Juan Plazuelo Pascual, Carlos Tardon Rubio, Juan Toro Cebada, Angel Hernando Veciana
This document analyzes price discovery in cryptocurrency markets by comparing centralized and decentralized exchanges, as well as spot and futures markets. The study focuses first on Ethereum (ETH) and then applies a similar approach to Bitcoin (BTC). Chapter 1 outlines the theoretical framework, emphasizing the structural differences between centralized exchanges and decentralized finance mechanisms, especially Automated Market Makers (AMMs). It also explains how to construct an order book from a liquidity pool in a decentralized setting for comparison with centralized exchanges. Chapter 2 describes the methodological tools used: Hasbrouck's Information Share, Gonzalo and Granger's Permanent-Transitory decomposition, and the Hayashi-Yoshida estimator. These are applied to explore lead-lag dynamics, cointegration, and price discovery across market types. Chapter 3 presents the empirical analysis. For ETH, it compares price dynamics on Binance and Uniswap v2 over a one-year period, focusing on five key events in 2024. For BTC, it analyzes the relationship between spot and futures prices on the CME. The study estimates lead-lag effects and cointegration in both cases. Results show that centralized markets typically lead in ETH price discovery. In futures markets, while they tend to lead overall, high-volatility periods produce mixed outcomes. The findings have key implications for traders and institutions regarding liquidity, arbitrage, and market efficiency. Various metrics are used to benchmark the performance of modified AMMs and to understand the interaction between decentralized and centralized structures.
Sosyal medya, insanları alış veriş alışkanlıklarından yatırım kararlarına kadar birçok ticari niyetleri üzerinde yüksek etki düzeyi olduğu güncel birçok çalışmada araştırılmaya başlanmıştır ve bu ilişki ortaya konmuştur. Bu ilişki üzerine inşa edilerek geliştirilen güncel analiz yöntemleri yatırım araçlarının gelecek değerlerini tahmin ederek yatırım kararları almada bir destek mekanizması olarak kullanılması çok cazip bir konudur. Bu sebeple bu ilişki yatırımcı ve analistlerden akademisyenlere kadar güncel bir ilgi konusu olmuştur. Bu çalışmanın amacı da sosyal medya ile yatırım kararları arasındaki ilişkiyi metinsel ve finansal analiz aracılığı ile görmeye çalışmaktır. Bu çalışmada Twitter üzerinden metin madenciliği ile veri çekilmiş ve sentiment(duygu) analizi ile yorumların olumlu ya da olumsuz olma durumu incelenmiştir. Sentiment analizinden elde edilen sayısal değerler ile güncel ve küresel bir yatırım aracı olan Bitcoin fiyatları arasındaki ilişkinin varlığını sorgulamak adına Granger Nedensellik analizi gibi finansal analizler kullanılmıştır.
Dhyanendra Jain, Ashu Jain, Amit Pandey, Jogender Kumar
Abstract Bitcoin is one of the crypto currencies and is most unpredictable currencies. In the world of crypto currencies, the value of a coin can unpredictably upgrade or degrade. In this study, the model has been trained to predict the value of Bitcoin in USD at any given time stamp. For this prediction, three algorithms of machine learning - Linear Regression (LR), Support Vector Regression (SVR) and Neural Network Regression (NNR) have been used. The model is trained using the collected dataset. After the collection of data set, we first applied SVR algorithm, then we used LR and then NNR to calculate the error compared to the actual value. Root mean squared error (RMSE) is used as the predictive measure. Out of the three algorithms, LR was found out to be more accurate for predicting the value of bitcoin.
Recent studies have found that the log-volatility of asset returns exhibit roughness. This study investigates roughness or the anti-persistence of Bitcoin volatility. Using the multifractal detrended fluctuation analysis, we obtain the generalized Hurst exponent of the log-volatility increments and find that the generalized Hurst exponent is less than $1/2$, which indicates log-volatility increments that are rough. Furthermore, we find that the generalized Hurst exponent is not constant. This observation indicates that the log-volatility has multifractal property. Using shuffled time series of the log-volatility increments, we infer that the source of multifractality partly comes from the distributional property.
Long-term prediction of multivariate time series is still an important but challenging problem. The key to solve this problem is to capture the spatial correlations at the same time, the spatio-temporal relationships at different times and the long-term dependence of the temporal relationships between different series. Attention-based recurrent neural networks (RNN) can effectively represent the dynamic spatio-temporal relationships between exogenous series and target series, but it only performs well in one-step time prediction and short-term time prediction. In this paper, inspired by human attention mechanism including the dual-stage two-phase (DSTP) model and the influence mechanism of target information and non-target information, we propose DSTP-based RNN (DSTP-RNN) and DSTP-RNN-2 respectively for long-term time series prediction. Specifically, we first propose the DSTP-based structure to enhance the spatial correlations between exogenous series. The first phase produces violent but decentralized response weight, while the second phase leads to stationary and concentrated response weight. Secondly, we employ multiple attentions on target series to boost the long-term dependence. Finally, we study the performance of deep spatial attention mechanism and provide experiment and interpretation. Our methods outperform nine baseline methods on four datasets in the fields of energy, finance, environment and medicine, respectively.
Bitcoin is considered the most valuable currency in the world. Besides being highly valuable, its value has also experienced a steep increase, from around 1 dollar in 2010 to around 18000 in 2017. Then, in recent years, it has attracted considerable attention in a diverse set of fields, including economics and computer science. The former mainly focuses on studying how it affects the market, determining reasons behinds its price fluctuations, and predicting its future prices. The latter mainly focuses on its vulnerabilities, scalability, and other techno-crypto-economic issues. Here, we aim at revealing the usefulness of traditional autoregressive integrative moving average (ARIMA) model in predicting the future value of bitcoin by analyzing the price time series in a 3-years-long time period. On the one hand, our empirical studies reveal that this simple scheme is efficient in sub-periods in which the behavior of the time-series is almost unchanged, especially when it is used for short-term prediction, e.g. 1-day. On the other hand, when we try to train the ARIMA model to a 3-years-long period, during which the bitcoin price has experienced different behaviors, or when we try to use it for a long-term prediction, we observe that it introduces large prediction errors. Especially, the ARIMA model is unable to capture the sharp fluctuations in the price, e.g. the volatility at the end of 2017. Then, it calls for more features to be extracted and used along with the price for a more accurate prediction of the price. We have further investigated the bitcoin price prediction using an ARIMA model, trained over a large dataset, and a limited test window of the bitcoin price, with length $w$, as inputs. Our study sheds lights on the interaction of the prediction accuracy, choice of ($p,q,d$), and window size $w$.
The goal of this paper is to explore the relationship between momentum effects and liquidity in cryptocurrency markets. Portfolios based on momentum-liquidity bivariate sorts are formed and rebalanced on a varying number of cryptocurrencies through time. We find a strong momentum effect in the most liquid cryptocurrencies, which supports the theories of investor herding behavior. Moreover, we propose two profitable long-only strategies: the illiquid losers and liquid winners, which exhibit improved risk adjusted performance over the market capitalization weighted portfolio.
Ferdiansyah Ferdiansyah, Edi Surya Negara, Yeni Widyanti
Cryptocurrency trade is now a popular type of investment. Cryptocurrency market has been treated similar to foreign exchange and stock market. The Characteristics of Bitcoin have made Bitcoin keep rising In the last few years. Bitcoin exchange rate to American Dollar (USD) is $3990 USD on November 2018, with daily pice fluctuations could reach 4.55%2. It is important to able to predict value to ensure profitable investment. However, because of its volatility, there’s a need for a prediction tool for investors to help them consider investment decisions for cryptocurrency trade. Nowadays, computing based tools are commonly used in stock and foreign exchange market predictions. There has been much research about SVM prediction on stocks and foreign exchange as case studies but none on cryptocurrency. Therefore, this research studied method to predict the market value of one of the most used cryptocurrency, Bitcoin. The preditct methods will be used on this research is regime prediction to develop model to predict the close value of Bitcoin and use Support vector classifier algorithm to predict the current day’s trend at the opening of the market
This paper investigates the importance of "time of execution" and the relevance of "precision time" in order driven transactions done over distributed ledgers. We created a distributed marketplace using stock market price data from the Toronto Stock Exchange (TMX). We then proceeded to test and measure the impact of timing of orders at the nanosecond level. Whilst price discovery in order driven markets is done instantaneously, with distributed markets, it is necessary to know which order to process first to avoid "front-running". We argue that a protocol for the time of order of receipt and execution should be subject to nanosecond stacking. Our approach incorporates both transitory and permanent price discovery components. It allows for the efficient processing of transactions and the order that are received by a market clearing distributed ledger.
Kalpanasonika R, Sayasri S M, Vinothini A, Suga Priya H
The accusative of this paper is to predict the bitcoin price accurately by taking various parameters into consideration which affects the bitcoin value. Here multi-layer perceptron algorithms under deep learning are used to predict the price of crypto-currency. Many researchers have analysed the crypto-currency features in many ways such as, market price prediction, the impact of cryptocurrency in real life. It has the ability to make long-term prediction of the exchange price in crypto-currencies particularly in US dollar, based on historical trends. The bitcoin cost prediction is done based on the data set which consists of 13 features relating to the crypto-currency price recorded daily over the period of particular range.
Cryptocurrency is playing an increasingly important role in reshaping the financial system due to its growing popular appeal and mechant acceptance. While many people are making investments in Cryptocurrency, the dynamical features, uncertainty, the predictability of Cryptocurrency are still mostly unknown, which dramatically risk the investments. It is a matter to try to understand the factors that infiuence the value formation. In this study, we use advanced artificial intelligence frameworks of fully connected Artificial Neural Network (ANN) and Long Short-Term Memory (LSTM) Recurrent Neural Network to analyse the price dynamics of Bitcoin, Etherum, and Ripple. We find that ANN tends to rely more on long-term history while LSTM tends to rely more on short-term dynamics, which indicate the efficiency of LSTM to utilise useful information hidden in historical memory is stronger than ANN. However, given enough historical information ANN can achieve a similar accuracy, compared with LSTM. This study provides a unique demonstration that Cryptocurrency market price is predictable. However, the explanation of the predictability could vary depending on the nature of the involved machine-learning model.
This letter investigates the dynamic relationship between market efficiency, liquidity, and multifractality of Bitcoin. We find that before 2013 liquidity is low and the Hurst exponent is less than 0.5, indicating that the Bitcoin time series is anti-persistent. After 2013, as liquidity increased, the Hurst exponent rose to approximately 0.5, improving market efficiency. For several periods, however, the Hurst exponent was found to be significantly less than 0.5, making the time series anti-persistent during those periods. We also investigate the multifractal degree of the Bitcoin time series using the generalized Hurst exponent and find that the multifractal degree is related to market efficiency in a non-linear manner.
Avinash Barnwal, Hari Pad Bharti, Aasim Ali, Vishal Krishna Singh
Predicting the direction of assets have been an active area of study and a difficult task. Machine learning models have been used to build robust models to model the above task. Ensemble methods is one of them showing results better than a single supervised method. In this paper, we have used generative and discriminative classifiers to create the stack, particularly 3 generative and 6 discriminative classifiers and optimized over one-layer Neural Network to model the direction of price cryptocurrencies. Features used are technical indicators used are not limited to trend, momentum, volume, volatility indicators, and sentiment analysis has also been used to gain useful insight combined with the above features. For Cross-validation, Purged Walk forward cross-validation has been used. In terms of accuracy, we have done a comparative analysis of the performance of Ensemble method with Stacking and Ensemble method with blending. We have also developed a methodology for combined features importance for the stacked model. Important indicators are also identified based on feature importance.
Thomas Fischer, Christopher Krauß, Alexander Deinert
Machine learning research has gained momentum—also in finance. Consequently, initial machine-learning-based statistical arbitrage strategies have emerged in the U.S. equities markets in the academic literature, see e.g., Takeuchi and Lee (2013); Moritz and Zimmermann (2014); Krauss et al. (2017). With our paper, we pose the question how such a statistical arbitrage approach would fare in the cryptocurrency space on minute-binned data. Specifically, we train a random forest on lagged returns of 40 cryptocurrency coins, with the objective to predict whether a coin outperforms the cross-sectional median of all 40 coins over the subsequent 120 min. We buy the coins with the top-3 predictions and short-sell the coins with the flop-3 predictions, only to reverse the positions after 120 min. During the out-of-sample period of our backtest, ranging from 18 June 2018 to 17 September 2018, and after more than 100,000 trades, we find statistically and economically significant returns of 7.1 bps per day, after transaction costs of 15 bps per half-turn. While this finding poses a challenge to the semi-strong from of market efficiency, we critically discuss it in light of limits to arbitrage, focusing on total volume constraints of the presented intraday-strategy.
Matheus José Silva de Souza, Fahad Almudhaf, Bruno Miranda Henrique, Ana Beatriz Silveira Negredo · 7 authors
This paper aims to investigate how Machine Learning (ML) techniques perform in the prediction of cryptocurrency prices. We answer if Support Vector Machines (SVM) and Artificial Neural Networks (ANN) based strategies can generate abnormal risk-adjusted returns when applied to Bitcoin, the largest decentralized digital currency in terms of market capitalization. Findings indicate that traders are able to earn conservative returns on the risk adjusted basis, even accounting for transaction costs, when using SVM. Furthermore, the study suggests that ANN can explore short run informational inefficiencies to generate abnormal profits, being able to beat even buy-and-hold during strong bull trends.
Jorge Salmerón, Departamento de Economía, División de Ciencias Sociales y Humanidades, Universidad Autónoma Metropolitana -Unidad Azcapotzalco
En el siguiente trabajo se realiza un análisis sobre el comportamiento de los precios de seis criptomonedas: el Bitcoin, el Etherum, el Dash, el Ripple, el Litecoin y el Zcash, haciendo uso de vectores autorregresivos, funciones de impulso-respuesta, pruebas de causalidad de Granger y la prueba de cointegración de Johansen. El objetivo de la investigación es mostrar si existe relación de corto y/o largo plazo entre las variaciones en los precios de las criptomonedas, de tal forma que cuando los precios de alguna de éstas varían, las otras también fluctúan en relación a la variación de los precios de la criptomoneda en cuestión, y por tanto, intentar anticipar las fluctuaciones, en los precios de una criptomoneda, a partir de los movimientos de las demás.
We provide a trend prediction classification framework named the random sampling method (RSM) for cryptocurrency time series that are non-stationary. This framework is based on deep learning (DL). We compare the performance of our approach to two classical baseline methods in the case of the prediction of unstable Bitcoin prices in the OkCoin market and show that the baseline approaches are easily biased by class imbalance, whereas our model mitigates this problem. We also show that the classification performance of our method expressed as the F-measure substantially exceeds the odds of a uniform random process with three outcomes, proving that extraction of deterministic patterns for trend classification, and hence market prediction, is possible to some degree. The profit rates based on RSM outperformed those based on LSTM, although they did not exceed those of the buy-and-hold strategy within the testing data period, and thus do not provide a basis for algorithmic trading.
Higor Y. D. Sigaki, Matjaž Perc, Haroldo V. Ribeiro
The efficient market hypothesis has far-reaching implications for financial trading and market stability. Whether or not cryptocurrencies are informationally efficient has therefore been the subject of intense recent investigation. Here, we use permutation entropy and statistical complexity over sliding time-windows of price log returns to quantify the dynamic efficiency of more than four hundred cryptocurrencies. We consider that a cryptocurrency is efficient within a time-window when these two complexity measures are statistically indistinguishable from their values obtained on randomly shuffled data. We find that 37% of the cryptocurrencies in our study stay efficient over 80% of the time, whereas 20% are informationally efficient in less than 20% of the time. Our results also show that the efficiency is not correlated with the market capitalization of the cryptocurrencies. A dynamic analysis of informational efficiency over time reveals clustering patterns in which different cryptocurrencies with similar temporal patterns form four clusters, and moreover, younger currencies in each group appear poised to follow the trend of their 'elders'. The cryptocurrency market thus already shows notable adherence to the efficient market hypothesis, although data also reveals that the coming-of-age of digital currencies is in this regard still very much underway.
This paper analyses the efficiency of cryptocurrency markets by applying econometric models to different short-term investment horizons. A number of experiments are carried out to demonstrate that small training sets can still be used to build efficient and useful forecasts, which in turn can be transformed into straight-forward investment strategies. It also compares the application of selected models on cryptocurrency and mature stock markets. The forecasting accuracy of the models is explored using different error metrics and different horizons. The results suggest that the variation of the error estimates doesn’t appear to be tightly related to the maturity of the markets, but rather depends on the intrinsic characteristics of the analyzed time series.
In this study, a methodology is presented where a hybrid system combining an evolutionary algorithm with artificial neural networks (ANNs) is designed to make weekly directional change forecasts on the USD by inferring a prediction using closing spot rates of three currency pairs: EUR/USD, GBP/USD and CHF/USD. The forecasts made by the genetically trained ANN are compared to those made by a new variation of the simple moving average (MA) trading strategy, tailored to the methodology, as well as a random model. The same process is then repeated for the three major cryptocurrencies namely: BTC/USD, ETH/USD and XRP/USD. The overall prediction accuracy, uptrend and downtrend prediction accuracy is analyzed for all three methods within the fiat currency as well as the cryptocurrency contexts. The best models are then evaluated in terms of their ability to convert predictive accuracy to a profitable investment given an initial investment. The best model was found to be the hybrid model on the basis of overall prediction accuracy and accrued returns.
Objetivo: Presentar tres alternativas para el diseño de un sistema difuso, con la intención de predecir la señal precio del Bitcoin (BTC) en dólares (USD). Metodología: Se realizó un análisis matemático y grafico de la señal del precio del Bitcoin [BTC]. Los datos de entrada se han seleccionado como precios de cierre para cada periodo, donde el periodo es constante y de 30 minutos. Con esto se busca identificar posibles características de interés, patrones periódicos o presencia de ruido. Resultados: En la primera aproximación a la solución del problema, se construyó un sistema difuso con motor de inferencia Mamdani en el que los parámetros de la máquina se ajustan manualmente, basándose en los patrones encontrados en el estudio previo de la señal y en el conocimiento de un experto en el área. La segunda solución se generó mediante el algoritmo ANFIS, que realiza el ajuste automático de los parámetros empleando algoritmos de gradiente descendiente; y la tercera solución se diseñó a través de la implementación de un algoritmo bioinspirado, conocido como algoritmo genético simple. Conclusiones: En los tres casos, son cinco las entradas del sistema, cada una de las cuales corresponde a muestras de la señal del precio; la salida, por su parte, es la predicción del precio para el periodo siguiente.