Maria Letizia Guerra, Laerte Sorini, Luciano Stefanini
Sentiment analysis to characterize the properties of Bitcoin prices and their forecasting is here developed thanks to the capability of the Fuzzy Transform (F-transform for short) to capture stylized facts and mutual connections between time series with different natures. The recently proposed Lp-norm F-transform is a powerful and flexible methodology for data analysis, non-parametric smoothing and for fitting and forecasting. Its capabilities are illustrated by empirical analyses concerning Bitcoin prices and Google Trend scores (six years of daily data): we apply the (inverse) F-transform to both time series and, using clustering techniques, we identify stylized facts for Bitcoin prices, based on (local) smoothing and fitting F-transform, and we study their time evolution in terms of a transition matrix. Finally, we examine the dependence of Bitcoin prices on Google Trend scores and we estimate short-term forecasting models; the Diebold–Mariano (DM) test statistics, applied for their significance, shows that sentiment analysis is useful in short-term forecasting of Bitcoin cryptocurrency.
In the finance sector, in general, a single VaR method is used for one single portfolio or for all similar portfolios and it hampers the opportunity for comparison. Such shortcoming deriving from trusting one single VaR method results in very incoherent results for the analysis as well as in untrustable transactions based upon those risk estimations. In order to overcome that, similar investments tools/portfolios should be analysed simultaneously by different VaR methods for comparison. Considering such overcome, this study is aimed to compare the VaR (value at risk) estimation methodologies for all 5 separated portfolios (which are similar considering their liquidity and investment process) holding USD, EUR, GOLD, BIST100 Index (Istanbul Stock Exchange Index) and BITCOIN considering their daily return on TRL (Turkish Lira). For performance measurement of different methodologies listed namely as extreme value VaR (GRPD-gnadenko theorem), ewma based volatility filtered historical simulation, historical simulation, delta normal, and bootstrapping; the 3 backtesting procedures and the related statistics are used.
Abstract The purpose of this research is to identify how effective the determinants of the improved price changes in cryptocurrencies are and if they are predictable. The study addresses several independent variables that are in our consideration which may impact the prices the most. To obtain the results, panel data has been used to run fixed effects models. Then we treated them as time series data to run dynamic trend indicator and first-differencing volatility regression model. Important political shocks and instabilities have been analyzed and interpreted in this paper. In the light of our findings we were able to comment on the complex relation between cryptocurrency prices and socio-political situations throughout the time range. The results address that cryptocurrency price changes are predictable. It is easy to say that major stakeholders (Apple, Amazon, Facebook, Google, Tesla) affect the most prices. Internet search trends seem to have an impact but at the end it has been found that the correlation is strong. We have evaluated all the major cryptocurrency prices with exact accuracy of 95.38% using the volatility regression model effectively. All the cryptocurrencies are evaluated against US dollars in regard of different cryptocurrency like Bitcoin, Ethereum, Litecoin and Ripple digital currency. Cryptocurrencies shouldn’t be seen as a gambling medium and should be taken more seriously like an investment medium. In some specific occasions investing in cryptocurrencies may lead lucrative income.
Abstract From the past two years with increasing geopolitical and economic issues, global currency values have been falling and stock markets have been having a poor run & investors losing wealth. This has led to a renewal of interest in digital currencies. Cryptocurrency one of the most prominent digital currency has found itself in spotlight with investors wanting a piece of it and business establishments accepting it as a source of payment due to its stable performance in the last few years. This research has been done on predicting cryptocurrency prices using machine learning based neural network which has a lowest the model loss over 100 epochs during training and Technical Trade Indicators (TTI) graphs depicts a real BTC value 5 to 10 times in 300-days of current fiscal year has further supported this increasing trader confidence and a shift in global cryptocurrency graph by predicted BTC values. On the same lines, we are analyzing bitcoin prices using Machine Learning and Sentiment Analysis. We also study stock market trends in order to better predict bitcoin prices quantitively. In this work we analyze the impact of global currencies like US Dollar, foreign exchanges on Bitcoin prices and whether Bitcoin has the stability to dethrone global currencies and become the single medium of transaction. This work is adequate enough to aid in predicting price and with results obtained from predicting Bitcoin prices using machine learning based neural network achieving an accuracy of 94.89% under all circumstances of technical trade indication thereby bringing down its price prediction by over 13.7% in April 2020 itself during evaluation.
Abstract Electronic payments is something that is currently in high demand by investors today, but transactions are often constrained due to various problems, especially from third parties. For this reason, cryptocurrency emerged, which is one of the solutions for conducting electronic payment transactions. Some types of cryptocurrency that are most in demand by investors are bitcoin, ethereum, and ripple. The fluctuation value of cryptocurrency is very difficult to predict so that investors often experience losses when making transactions. This study aims to predict cryptocurrency prices such as bitcoin, ethereum and ripple using data mining algorithms. The data mining algorithm used in this prediction process is K-NN, Neural Network, SVM, Linear Regression, Random Forest and Decision Tree. Data mining modeling is done by dividing the dataset into each type of commodity and then analyzed using each algorithm. The results of this study indicate that the accuracy value obtained from some data mining algorithms is good enough to predict cryptocurrency prices
This paper is discusses the problems of the short-term forecasting of financial time series using supervised machine learning (ML) approach. For this goal, we applied several the most powerful methods including Support Vector Machine (SVM), Multilayer Perceptron (MLP), Random Forests (RF) and Stochastic Gradient Boosting Machine (SGBM). As dataset were selected the daily close prices of two stock index: SP 500 and NASDAQ, two the most capitalized cryptocurrencies: Bitcoin (BTC), Ethereum (ETH), and exchange rate of EUR-USD. As features we used only the past price information. To check the efficiency of these models we made out-of-sample forecast for selected time series by using one step ahead technique. The accuracy rates of the forecasted prices by using ML models were calculated. The results verify the applicability of the ML approach for the forecasting of financial time series. The best out of sample accuracy of short-term prediction daily close prices for selected time series obtained by SGBM and MLP in terms of Mean Absolute Percentage Error (MAPE) was within 0.46-3.71 %. Our results are comparable with accuracy obtained by Deep learning approaches.
In recent years, bitcoin has become a very attractive investment in financial industry, which is not controlled by governments, but is based on trust between transfers under the technology of block chain. Hence, forecasting future bitcoin cryptocurrency values is a problem that has attracted the attention of many researchers in the field, while proving to be a very challenging problem. This work presents an experimental analysis using LSTM and GRUs for forecasting bitcoin values in a minute-granulated time for the entire next day. To this end we also present our methodology for conducting the experiments. The final goal is to create the core of a financial prediction tool around the RNNs. In our experiments, we achieved interesting results such as a SMAPE of 0.0002, a RMSE of US$ 3.844 and a rRMSE of 0.0028 in a day where bitcoin rates vary from US$ 13.2K and US$ 14.6K, surpassing the results of SMAPE found in the literature and proposed limit of SMAPE smaller than 0.007 for forecasts.
This paper provides a comprehensive state-of-the-art investigation of the recent advances in data science in emerging economic applications. The analysis is performed on the novel data science methods in four individual classes of deep learning models, hybrid deep learning models, hybrid machine learning, and ensemble models. Application domains include a broad and diverse range of economics research from the stock market, marketing, and e-commerce to corporate banking and cryptocurrency. Prisma method, a systematic literature review methodology, is used to ensure the quality of the survey. The findings reveal that the trends follow the advancement of hybrid models, which outperform other learning algorithms. It is further expected that the trends will converge toward the evolution of sophisticated hybrid deep learning models.
Trading cryptocurrencies (digital currencies) are currently performed by applying methods similar to what is applied to the stock market or commodities; however, these algorithms are not necessarily well-suited for predicting cryptocurrency prices. Unlike stock exchanges, which shut down for several hours or days at a time, digital currency prediction and trading seem to be of a more consistent and predictable nature. In this work, we benefit from sentiment analysis of tweets using both an existing sentiment analysis package and a manually tailored “objective analysis,” to calculate one impact value for each analysis every 15[Formula: see text]min. We then select the most appropriate training method by applying evolutionary techniques and discover the best subset of the generated features to include, as well as other parameters. One of the unique contributions of this work is the analysis of both English and Japanese tweets with a tailored “objective analysis” tool. This resulted in implementation of predictors which yielded 28% to 122% profit in a four-week simulation, much more than simply holding a digital currency for the same period of time.
BITCOIN has a different criterion than traditional systems that pay in states’ currencies. This payment system is a complex scheme designed to facilitate the transfer of value between the parties. In this study, firstly, brief information about technical analysis, BITCOIN and behavioral finance is given. Then, in the literature part of the study, studies on BITCOIN prices in the context of behavioral finance and technical analysis are given. In this study, it is examined Relative Strength Index (RSI) availability in Bitcoin transactions and evaluate in the context of behavioral finance findings. For describing the risk of trading in Bitcoin, were chosen Value at Risk (VaR) ratio. In application part of the study it is supposed 1 Bitcoin “Buy” orders were opened when RSI was under 30 and closed when RSI was above 70. And also, 1 Bitcoin “Sell” orders were opened when RSI was above 70 and closed when it was under 30. All obtained data from trades was used for revealing results on accuracy, total profitability. Positive trades were divided by total trades and multiplied by 100 for calculation of accuracy. Period of research is 2015-01-01 till 2019-08-31. As a result of the study we see the effects of biases in Bitcoin transactions. It is observed the examples of conservatism, over and underreaction, status quo effect and loss aversion. And also it is determined in this study that RSI works better in stable market when traders play safer. In other words, RSI works better when conservatism wins over overreaction.
Introduced in 2009, Bitcoin has demonstrated a huge potential as the world’s first digital currency and has been widely used as a financial investment. Our research aims to uncover the relationship between Bitcoin prices and people’s sentiments about Bitcoin on social media. Among various social media platforms, micro-blogging is one of the most popular. Millions of people use micro-blogging platforms to exchange ideas, broadcast views, and to provide opinions on different topics related to politics, culture, science, and technology. This makes them a potentially rich source of data for sentiment analysis. Therefore we chose one of the busiest micro-blogging platforms, Twitter, to perform sentiment analysis on Bitcoin. We used ELMo embedding model to convert Bitcoin-related tweets into a vector form and SVM classifier to divide the tweets into three sentiment categories - positive, negative, and neutral. We then used the sentiment data to find its relation with Bitcoin price fluctuation using the linear mixed model.
Price fluctuation is a necessity in every business, including the price of Bitcoin. Therefore humans need a device or application that can assist in making predictions quickly and accurately. Deep Learning can be used to make predictions which include setting the parameters used during training, choosing the best parameters for prediction, and choosing the prediction results with the smallest error level in the actual situation. This study performs a Bitcoin price prediction using Long Short Term Memory (LSTM). This study uses quantitative calculations on the prediction results. Measurement of accuracy is done by testing the previous price (back testing) and calculating the average error using RMSE and MAPE. The method for generating predictions is LSTM as a type of Recurrent Neural Network (RNN) which has the advantage of using long-term memory in handling time series data. LSTM implementation using Python is used for price forecasting with stages starting from data collection and normalization, input and output modeling. The result of this research is a prediction of Bitcoin price with an accuracy rate of 97.48% based on the best model with input layer 2, the number of epochs 100, the number of hidden layers 100, and activation using Softmax. These price predictions can be used as a reference and consideration for traders to create trading strategies and run them automatically on the digital currency market.
This study aims to investigate if the investors’ sentiment expressed on contentspecific online forum affects the return of Bitcoin. We use a large dataset consisting of 2.8 million forum posts sourced from “Bitcointalk.org” for a period between Jan 2016 and May 2020. The sentiment is derived daily with the Hu Liu lexical model after a detailed investigation of different lexicons. Using time-series data, we test for the relationship between the investors’ sentiment and Bitcoin price along with other financial variables that may inform about the direction of Bitcoin price. Our results show that sentiment derived from online forums do not present an autoregressive relationship with return of Bitcoin
Bitcoin, as one of the most popular cryptocurrency, is recently attracting\nmuch attention of investors. Bitcoin price prediction task is consequently a\nrising academic topic for providing valuable insights and suggestions. Existing\nbitcoin prediction works mostly base on trivial feature engineering, that\nmanually designs features or factors from multiple areas, including Bticoin\nBlockchain information, finance and social media sentiments. The feature\nengineering not only requires much human effort, but the effectiveness of the\nintuitively designed features can not be guaranteed. In this paper, we aim to\nmining the abundant patterns encoded in bitcoin transactions, and propose\nk-order transaction graph to reveal patterns under different scope. We propose\nthe transaction graph based feature to automatically encode the patterns. A\nnovel prediction method is proposed to accept the features and make price\nprediction, which can take advantage from particular patterns from different\nhistory period. The results of comparison experiments demonstrate that the\nproposed method outperforms the most recent state-of-art methods.\n
The Bitcoin (BTC) market presents itself as a new unique medium currency, and it is often hailed as the “currency of the future”. Simulating the BTC market in the price discovery process presents a unique set of market mechanics. The supply of BTC is determined by the number of miners and available BTC and by scripting algorithms for blockchain hashing, while both speculators and investors determine demand. One major question then is to understand how BTC is valued and how different factors influence it. In this paper, the BTC market mechanics are broken down using vector autoregression (VAR) and Bayesian vector autoregression (BVAR) prediction models. The models proved to be very useful in simulating past BTC prices using a feature set of exogenous variables. The VAR model allows the analysis of individual factors of influence. This analysis contributes to an in-depth understanding of what drives BTC, and it can be useful to numerous stakeholders. This paper’s primary motivation is to capitalize on market movement and identify the significant price drivers, including stakeholders impacted, effects of time, as well as supply, demand, and other characteristics. The two VAR and BVAR models are compared with some state-of-the-art forecasting models over two time periods. Experimental results show that the vector-autoregression-based models achieved better performance compared to the traditional autoregression models and the Bayesian regression models.
Education on cryptocurrency is essential for individuals to make informed decisions regarding foreign investment in digital assets. In recent years there is an exponential increase in the price of cryptocurrency due to its easy trading especially in developing countries so the trend of financial institutions buying cryptocurrency into their portfolios has grown in the past decade which results in economic growth. The first completely digital assets that asset managers have included are cryptocurrencies. Traders have a unique opportunity to forecast price swings due to social media’s impact on cryptocurrency prices. Trading using Al and Machine Learning has drawn more attention in recent years. One could investigate the above hypothesis to determine if it is feasible to capitalize on the Bitcoin market’s inefficiency for the purpose of generating unusually high profits. The advanced machine-learning techniques enable straightforward trading strategies to exceed conventional benchmarks. The findings demonstrate how basic computational processes might assist predict the near-term development of the bitcoin market. Further, there are prediction and comparison prices using SVM and Random Forest algorithms on the basics of efficiency while changing the number of days.
Uğur Kaya, Fırat Akba, İ̇hsan Tolga Medeni, Tunç D. Medeni
Son zamanlarda kullanımı oldukça yaygınlaşan blokzinciri teknolojisinin, İnternet teknolojisi ile beraber adı sıkça anılır olmaya başlamıştır. Blokzinciri teknolojisiyle geliştirilen Bitcoin, sanal para birimleri arasında en çok piyasa hacmini elinde bulunduran sanal para birimidir. Sanal para piyasalarının kontrolünü elinde bulunduran bir merkezi otoritenin olmaması sebebiyle fiyat manipülasyonlarına ve dışarıdan müdahalelere açık olan bu pazarda, en uçtaki yatırımcının yatırım yapabilmesi açısından yol gösterimine ihtiyaç duyulmaktadır. Son zamanlarda bu ihtiyacı karşılamak amacıyla birtakım yöntemler kullanılmaya başlanmıştır. Bu çalışmada makine öğrenmesi, zaman serileri analizi ve derin öğrenme yöntemleri kullanılarak Bitcoin fiyatlarındaki dalgalanma hakkında çeşitli tahminleme ve sınıflama yöntemleri beraber olarak değerlendirilmiştir. Bu bağlamda, koronavirüs pandemisi öncesi ve sonrasındaki Bitcoin kapanış fiyatları ve düşüş-yükseliş eğilimleri baz alınarak iki ayrı veri kümesi oluşturulmuştur. Bu veri kümeleri üzerinde tahmin ve sınıflama yöntemleri değerlendirilerek, başarıları karşılaştırılmıştır. Karşılaştırmalar sonucunda, pandemi öncesi verilerle yapılan çalışmada Destek Vektör Makineleri, pandemi sonrası verilerle yapılan çalışmada ise ARIMA en başarılı sonuçları vermiştir.
This research has examined the ability of two forecasting methods to forecast Bitcoin's price trends. The research is based on Bitcoin-USA dollar prices from the beginning of 2012 until the end of March 2020. Such a long period of time that includes volatile periods with strong up and downtrends introduces challenges to any forecasting system. We use particle swarm optimization to find the best forecasting combinations of setups. Results show that Bitcoin's price changes do not follow the "Random Walk" efficient market hypothesis and that both Darvas Box and Linear Regression techniques can help traders to predict the bitcoin's price trends. We also find that both methodologies work better predicting an uptrend than a downtrend. The best setup for the Darvas Box strategy is six days of formation. A Darvas box uptrend signal was found efficient predicting four sequential daily returns while a downtrend signal faded after two days on average. The best setup for the Linear Regression model is 42 days with 1 standard deviation.
Prosper Lamothe-Fernández, David Alaminos, Prosper Lamothe-López, Manuel Á. Fernández-Gámez
A precise prediction of Bitcoin price is an important aspect of digital financial markets because it improves the valuation of an asset belonging to a decentralized control market. Numerous studies have studied the accuracy of models from a set of factors. Hence, previous literature shows how models for the prediction of Bitcoin suffer from poor performance capacity and, therefore, more progress is needed on predictive models, and they do not select the most significant variables. This paper presents a comparison of deep learning methodologies for forecasting Bitcoin price and, therefore, a new prediction model with the ability to estimate accurately. A sample of 29 initial factors was used, which has made possible the application of explanatory factors of different aspects related to the formation of the price of Bitcoin. To the sample under study, different methods have been applied to achieve a robust model, namely, deep recurrent convolutional neural networks, which have shown the importance of transaction costs and difficulty in Bitcoin price, among others. Our results have a great potential impact on the adequacy of asset pricing against the uncertainties derived from digital currencies, providing tools that help to achieve stability in cryptocurrency markets. Our models offer high and stable success results for a future prediction horizon, something useful for asset valuation of cryptocurrencies like Bitcoin.
Alla A. Petukhina, Raphael C. G. Reule, Wolfgang Karl Härdle
This research analyses high-frequency data of the cryptocurrency market in regards to intraday trading patterns related to algorithmic trading and its impact on the European cryptocurrency market. We study trading quantitatives such as returns, traded volumes, volatility periodicity, and provide summary statistics of return correlations to CRIX (CRyptocurrency IndeX), as well as respective overall high-frequency based market statistics with respect to temporal aspects. Our results provide mandatory insight into a market, where the grand scale employment of automated trading algorithms and the extremely rapid execution of trades might seem to be a standard based on media reports. Our findings on intraday momentum of trading patterns lead to a new quantitative view on approaching the predictability of economic value in this new digital market.
Bitcoin has attracted extensive attention from investors, researchers, regulators, and the media. A well-known and unusual feature is that Bitcoin’s price often fluctuates significantly, which has however received less attention. In this paper, we investigate the Bitcoin price fluctuation prediction problem, which can be described as whether Bitcoin price keeps or reversals after a large fluctuation. In this paper, three kinds of features are presented for the price fluctuation prediction, including basic features, traditional technical trading indicators, and features generated by a Denoising autoencoder. We evaluate these features using an Attentive LSTM network and an Embedding Network (ALEN). In particular, an attentive LSTM network can capture the time dependency representation of Bitcoin price and an embedding network can capture the hidden representations from related cryptocurrencies. Experimental results demonstrate that ALEN achieves superior state-of-the-art performance among all baselines. Furthermore, we investigate the impact of parameters on the Bitcoin price fluctuation prediction problem, which can be further used in a real trading environment by investors.
In this paper, we study the volatility forecasts in the Bitcoin market, which has become popular in the global market in recent years. Since the volatility forecasts help trading decisions of traders who want a profit, the volatility forecasting is an important task in the market. For the improvement of the forecasting accuracy of Bitcoin’s volatility, we develop the hybrid forecasting models combining the GARCH family models with the machine learning (ML) approach. Specifically, we adopt Artificial Neural Network (ANN) and Higher Order Neural Network (HONN) for the ML approach and construct the hybrid models using the outputs of the GARCH models and several relevant variables as input variables. We carry out many experiments based on the proposed models and compare the forecasting accuracy of the models. In addition, we provide the Model Confidence Set (MCS) test to find statistically the best model. The results show that the hybrid models based on HONN provide more accurate forecasts than the other models.