The accurate imputation of missing values in time series data is paramount for maintaining the integrity and reliability of analyses and predictions. This article investigates the effica-cy of various missing values imputation methods, encom-passing well-known machine learning and statistical tech-niques. Moreover, for a better understanding, they imple-mented two financial data time series: S&P 500 and Bitcoin markets spanning from 2016 to 2023 on a daily frequency. Initially utilizing complete datasets, controlled missingness was introduced by randomly removing 45 data points. Then, these methods applied multiple imputation strategies for estimating and substituting these missing values. Experi-mental evaluation yielded insightful findings regarding the performance of the different methods. The examined ma-chine learning methods, including k-Nearest Neighbors (k-NN), Random Forest, Deep Learning, and Decision Trees, consistently outperformed their statistical counterparts, such as Mean Imputation, Regression Imputation, Hot-Deck Im-putation, and Expectation-Maximization Imputation. Nota-bly, Random Forest emerged as the most effective method, showcasing superior performance in terms of accuracy and robustness. Conversely, the Mean Imputation method exhibited com-paratively inferior outcomes, suggesting its limited suitabil-ity for financial time series data. This research contributes to the ongoing discourse on data integrity within finance ana-lytics and serves as a comprehensive guide for practitioners seeking optimal missing values imputation methods. The empirical evidence provided herein advances the under-standing of imputation techniques' relative performance and their application in financial data, facilitating enhanced de-cision-making processes and yielding more reliable predic-tions.
Fatih Ecer, Tolga Murat, Hasan Dınçer, Serhat Yüksel
Abstract Crypto assets have become increasingly popular in recent years due to their many advantages, such as low transaction costs and investment opportunities. The performance of crypto exchanges is an essential factor in developing crypto assets. Therefore, it is necessary to take adequate measures regarding the reliability, speed, user-friendliness, regulation, and supervision of crypto exchanges. However, each measure to be taken creates extra costs for businesses. Studies are needed to determine the factors that most affect the performance of crypto exchanges. This study develops an integrated framework, i.e., fuzzy best–worst method with the Heronian function—the fuzzy measurement of alternatives and ranking according to compromise solution with the Heronian function (FBWM’H–FMARCOS’H), to evaluate cryptocurrency exchanges. In this framework, the fuzzy best–worst method (FBWM) is used to decide the criteria’s importance, fuzzy measurement of alternatives and ranking according to compromise solution (FMARCOS) is used to prioritize the alternatives, and the Heronian function is used to aggregate the results. Integrating a modified FBWM and FMARCOS with Heronian functions is particularly appealing for group decision-making under vagueness. Through case studies, some well-known cryptocurrency exchanges operating in Türkiye are assessed based on seven critical factors in the cryptocurrency exchange evaluation process. The main contribution of this study is generating new priority strategies to increase the performance of crypto exchanges with a novel decision-making methodology. “Perception of security,” “reputation,” and “commission rate” are found as the foremost factors in choosing an appropriate cryptocurrency exchange for investment. Further, the best score is achieved by Coinbase, followed by Binance. The solidity and flexibility of the methodology are also supported by sensitivity and comparative analyses. The findings may pave the way for investors to take appropriate actions without incurring high costs.
Automated Machine Learning (AutoML) tools are revolutionizing the field of machine learning by significantly reducing the need for deep computer science expertise. Designed to make ML more accessible, they enable users to build high-performing models without extensive technical knowledge. This study delves into these tools in the context of time series analysis, which is essential for forecasting future trends from historical data. We evaluate three prominent AutoML tools—AutoGluon, Auto-Sklearn, and PyCaret—across various metrics, employing diverse datasets that include Bitcoin and COVID-19 data. The results reveal that the performance of each tool is highly dependent on the specific dataset and its ability to manage the complexities of time series data. This thorough investigation not only demonstrates the strengths and limitations of each AutoML tool but also highlights the criticality of dataset-specific considerations in time series analysis. Offering valuable insights for both practitioners and researchers, this study emphasizes the ongoing need for research and development in this specialized area. It aims to serve as a reference for organizations dealing with time series datasets and a guiding framework for future academic research in enhancing the application of AutoML tools for time series forecasting and analysis.
Bitcoin is known for its high volatility, which makes it challenging to accurately predict future prices. In this study, we aim to forecast Bitcoin prices for a month by incorporating exogenous variables, specifically the interest rate and recession probability. Our primary objective is to explore whether these variables have a positive impact on the prediction of Bitcoin prices. We used two popular time series forecasting models: Long Short-Term Memory (LSTM) and Facebook Prophet. Our approach involves exploring the impact of these exogenous variables on the performance of the models and comparing their results through plots and cross-validation. We trained the models using historical Bitcoin price data along with exogenous variables and evaluated their performance on a test dataset. Our results indicate that LSTM outperforms Facebook Prophet in terms of Bitcoin price prediction accuracy. This is because, while Facebook Prophet is optimized for statistical forecasting modeling, LSTM has the capability to learn intricate patterns and relationships given the right architecture with sufficient neurons. Importantly, we demonstrate that incorporating interest rates and recession probabilities significantly enhances the predictive capability of our models. Our findings suggest that changes in interest rates and recession probabilities have an impact on Bitcoin prices, and our models perform better when equipped with this valuable information.
Omar M. Ahmed, Lailan M. Haji, Ayah M. Ahmed, Nashwan M. Salih
The field of finance makes extensive use of real-time prediction of stock price tools, which are instruments that are put to use in the process of creating predictions. In this article, we attempt to predict the price of Bitcoin in a manner that is both accurate and reliable. Deep learning models, as opposed to more traditional methods, are used to manage enormous volumes of data and to generate predictions. The purpose of this research is to develop a method for predicting stock prices using the Hybrid Convolutional Recurrent Model (HCRM) architecture. This model architecture integrates the advantages of two separate deep learning models: The 1-Dimensional-Convolusional Neural Network (1D-CNN) and the Long-Short Term Memory (LSTM). The 1D-CNN is responsible for the feature extraction, while the LSTM is in charge of the temporal regression. The developed 1D-CNN-LSTM model has an outstanding performance in predicting stock values.
Summary This study explores various machine learning and deep learning applications on financial data modelling, analysis and prediction processes. The main focus is to test the prediction accuracy of cryptocurrency hourly returns and to explore, analyse and showcase the various interpretability features of the ML models. The study considers the six most dominant cryptocurrencies in the market: Bitcoin, Ethereum, Binance Coin, Cardano, Ripple and Litecoin. The experimental settings explore the formation of the corresponding datasets from technical, fundamental and statistical analysis. The paper compares various existing and enhanced algorithms and explains their results, features and limitations. The algorithms include decision trees, random forests and ensemble methods, SVM, neural networks, single and multiple features N‐BEATS, ARIMA and Google AutoML. From experimental results, we see that predicting cryptocurrency returns is possible. However, prediction algorithms may not generalise for different assets and markets over long periods. There is no clear winner that satisfies all requirements, and the main choice of algorithm will be tied to the user needs and provided resources.
Karar bilimi karar verme işini kolaylaştırmak ve geliştirmek için eldeki sınırlı bilgiyi kullanarak pek çok teknikten faydalanır. Bu nedenle ekonomi, istatistik, üretim yönetimi ve kontrolü ve psikoloji gibi bilim dallarını da içeren disiplinler arası bir alandır. Sürekli olarak karşı karşıya kalınan karar verme durumu neticesinde verilen kararlar ve sonrasında atılan adımlar ise geleceği şekillendirmektedir. Bu nedenle karar biliminin günümüzdeki yeri oldukça önemlidir. Bu çalışmada 2012-2021 yılları içerisinde karar bilimi alanında üretilen bilimsel çıktıların değerlendirilmesi amaçlanmaktadır. Bu amaçla Scopus/SciVal veri tabanı üzerinden ulaşılan 508.220 bilimsel çıktı incelenmiş, yıllara göre bilimsel çıktı sayısı, atıf sayısı, görüntülenme sayısı bilgileri paylaşılmıştır. Dünya genelinde üretilen bilimsel çıktıları kapsayan bu çalışmada karar bilimi alanında en fazla bilimsel çıktının 2021 yılında (95.109) üretildiği ve en fazla bilimsel çıktıya sahip ülkenin Çin (106.752) olduğu sonucuna ulaşılmıştır. Ayrıca en fazla bilimsel çıktıya sahip enstitü/üniversitenin CNRS (10.411) ve en fazla bilimsel çıktıya yer veren derginin “IFIP Advances in Information and Communication Technology” (10.084) olduğu belirlenmiştir. Bilimsel çıktı sayısı dikkate alındığında yapılan çalışmalarda daha çok kurumsal işbirliklerinin tercih edildiği (201.933) ve karar bilimi alanı içerisinde en fazla çalışılan konuların “Bitcoin; Ethereum; Nesnelerin İnterneti” (16473) olduğu sonucuna ulaşılmıştır. Genel olarak yapılan bu çalışma karar bilimi alanında çalışan araştırmacılar için bilgilendirme, değerlendirme ve yönlendirme özelliklerini taşımaktadır.
Richard De Freitas Pinto, Viviane Fagundes de [UNESP] Mattos, Luiz Ricardo Nakamura
Esse trabalho apresenta a modelagem do volume mensal de transações da criptmoeda Ethereum por meio da metodologia de Box-Jenkins, envolvendo as etapas: análise exploratória, identificação, estimação e validação, algumas das quais executadas com a utilização de diferentes técnicas. O modelo encontrado pela modelagem SARIMA (Modelo autoregressivo integrado de médias móveis sazonal) conseguiu descrever o comportamento linear dos dados de forma satisfatória, mas não foi suficiente para descrever o comportamento da série, composta por movimento linear e não linear, sendo melhor representada por um modelo híbrido.
The rapid expansion of the non-fungible token (NFT) market has attracted many investors. However, studies on the NFT price fluctuations have been relatively limited. To date, the machine learning approach has not been used to demonstrate a specific error in NFT sale price fluctuation prediction. The aim of this study was to develop a prediction model for NFT price fluctuations using the NFT trading information obtained from OpenSea, the world’s largest NFT marketplace. We used Python programs to collect data and summarized them as: NFT information, collection information, and related account information. AdaBoost and Random Forest (RF) algorithms were employed to predict the sale price and price fluctuation of NFTs using regression and classification models, respectively. We found that the NFT related account information, especially the number of favorites and activity status of creators, confer a good predictive power to both the models. AdaBoost in the regression model had more accurate predictions, the root mean square error (RMSE) in predicting NFT sale price was 0.047. In predicting NFT sale price fluctuations, RF performed better, which the area under the curve (AUC) reached 0.956. We suggest that investors should pay more attention to the information of NFT creators. We anticipate that these prediction models will reduce the number of investment failures for the investors.
The purpose of this research is to compare ARIMA and Prophet algorithms and find the best algorithm for forecasting bitcoin prices. The dataset is two years historical bitcoin data between February 2019 and 2021. The data is segmented into daily, weekly, and monthly period category. Both algorithms are built into a univariate model that only receive 2 features for training the model. Several ARIMA models is developed for each dataset interval. After that, the parameter of each model will be cross-referenced to each other to obtain the best parameter combination. Meanwhile, Prophet model will be developed using automatic and manual tuning. Then again parameter value of each model will be cross-referenced to each other to obtain the best parameter combination. Evaluation of the training model is done by calculating the Mean Absolute Error (MAE), Mean Absolute Percentage Error (MAPE), Mean Squared Error (MSE), and Rooted Mean Squared Error (RMSE). The results showed that the best model for the daily and weekly data category was the Prophet algorithm, while for the monthly data category was the ARIMA algorithm.
This study uses the API of Upbit, one of Korea’s cryptocurrency exchanges, to predict continuous time series for a limited period and cryptocurrencies using LSTM, a machine learning technique. The trading (buying and selling) point algorithm presented in this study was used to conduct experimental research on efficient profit creation for cryptocurrency investment. Several related studies have shown the results of time series prediction for long-term forecasts, such as a week or several months. Still, they have not attempted to make an ultra-short-term prediction in units of one minute. This paper attempts such a 1 min prediction. This is an experiment to create efficient profits by setting efficient trading (buying and selling) points using machine learning techniques and repeating these operations by an algorithm. Applying it to cryptocurrency shows the possibility of time series prediction.
This study analysis forecasting the bitcoin exchange rate against the USD. The dataset selected for this study starts from January 2015 to June 2022. This study's methodology uses autoregressive integrated moving average forecasting (ARIMA). The overall outcomes of this study were gathered from the statistical software Minitab 21.1. The Box Jenkins approaches are also used to predict the best model. To determine the ARIMA model parameter, this study did autocorrelation function (ACF) and partial autocorrelation function (PACF) analyses. According to the Box-Cox transformation method, log transformation was selected. The outcome demonstrates that the seasonal with the regular difference in the Bitcoin exchange rate against the USD is a stationary data series. The forecasting model used in this study is ARIMA (1,1,0) (2,1,1)12. This predicted model is identified through the Mean squared error by comparing the other guessing ARIMA models. After the prediction, 5 Month bitcoin exchange rate against the USD. Investors will be able to estimate the bitcoin exchange rate against the USD with the use of this information, but volatility must also be properly watched. This will aid investors in making better investment decisions and increase profits. In future studies, better consider another exchange rate of BTC and software experts will develop such type of software based on ARIMA models for prediction.
How to predict the change trend of asset prices in the future and decide different operation modes in advance to obtain the maximum benefits is the concern of investors. Taking gold and bitcoin as examples, this paper develops an appropriate mathematical model that uses only the past daily price stream to help traders determine whether to buy, hold or sell assets in their portfolio every day. At the same time, the robustness of the model is analyzed by robustness. The study found that holding US $1000 on September 11, 2016 will eventually maximize profits on September 10, 2021.
To be or not to be is the question that Hamlet thinks about day and night. Gold or Bitcoins is an inescapable choice for investors. With the ever rising and falling price of gold and bitcoin, making good trading decisions is of paramount importance. In this paper, we systematically investigate how data can be used to quantify the factors that influence trading and make the final decision. We build time series with the prices of gold and bitcoin for the past five years. We obtained forecast curves with excellent fit by seasonality analysis and ARIMA time series model forecasts.
Portfolio optimization is one of the most complex problems in the financial field, and technical analysis is a popular tool to find an optimal solution that maximizes the yields. This paper establishes a portfolio optimization model consisting of a weighted unidirectional dual-layer LSTM model and an SMA-slope strategy. The weighted unidirectional dual-layer LSTM model is developed to predict the daily prices of gold/Bitcoin, which addresses the traditional problem of prediction lag. Based on the predicted prices and comparison of two representative investment strategies, simple moving average (SMA) and Bollinger bands (BB), this paper adopts a new investment strategy, SMA-slope strategy, which introduces the concept of k-slope to measure the daily ups and downs of gold/Bitcoin. As two typical financial products, gold and Bitcoin are opposite in terms of their characteristics, which may represent many existing financial products in investors’ portfolios. With a principle of $1000, this paper conducts a five-year simulation of gold and Bitcoin trading from 11 September 2016 to 10 September 2021. To compensate for the SMA and BB that may miss buying and selling points, 4 different parameters’ values in the k-slope are obtained through particle swarm optimization simulation. Also, the simulation results imply that the proposed portfolio optimization model contributes to helping investors make investment decisions with high profitability.
In this paper, we aimed to build a quantitative investment trading model based on a combination of a multivariate cycle ARMA model and Apriori. We first note that in order to have a sound investment strategy, a forecast for the next trading day needs to be made. To do this, a basic time series forecasting model was first built to predict the value of gold and bitcoin for the next day based on the market volatility of the previous 40 days. The next step is developing a trading strategy model with a stable rate of return and some risk tolerance. At the same time, we developed a fixed stop-loss strategy to protect the strategy's stability and improve the risk resistance performance. Ultimately, using this model, we calculated that on 10 September 2021, we will have a return of $4816941 in Bitcoin and $1129.0503 in gold.
Yash Wadalkar, Yellamraju V H Sai Tarun, Jaiesh Singhal, Reena Sonkusare
Bitcoin, one of the most famous and high-in- demand cryptocurrencies, is a type of digital asset that is extremely difficult to track and make predictions upon. In addition, Bitcoin price does not correlate with market- movements, therefore, predicting its price action and its locus is an ordeal. In this paper, we have followed a comparative analysis approach, wherein we are using four different models to predict the trend of BTC Time series data. The results justify that the models have achieved accurate forecasting trends. During the period of 16th to 31st December 2020, Bitcoin prices experienced considerably high swings, due to the increased demand for it. In quantitative terms, the prices experienced fluctuations to the tune of 8000 USD. Despite these enormous price changes, we were able to achieve a model, that helped us attain a Mean Absolute Error (MAE) of 153.55 USD and Mean Square Error (MSE) of 43231.80 USD. Conventional Bitcoin price predicting researches follow a single to two model approach. However, for a highly volatile asset like Bitcoin, making long-term predictions and generalizing them based on limited number of models results in low accuracy outputs. This gap has been bridged in our research, we have worked with different models, as well as fragmented the time intervals into smaller portions, post which the prediction was made for only 2 days. Using this approach, we attained results with least error rates. The results obtained clearly show that ARIMA is the best model for predicting the future trends for BTC time series data. It takes into account the different types of decompositions like Regular Trend, Sessional and Residual Trend making the model give the best results.
This study interrelates three adjacent topics in data evaluation. The first is the establishment of a relationship between Bregman divergence and probabilistic alpha-divergence. In particular, we demonstrate that square-root-order probability normalization enables the unification of these two divergence families. This yields a new alpha-divergence, which can be used to jointly derive the alpha-EM algorithm (alpha-expectation-maximization algorithm) and the traditional log-EM algorithm. The second topic is the application of the alpha-EM algorithm in the evaluation of graders scoring raw data over a network. We estimate multinomial mixture distributions in this evaluation problem. We note that the convergence speed of the alpha-EM algorithm is significantly higher than that of the log-EM algorithm. Finally, the third topic is the use of this increase in convergence speed to assign the winning evaluator and miner in a blockchain environment. This is achieved by proof-of-review using evaluation scores, which is a class of proof-of-stake. In the second and third topics, we select terminology from wine tasting for brevity in the exposition. However, this formulation can be applied to a broader class of data in a network environment comprising blockchains.
Abstract— After being introduced in 2008, the rise in the price of bitcoin and the popularity of other cryptocurrencies triggered a growing discussion about how much energy was consumed during the production of this currency. Making cryptocurrency the most expensive and most popular, both the business world and the research community have begun to study the devel-opment of bitcoin. In this study bitcoin price predictions are performed using the double exponential smoothing method based on the mean absolute percentage error (MAPE). The MAPE value is used to find the best alpha (α) parameter as the basis for bitcoin price forecasting. The dataset used is the price of bitcoin from 2017 to 2019. The dataset was obtained from www.cryptocompare.com. As for the value of the alpha parameter (α), using a value of 0.1 to 0.9. Based on the test results using the double exponential smoothing method obtained the smallest MAPE value of 2.89%, with the best alpha (α) at 0.9. The prediction is done to see the price of bitcoin on January 1, 2020. The error rate generated on the predicted price of bitcoin uses an amount of 0.0373%. This shows that the system built can be used as a support for decision making when trading bitcoin.
Tri Wijayanti Septiarini, Muhammad Rifki Taufik, Mufti Afif, Atika Rukminastiti Masyrifah
Abstract The objective of this study were (i) to construct the classical statistic and artificial intelligent model for predicting bitcoin cryptocurrency, and (ii) to compare the predicting performance by using root mean square error (RMSE) and mean square error (MSE) as forecasting evaluation tool. The observation data used in this study were collected during January, 5 2017 to October, 1 2019 (in total 1,000 daily observation data). The statistical method used in this study were ARIMA (Autoregressive Moving Average) and Exponential Smoothing. The artificial intelligent model were used in this study were fuzzy time series and ANFIS (Adaptive Neuro Fuzzy Inference System). The partitions data set were of 75%-25% of training and testing, respectively. The cryptocurrency investigated was bitcoin (BTC) which is the top three of most widely traded cryptocurrency. The forecasting results show that the classical method has the smallest value of RMSE and MSE which is exponential smoothing with 9749.81 for MSE and 98.74 for RMSE. However, the performance of forecasting method cannot be guaranteed from either classical or modern forecasting method. Analyzing with different method can be considered for future study, for example machine learning, neural network, modified fuzzy time series, etc.
Bitcoin is electronic money that can be used as an alternative for investment. Investors will get benefit buying bitcoin when the price of bitcoin is down and reselling it when bitcoin prices are increasing. The fluctuating bitcoin prices cause forecasting as a basis for investors to make decisions, where the time series method is used as a forecasting model, then a pattern can be found to predict future events. The classical time series methods are often violating the statistical assumptions. To face these problems, then it is used free assumptions methods, the method with the Fuzzy Time Series Markov Chain, the Chen Logical Method, and its segmented methods due to unbalancing forecasting results. This study is built the forecasting model of the price of bitcoin for the coming period based on the data from 2010 to 2020. The proposed methods have a better fit for bitcoin time series data prices. Besides, the Fuzzy Time Series Markov Chain method has the slightly smallest accuracy error based on Mean Absolute Percentage Error (MAPE) comparing to the Fuzzy Time Series Segmented Chen Logical Method and Fuzzy Time Series Chen Logical Method.
Ioannis Chalkiadakis, Anna Zaremba, Gareth W. Peters, Michael J. Chantler
This paper establishes a new framework for assessing multimodal statistical causality between cryptocurrency market (cryptomarket) sentiment and cryptocurrency price processes. In order to achieve this, we present an efficient algorithm for multimodal statistical causality analysis based on Multiple-Output Gaussian Processes. Signals from different information sources (modalities) are jointly modelled as a Multiple-Output Gaussian Process, and then using a novel approach to statistical causality based on Gaussian Processes (GPs), we study linear and non-linear causal effects between the different modalities. We demonstrate the effectiveness of our approach in a machine learning application by studying the relationship between cryptocurrency spot price dynamics and sentiment time-series data specific to the crypto sector, which we conjecture influences retail investor behaviour. The investor sentiment is extracted from cryptomarket news data via methods developed in the area of statistical machine learning known as Natural Language Processing (NLP). To capture sentiment, we present a novel framework for text to time-series embedding, which we then use to construct a sentiment index from publicly available news articles. We conduct a statistical analysis of our sentiment statistical index model and compare it to alternative state-of-the-art sentiment models popular in the NLP literature. In regard to the multimodal causality, the investor sentiment is our primary modality of exploration, in addition to price and a blockchain technology-related indicator (hash rate). Analysis shows that our approach is effective in modelling causal structures of variable degree of complexity between heterogeneous data sources and illustrates the impact that certain modelling choices for the different modalities can have on detecting causality. A solid understanding of these factors is necessary to gauge cryptocurrency adoption by retail investors and provide sentiment- and technology-based insights about the cryptocurrency market dynamics.
This paper mainly studies the retailer's reorder quantity and the profits of retailer, supplier and entire supply chain with inventory financing under incomplete information based on entropy. First we establish a inventory financing model to derive the reorder quantity of retailer under centralized supply chain and decentralized supply chain respectively. Then introduce the maximum entropy method to predict the market demand distribution. Fianlly, the results of our experiment indicate that information value can be higher when the market demand fluctuates severely and the best reorder quantity decision for retailer should be considered from entire supply chain except the fluncation of demand is big under incomplete information, while the retailer's decentralized decision under incomplete infromation will lead the entire supply chian to the worst profit.