNNs). This paper contributes to the existing literature by defining the appropriate NN structure comparable across different NN architectures, which yields the optimal NN model for Bitcoin return forecasting. Moreover, by incorporating turbulent events such as COVID and war, this paper emerges as a stress test for NNs. Finally, inputs are carefully selected, mostly covering macroeconomic and market variables, as well as different attractiveness measures, the importance of which in cryptocurrency forecasting is tested. The main results indicate that all NNs perform the best in an environment of bullish market, where CNNs stand out as the optimal models for continuous dataset, and LSTMs emerge as optimal in direction forecasting. In the downturn periods, CNNs stand out as the best models. Additionally, Tweets, as an attractiveness measure, enabled the models to attain superior performance.
This paper examines the efficiency and asymmetric multiracial features of NFTs (Mana, Tezos), and traditional assets (EGX30, Oil index) using Asymmetric Multiracial Cross-Correlations Analysis covering the period from January 2020 to May 2021. Considering the full sample with a significant variation among asset classes. (Oil-Tezos)and (Mana-Tezos) is the most efficient.Since their inception, the blockchain-based digital asset classes have received immense interest from investors and portfolio managers as an alternative investment platform. Along with other established traditional cryptocurrencies such as Bitcoin, Litecoin, Ripple, and Ethereum, new blockchain asset classes such as Decentralized Finance (DeFi) and Non-Fungible Tokens (NFTs) have made a considerable contribution tothe asset market’s recent expansion (Aharon & Demir, 2021; Alam, Chowdhury, Abdullah, & Masih, 2023; Maouchi, Charfeddine, & el Montasser, 2021; Yousaf & Yarovaya, 2022).Fundamentally, NFTs and DeFi differ from traditional cryptocurrencies as they are not virtual currency. Where NFTs are non-transferable cryptographic digital assetscreated by Ethereum smart contracts and can be sold and traded, the interchangeability of NFTs when comparing the other cryptocurrencies is very low (Karim, Lucey, Naeem, & Uddin, 2022; Q. Wang, Li, Wang, & Chen, 2021; Y. Wang, 2022).The NFTs and DeFi are relatively contemporary and unexplored asset classes, but their market capitalization has grown substantially as risk minimizing assets, particularly during the COVID-19 period. In the NFT space, the KeywordsVolatility, NFTs, Traditional Financial Assets, and MGARCH
With particular attention to variables like volatility and the performance of the U.S. stock market, this study attempts to conduct a thorough quantitative examination of the relationship between the cryptocurrency market. By using sophisticated mathematical modeling approaches, such as regression analysis and correlation methodologies, it is hoped to identify the key characteristics of these markets as well as the degree to which cryptocurrency volatility and stock market success are causally related. The use of historical data, spanning a specific time (from July 1st, 2019, to July 1st, 2023) around 4 index price-day transaction data will be made, with a focus on high-frequency data for improved accuracy. The results of this study, which examine each option's characteristics or attributes, will add to the larger body of scholarly literature on the integration of cryptocurrencies into conventional financial markets. Moreover, drawing conclusions about some effects or prospective connections between cryptocurrencies and the financial industry based on their similarity to the American stock market. To pave the way for better-informed financial decision-making, this research aims to deepen our understanding of the interactions and spillover effects between cryptocurrency volatility and the American stock market.
Zahra Ghorrati, Kourosh Shahnazari, Ahmad Esmaeili, Eric T. Matson
One of the simplest approach in Reinforcement Learning (RL) is updating Q-table using Bellman operator. While theoretical expectations hint at the potential convergence achieved by modeling the discrete Q-table with the Bellman operator, practical limitations surface in real-world scenarios. The main challenges associated with it include the exponential growth of the Q-table size with an increasing number of state dimensions and the inability to use the Q-table in continuous state spaces. Alternative approaches, such as employing neural networks to approximate the parameterized Q-function, may not necessarily result in convergence.In response to these challenges, this paper introduces an simple innovative methodology inspired by the Bellman method updating. The proposed method utilizes fuzzy rules to discretize the state space, leading to the direct use of the Bellman operator for updating the fuzzy neural network weights, effectively acting as the Fuzzy Q-table. Instead of approximating the Q-function utilizing neural network/deep neural network based on gradient approaches, the proposed method establishes a Fuzzy Q-table and updates it using the Bellman equation. This strategic decision helps to solve the convergence problem in addition to prevent entrapment in local minima problems, a common challenge faced by conventional gradient methods. The efficacy of the proposed approach is demonstrated through its application to trading in the Bitcoin Futures Market, showcasing its ability to navigate complexities and uncertainties. Beyond financial markets, this methodology presents a versatile solution applicable to a diverse range of reinforcement learning problems, addressing limitations faced by traditional Q-tables or DQN.
Financial markets are increasingly interlinked. Therefore, this study explores the complex relationships between the Tadawul All Share Index (TASI), West Texas Intermediate (WTI) crude oil prices, and Bitcoin (BTC) returns, which are pivotal to informed investment and risk-management decisions. Using copula-based models, this study identified Student’s t copula as the most appropriate one for encapsulating the dependencies between TASI and BTC and between TASI and WTI prices, highlighting significant tail dependencies. For the BTC–WTI relationship, the Frank copula was found to have the best fit, indicating nonlinear correlation without tail dependence. The predictive power of the identified copulas were compared to that of Long Short-Term Memory (LSTM) networks. The LSTM models demonstrated markedly lower Root Mean Squared Error (RMSE), Mean Absolute Error (MAE), and Mean Absolute Scaled Error (MASE) across all assets, indicating higher predictive accuracy. The empirical findings of this research provide valuable insights for financial market participants and contribute to the literature on asset relationship modeling. By revealing the most effective copulas for different asset pairs and establishing the robust forecasting capabilities of LSTM networks, this paper sets the stage for future investigations of the predictive modeling of financial time-series data. The study highlights the potential of integrating machine-learning techniques with traditional econometric models to improve investment strategies and risk-management practices.
Mamun Ahmed, Sayma Alam Suha, Fahamida Hossain Mahi, Forhad Uddin Ahmed
Since entering the market in 2009, Bitcoin has had a price that is extremely erratic. Its price is influenced by factors such as adoption rates, regulatory changes, geopolitical occurrences, and macroeconomic developments. Experts believe that Bitcoin's price will rise in the long run due to limited supply and rising demand. Therefore, the aim of this study is to propose an ensemble feature selection and machine learning-based approach to predict bitcoin price. For this research purpose, the cryptocurrency-based dataset has been used, visualized, and preprocessed. Five different feature selection approaches (Pearson, RFE, Embedded Random Forest, Tree-based and Light GBM) are followed by ensemble methodology, with the maximum voting approach to extract the most significant features and generate a dataset with reduced attributes. Then the dataset with or without feature selection is used for bitcoin price prediction by applying ten different machine learning regressing models, which includes six traditional, four bagging and boosting ensemble techniques. The comparative result analysis through multiple performance parameters reveals that the decreased number of features improves the performance for each of the models and the ensemble models outperform other types of models. Therefore, Random Forest regression ensemble ML model can get the best prediction accuracy with 0.036018 RMSE, 0.029470 MAE and 0.934512 R2 employing the dataset with reduced features for estimating the value of bitcoin.
In the rapidly evolving domain of cryptocurrency trading, accurate market data analysis is crucial for informed decision making. Candlestick patterns, a cornerstone of technical analysis, serve as visual representations of market sentiment and potential price movements. However, the sheer volume and complexity of cryptocurrency price time-series data presents a significant challenge to traders and analysts alike. This paper introduces an innovative rule-based methodology for recognizing candlestick patterns in cryptocurrency markets using Python. By focusing on Ethereum, Bitcoin, and Litecoin, this study demonstrates the effectiveness of the proposed methodology in identifying key candlestick patterns associated with significant market movements. The structured approach simplifies the recognition process while enhancing the precision and reliability of market analysis. Through rigorous testing, this study shows that the automated recognition of these patterns provides actionable insights for traders. This paper concludes with a discussion on the implications, limitations, and potential future research directions that contribute to the field of computational finance by offering a novel tool for automated analysis in the highly volatile cryptocurrency market.
n the rapidly evolving world of cryptocurrency markets, the precise forecasting of Bitcoin's value against the US Dollar acquires paramount importance, catering to the interests of diverse stakeholders including investors, regulatory agencies, and academia. This study ventures into a comprehensive assessment of various time series forecasting methodologies, including but not limited to Random Forest Regression, ARIMA, Linear Regression, and XGBoost. Notably, our investigation unveils a pivotal revelation: the foundational models like Linear Regression and Random Forest Regression, traditionally con-sidered less complex, not only contend but also surpass the forecast accuracy of ARIMA models in the realm of Bitcoin. This paper aims to demystify the underpinnings of this superior performance, especially in mitigating the inherent volatility and unpredictability characteristic of Bitcoin. Our findings herald a transformative perspective in financial time series forecasting, potentially reshaping investment strategies and predictive analytics in the digital currency landscape.n the rapidly evolving world of cryptocurrency markets, the precise forecasting of Bitcoin's value against the US Dollar acquires paramount importance, catering to the interests of diverse stakeholders including investors, regulatory agencies, and academia. This study ventures into a comprehensive assessment of various time series forecasting methodologies, including but not limited to Random Forest Regression, ARIMA, Linear Regression, and XGBoost. Notably, our investigation unveils a pivotal revelation: the foundational models like Linear Regression and Random Forest Regression, traditionally considered less complex, not only contend but also surpass the forecast accuracy of ARIMA models in the realm of Bitcoin. This paper aims to demystify the underpinnings of this superior performance, especially in mitigating the inherent volatility and unpredictability characteristic of Bitcoin. Our findings herald a transformative perspective in financial time series forecasting, potentially reshaping investment strategies and predictive analytics in the digital currency landscape.I
In recent years, cryptocurrencies have received substantial attention from investors, researchers and the media due to their volatile behaviour and potential for high returns. This interest has led to an expanding body of research aimed at predicting cryptocurrency prices, which are notably influenced by a wide array of technical, sentimental, and legal factors. This paper reviews scholarly content from 2014 to 2024, employing a systematic approach to explore advanced quantitative methods for cryptocurrency price prediction. It encompasses a broad spectrum of predictive models, from early statistical analyses to sophisticated machine and deep learning algorithms. Notably, this review identifies and discusses the integration of emerging technologies such as Transformers and hybrid deep learning models, which offer new avenues for enhancing prediction accuracy and practical applicability in real-world scenarios. By thoroughly investigating various methodologies and parameters influencing cryptocurrency price predictions, including market sentiment, technical indicators, and blockchain features, this review highlights the field’s complexity and rapid evolution. The analysis identifies significant research gaps and under-explored areas, providing a foundational guideline for future studies. These guidelines aim to connect theoretical advancements with practical, profit-driven applications in cryptocurrency trading, ensuring that future research is both innovative and applicable.