This paper investigates how to use deep learning methods to combine with traditional multi-factor models and construct a quantitative trading model based on an AutoEncoder algorithm (AE) to classify cryptocurrencies since 2009, so as to screen out ones with investment value and then construct an effective investment portfolio. The AE algorithm is capable of handling high-dimensional data and mining interfactor non-linearities. Our empirical results on cryptocurrencies show that the model outperforms single-type factors and benchmark in terms of Cumulative Returns and the Sharpe Ratio.
This study explores whether and to what extent cryptocurrency ecosystem network connectivity predicts Bitcoin returns across quantiles of the return distribution. The facets of cryptocurrency ecosystem network connectivity we consider include connectivity between the on- and off-chain segments of the Bitcoin market, the intensity and synchronization of social and traditional crypto-focused media activity, the intensity of network correlations between cryptocurrencies. We identify tail behaviour predictors employing a quantile regression approach. The results demonstrate the effectiveness of several connectivity measures in predicting both price spikes and downfalls, but in a different way before and during the COVID-19 outbreak.
Cryptocurrencies have taken over the whole financial world with their high risk and high reward quality. Cryptocurrency markets are one of the most complex and volatile markets in the world. In this paper, we make an attempt to predict financial time series for popular cryptocurrencies like Ethereum, Binance coin and Bitcoin, using deep learning and multifractal detrended fluctuation analysis techniques. After testing for non-linearity in the financial time series of popular cryptocurrencies, it was revealed that cryptocurrency time series manifest latent relations, short term, and long term memory. The method uses Transformers and Long-short term Neural networks(LSTM) to forecast the prices of various cryptocurrencies. Although using LSTM along with transformers leads to longer computational times, but the predictive accuracy is better as compared to traditional regression neural networks and kNN forecasting models. Experimentation of the proposed model reveals that the deep learning model is profoundly efficient in predicting the intrinsic dynamics of cryptocurrency financial time series.
This article aims to test a causal nexus between bitcoin market and economic policy uncertainty. We use the continuous wavelet analysis to investigate lead-lag relationship between bitcoin market and economic policy uncertainty in different time-frequency domains. Our findings show the negative relationship between bitcoin returns and economic policy uncertainty around the period of bitcoin’s currency recognition and COVIC-19 pandemic crisis both daily and monthly time series test. Furthermore, we find that the causality relationship between bitcoin and economic policy uncertainty is relatively indistinct around the period of bitcoin’s currency recognition, while bitcoin returns are leading economic policy uncertainty changes during COVID-19 pandemic crisis, indicating the economic policy uncertainty fluctuation trend can refer to the fluctuation of bitcoin, bitcoin can be viewed as a leading indicator, but it could not be employed as a safe-haven asset hedge against uncertainty during the period of COVID-19 pandemic.
Marcin Wątorek, Jarosław Kwapień, Stanisław Drożdż
Unlike price fluctuations, the temporal structure of cryptocurrency trading has seldom been a subject of systematic study. In order to fill this gap, we analyse detrended correlations of the price returns, the average number of trades in time unit, and the traded volume based on high-frequency data representing two major cryptocurrencies: bitcoin and ether. We apply the multifractal detrended cross-correlation analysis, which is considered the most reliable method for identifying nonlinear correlations in time series. We find that all the quantities considered in our study show an unambiguous multifractal structure from both the univariate (auto-correlation) and bivariate (cross-correlation) perspectives. We looked at the bitcoin--ether cross-correlations in simultaneously recorded signals, as well as in time-lagged signals, in which a time series for one of the cryptocurrencies is shifted with respect to the other. Such a shift suppresses the cross-correlations partially for short time scales, but does not remove them completely. We did not observe any qualitative asymmetry in the results for the two choices of a leading asset. The cross-correlations for the simultaneous and lagged time series became the same in magnitude for the sufficiently long scales.
Jul 20, 2022·Proceedings of the 15th International Conference on Computer Graphics, Visualization, Computer Vision and Image Processing (CGVCVIP 2021), the 7th International Conference on Connected Smart Cities (CSC 2021) and 6th International Conference on Big Data Analytics, Data Mining and Computational Intelligence (BigDaCI’21)
Ningbo Zhu, Fei Yang, Mingzhi Zhu, Xinyao Sun and Irene Cheng
The cryptocurrency industry has evolved rapidly in recent years, and it is increasingly popular as a convenient tool tocomplement the traditional stock and futures exchanges. Accurate market research enables traders to make moreinformed decisions
The Bitcoin (BTC) is one of the most popular cryptocurrencies and now is one type of investment on the stock market. The price prediction is a real challenge as it depends by too many factors, an investment to BTC characterized too risky because the price has too many upside-downs. However BTC was the occasion for many people to start explore the stock market. In the last few years, the prediction of BTC has occupied the scientific community and many approaches have been made. In this research we try to predict the price of BTC per minute with long short-term memory networks, and then pass these predictions to a recurrent reinforcement learning (RRL) model to trade the BTC with United States dollars (USD). In the end of the paper we present the profits that the models made.
Japjeet Singh, Sulalitha Bowala, A. Thavaneswaran, Ruppa K. Thulasiram · 5 authors
The forecasting problems in Computational Finance involve modelling the vagueness and imprecision inherent to the financial markets. Fuzzy set theory has a unique ability to quantitatively and qualitatively model and analyze such problems. Volatility forecasting plays an important role in financial risk management and in option pricing. Recently, there has been a growing interest in data-driven volatility models and neurovolatility models for risk forecasting of stocks and index funds. However, even these state-of-the-art models do not take into account the fuzzy volatility in their risk forecasts.Cryptocurrencies are a novel financial asset class based on the Blockchain technology. Cryptocurrencies have gained popularity among retail investors as a financial asset with high risks and high returns. The extremely volatile nature of cryptocurrencies (compared to traditional assets) makes forecasting their volatility more challenging. A simple algorithmic trading approach, Simple Moving Average (SMA) crossover strategy, is used to calculate the Algo returns. This paper provides fuzzy forecasts of the volatility of Algo returns using the data-driven Exponentially Weighted Moving Average (DD-EWMA) and neuro models for six major cryptocurrencies. We also compute and compare fuzzy volatility forecasts of four major tech stocks and Chicago Board Options Exchange’s (CBOE) volatility index (VIX) using DD-EWMA and neuro models. Our experimental results show that the data-driven models produce better forecasts for cryptocurrencies as compared to the neuro models, while for the regular stocks and indexes, no such definitive conclusion could be drawn.
This is the first study to examine the quantile connectedness for returns-volume and volatility-volume pairs for the three non-fungible tokens (THETA, Tezos, and Enjin Coin) using the quantile VAR approach. The results report the highest connectedness of volume with returns and volatility in the extreme upper quantile compared to other quantiles, implying the asymmetric connectedness. The spillover effect is observed from volume to returns and volatilities in extreme upper and lower market conditions, whereas opposite direction of spillovers is evident for the selected non-fungible tokens at median quantile. Our findings are useful for investors in predicting the returns and risk of NFTs using trading volume in the extreme market conditions.
Cryptocurrencies are new kinds of electronic currencies based on communication technologies. These currencies have attracted the attention of investors. However, cryptocurrencies are very volatile and unpredictable. For investors, it is very difficult to make investment decisions in cryptocurrency market. Therefore, revealing changes in the dynamics of cryptocurrencies are valuable for investors. Bitcoin is the most popular and representative cryptocurrency in cryptocurrency market. In this study how dynamical properties of Bitcoin changed through time is analyzed with recurrence quantification analysis (RQA). RQA is a pattern recognition-based time series analysis method that reveals dynamics of the time series by calculating some metrics called RQA measures. This method has been successfully applied to nonlinear, nonstationary, short and chaotic time series and does not assume a statistical model. RQA can reveal important properties of time series data such as determinism, laminarity, stability, randomness, regularity and complexity. By using sliding window RQA we show that in 2021 RQA measures for Bitcoin prices collapse and Bitcoin becomes more unpredictable, more random, more unstable, more irregular and less complex. Therefore, dynamics and stability of the Bitcoin prices significantly changed in 2021.
The main contribution of this research is to investigate whether an Artificial Neural Network is an option to predict Ethereum cryptocurrency close price on a time constrained scenario. The ANN training time and time lagged data availability are considered as constraints on finding the fastest and the most accurate regression model using ARIMA results as a baseline. As part of the study, hourly aggregated data is processed to generate a step-ahead forecast and then processing time is compared for each architecture. Previous work related to cryptocurrency forecasting usually focus the analysis only on accuracy, and use coarser data granularity. Results have shown that convolutional neural networks over performed other architectures for accuracy and time objectives.
Digital technology developments shape the behaviour, performances, standards of society, organizations and individuals imposing new ways of payments and new forms of money. In this environment in 2008 was developed a new type of currency, namely Bitcoin. Cryptocurrency, as this new form of money has been generically called, puts pressure on the traditional concept of money. Today, the economic value of cryptocurrencies is attested by their circulation and acceptance by user communities for trade. However, establishing this value raises debates in the literature. The research from this paper investigates and analyses if there is a strong enough connectedness between Bitcoin price evolution and energy consumption tendency (for mining), to influence Bitcoin value. Public data from January 2014 to July 2021 is used. An Artificial Neural Network (ANN) was used to study and predict the tendency of Bitcoin price and energy consumption. A comparison between the forecasting trend and the real trend (the evolution of energy consumption and Bitcoin price) was made. The conducted research starts with a quantitative one and ends with a qualitative one (trends). The obtained results show that qualitatively, there is a good correlation between monthly average values of BTC prices and electricity consumption for mining.
We investigate which factors contribute most to the liquidity of Bitcoin, using a diverse universe of candidate factors reflecting key developments in the crypto market and the global economy. The empirical analysis relies on three regularized linear regression methods, viz. LASSO, adaptive LASSO, and elastic net. We also apply a cross-fit partialing-out LASSO instrumental-variables regression model, as a supplementary approach to handle endogeneity. Findings reveal that trading volume and realized volatility of Bitcoin, cryptocurrency hacks, Ethereum liquidity, and public attention are the most common drivers of liquidity, irrespective of the penalized regression approach and liquidity proxy adopted. Our evidence confirms the paramountcy of cryptocurrency-specific factors over global economic and financial ones in influencing Bitcoin liquidity.
This research assesses the diversification patterns for tail dependence between Bitcoin return and trading volume by utilizing a dynamic mixture copula approach with spillover effect and asymmetric volatility effect. There are four main empirical findings. First, the spillover effect between return and trading volume exists. Second, the leverage effect is statistically significant for return and trading volume. Third, the linkages between return and trading volume are diversified. Both positive and negative tail dependence structures are observed, and the frequency of a positive tail dependence occurring is higher. Furthermore, the asymmetric tail dependence structure exists in positive and negative dependence situations. In the positive dependence structure, a co-movement in the increasing direction is stronger than a co-movement in the decreasing direction. In the negative dependence structure, a situation of a large return with low volume occurs more often than a situation of a small return with high volume. Fourth, the volatility of trading volume positively predicts the magnitude of positive dependence.
Tomas Scagliarini, Giuseppe Pappalardo, Alessio Emanuele Biondo, Alessandro Pluchino · 6 authors
In this paper we analyse the effects of information flows in cryptocurrency markets. We first define a cryptocurrency trading network, i.e. the network made using cryptocurrencies as nodes and the Granger causality among their weekly log returns as links, later we analyse its evolution over time. In particular, with reference to years 2020 and 2021, we study the logarithmic US dollar price returns of the cryptocurrency trading network using both pairwise and high-order statistical dependencies, quantified by Granger causality and O-information, respectively. With reference to the former, we find that it shows peaks in correspondence of important events, like e.g., Covid-19 pandemic turbulence or occasional sudden prices rise. The corresponding network structure is rather stable, across weekly time windows in the period considered and the coins are the most influential nodes in the network. In the pairwise description of the network, stable coins seem to play a marginal role whereas, turning high-order dependencies, they appear in the highest number of synergistic information circuits, thus proving that they play a major role for high order effects. With reference to redundancy and synergy with the time evolution of the total transactions in US dollars, we find that their large volume in the first semester of 2021 seems to have triggered a transition in the cryptocurrency network toward a more complex dynamical landscape. Our results show that pairwise and high-order descriptions of complex financial systems provide complementary information for cryptocurrency analysis.
This paper investigates the pricing of liquidity risk in the cross-section of cryptocurrencies from January 2017 to December 2020. The cryptocurrencies with high liquidity risk (beta) earned a risk-adjusted return of 4.4% higher weekly than those with low liquidity risk after controlling for the market, size, and reversal factors. Furthermore, the positive relation between expected cryptocurrency returns and liquidity risk is robust when I use cross-sectional regression tests for individual cryptocurrencies and alternative liquidity measures. The results suggest that liquidity risk is an important determinant of expected cryptocurrency returns.