Elie Bouri, Rangan Gupta, Chi Keung Marco Lau, David Roubaud
We study whether level of risk aversion can be used to predict Bitcoin returns using copulas and quantile-based models. We find evidence of predictability when the market return is at extreme quantiles. Further analyses show that the cross-quantilogram is similar when risk aversion is at the low or medium level for various quantiles of Bitcoin returns. The predictability is positive when the risk aversion is at very low level. However, predictability becomes negative when both the risk aversion and Bitcoin returns are very high, suggesting that when risk aversion and Bitcoin returns are at very high levels, Bitcoin is less likely to have large gains.
Erdinc Akyildirim, Oğuzhan Çepni, Shaen Corbet, Gazi Salah Uddin
In the aftermath of the global financial crisis and ongoing COVID-19 pandemic, investors face challenges in understanding price dynamics across assets. This paper explores the performance of the various type of machine learning algorithms (MLAs) to predict mid-price movement for Bitcoin futures prices. We use high-frequency intraday data to evaluate the relative forecasting performances across various time frequencies, ranging between 5 and 60-min. Our findings show that the average classification accuracy for five out of the six MLAs is consistently above the 50% threshold, indicating that MLAs outperform benchmark models such as ARIMA and random walk in forecasting Bitcoin futures prices. This highlights the importance and relevance of MLAs to produce accurate forecasts for bitcoin futures prices during the COVID-19 turmoil.
In <b>The Bitcoin VIX and Its Variance Risk Premium</b>, published in the Spring 2021 issue of <b><i>The Journal of Alternative Investments</i></b>, <b>Carol Alexander</b> and <b>Arben Imeraj</b> (both of the <b>University of Sussex</b>) introduce the bitcoin volatility index. CryptoCompare now streams this index every 15 seconds, under the ticker BVIN. Alexander and Imeraj are the first to investigate the bitcoin variance risk premiums and the behavior of the term structure of fair-value variance swap rates. The authors collect price data on bitcoin derivatives traded on the Deribit exchange via its application programming interface. They construct a family of indexes for different maturities using the same methodology used by CBOE’s equity volatility index, the VIX. They describe the methodology, noting that it accounts for information in volatility skews but assumes no jumps in prices. They also compare the indexes with those created with an alternative technique that does not rely on the no-jump assumption. In addition, they explore the diversification potential of bitcoin variance through correlation matrixes with other assets’ volatility indexes, realized volatilities, and other variance risk premiums. <b>TOPICS:</b>Currency, mutual funds/passive investing/indexing, statistical methods, performance measurement
Ali Raheman, Anton Kolonin, Ben Goertzel, Gergely Hegykozi · 5 authors
We present the cognitive architecture of an autonomous agent for active portfolio management in decentralized finance, involving activities such as asset selection, portfolio balancing, liquidity provision, and trading. Partial implementation of the architecture is provided and supplied with preliminary results and conclusions.
This paper proposes the PROMETHEE II based multicriteria approach for cryptocurrency portfolio selection. Such an approach allows considering a number of variables important for cryptocurrencies rather than limiting them to the commonly employed return and risk. The proposed multiobjective decision making model gives the best cryptocurrency portfolio considering the daily return, standard deviation, value-at-risk, conditional value-at-risk, volume, market capitalization and attractiveness of nine cryptocurrencies from January 2017 to February 2020. The optimal portfolios are calculated at the first of each month by taking the previous 6 months of daily data for the calculations yielding with 32 optimal portfolios in 32 successive months. The out-of-sample performances of the proposed model are compared with five commonly used optimal portfolio models, i.e., naïve portfolio, two mean-variance models (in the middle and at the end of the efficient frontier), maximum Sharpe ratio and the middle of the mean-CVaR (conditional value-at-risk) efficient frontier, based on the average return, standard deviation and VaR (value-at-risk) of the returns in the next 30 days and the return in the next trading day for all portfolios on 32 dates. The proposed model wins against all other models according to all observed indicators, with the winnings spanning from 50% up to 94%, proving the benefits of employing more criteria and the appropriate multicriteria approach in the cryptocurrency portfolio selection process.
This study investigates asymmetric multifractality and market efficiency of the major cryptocurrencies during the COVID-19 pandemic while accounting for different investment horizons. By applying the asymmetric multifractal detrended fluctuation analysis, we show that the outbreak affected the efficiency property of price behaviors differently between short- and long-term horizons. After the outbreak, the markets exhibited stronger multifractality in the short-term but weaker multifractality in the long-term. We also analyze asymmetric market patterns between upward and downward trends and between small and large price fluctuations and confirm that the outbreak has greatly changed the level of asymmetry in cryptocurrency markets.
Purpose This paper aims to demonstrate a dynamic cointegration-based pairs trading strategy, including an optimal look-back window framework in the cryptocurrency market and evaluate its return and risk by applying three different scenarios. Design/methodology/approach This study uses the Engle-Granger methodology, the Kapetanios-Snell-Shin test and the Johansen test as cointegration tests in different scenarios. This study calibrates the mean-reversion speed of the Ornstein-Uhlenbeck process to obtain the half-life used for the asset selection phase and look-back window estimation. Findings By considering the main limitations in the market microstructure, the strategy of this paper exceeds the naive buy-and-hold approach in the Bitmex exchange. Another significant finding is that this study implements a numerous collection of cryptocurrency coins to formulate the model’s spread, which improves the risk-adjusted profitability of the pairs trading strategy. Besides, the strategy’s maximum drawdown level is reasonably low, which makes it useful to be deployed. The results also indicate that a class of coins has better potential arbitrage opportunities than others. Originality/value This research has some noticeable advantages, making it stand out from similar studies in the cryptocurrency market. First is the accuracy of data in which minute-binned data create the signals in the formation period. Besides, to backtest the strategy during the trading period, this study simulates the trading signals using best bid/ask quotes and market trades. This study exclusively takes the order execution into account when the asset size is already available at its quoted price (with one or more period gaps after signal generation). This action makes the backtesting much more realistic.
We provide an empirical analysis of herding behavior in cryptocurrency markets during COVID-19 and periods of cyber-attacks, differentiating between fundamental and nonfundamental herding. The results show that herding behavior is driven by fundamental information during the full sample period and the cyber-attack days. However, herding is not prevalent during the COVID-19 outbreak, either when reacting to fundamental or nonfundamental information. This finding suggests heterogeneity in the behaviors of participants in the cryptocurrency markets during the COVID-19 period.
Dirk Gerritsen, Rick A.C. Lugtigheid, Thomas Walther
Using a hand-collected dataset containing bullish, neutral, and bearish predictions for Bitcoin published by crypto experts, we show that neutral and bearish predictions are followed by negative abnormal returns whereas bullish predictions are not associated with nonzero abnormal returns. Based on all outstanding predictions, we compute prediction revisions relative to (i) the latest issued prediction and (ii) the outstanding consensus prediction. Downward revisions are followed by negative abnormal returns. We conclude that crypto experts are skilled information intermediaries on the Bitcoin market.
Shreyas Pillai, Darshan Biyani, Ria Motghare, Deepak C. Karia
Cryptocurrencies are considered to be the next big thing in the financial sector and are the most emerging market in the current world. The amount of data available and the sophisticated architecture behind it make cryptocurrencies an excellent subject for research and thus an easy share to get deep insights of its value using machine learning for price prediction and sentiment analysis. While the previous works only used mathematical methods and various machine learning algorithms for predicting the price of cryptocurrency but forgot a vital and inseparable part that is the sentiments of the trading community which plays a vital and important role in determining and calculating the price of the share. This paper also included sentiment analysis of that share. The paper uses Long short term memory algorithm for predicting the price of the cryptocurrency and Vader sentiment analysis to predict the sentiment of the people by scrapping a news website. This paper also included a proposed methodology for creating a notification system using the dual moving cross-over technique. The result is an application which combines all three algorithms to create an efficient and accurate trading application.
This study investigates the performance of Bitcoin as a diversifier under different constraining portfolio optimization frameworks. The study employs different constraining optimization frameworks that seek to maximize risk-adjusted returns (Sharpe ratio) of the portfolio by optimizing allocations to each asset class (asset allocation). The performance attributes are evaluated by comparing the portfolios both with and without Bitcoin under frameworks ranging from equal-weighted, risk-parity, and semi-constrained to unconstrained. This study suggests that Bitcoin, due to its exotic nature, unwavering appeal, and unknown set of drivers, could act as a diversifier in normal market conditions, and it might also have some borderline hedge to safe haven properties. The results further suggest that while Bitcoin may be a potential diversifier for a risk-seeking investor, the risk-averse investor must exercise caution by limiting their exposure to Bitcoin in their portfolios, as unnecessary exposure may increase the probability of losses in extreme market conditions.
Paul Delfabbro, Daniel L. King, Jennifer N. Williams
BACKGROUND AND AIMS: Crypto-currency trading is a rapidly growing form of behaviour characterised by investing in highly volatile digital assets based largely on blockchain technology. In this paper, we review the particular structural characteristics of this activity and its potential to give rise to excessive or harmful behaviour including over-spending and compulsive checking. We note that there are some similarities between online sports betting and day trading, but also several important differences. These include the continuous 24-hour availability of trading, the global nature of the market, and the strong role of social media, social influence and non-balance sheet related events as determinants of price movements. METHODS: We review the specific psychological mechanisms that we propose to be particular risk factors for excessive crypto trading, including: over-estimations of the role of knowledge or skill, the fear of missing out (FOMO), preoccupation, and anticipated regret. The paper examines potential protective and educational strategies that might be used to prevent harm to inexperienced investors when this new activity expands to attract a greater percentage of retail or community investors. DISCUSSION AND CONCLUSIONS: The paper suggests the need for more specific research into the psychological effects of regular trading, individual differences and the nature of decision-making that protects people from harm, while allowing them to benefit from developments in blockchain technology and crypto-currency.
\nAmaç – Çalışmada önemli fiyat dalgalanmalarına sahip kripto paralardan en yüksek işlem hacmine sahip olan Bitcoin’in volatilite dinamiklerini tespit etmek için Bitcoin getirilerinin yükseliş/kazandıran ve düşüş/kaybettiren rejimleri, rejim geçiş olasılıkları ve rejimde kalma sürelerinin tespit edilmesi amaçlanmıştır. Yöntem – Çalışmada Bitcoin getirilerinin yükseliş/kazandıran ve düşüş/kaybettiren rejimleri, rejim geçiş olasılıkları ve rejimde kalma süreleri hem değişimlerin hem de rejim geçiş olasılıklarının hesaplanmasına imkan veren Markov Rejim Değişim Modeli kullanılmıştır. Bulgular – Çalışma kapsamında çalışmaya konu periyotta Bitcoin getiri serisi için en uygun modelin üç rejimli MSIH(3)-AR(1) modeli olduğu tespit edilmiştir. Modele ilişkin analizler yapıldığında ise söz konusu modelin doğrusal modele göre daha güçlü sonuçlar verdiği görülmektedir. Üç rejimden oluşan modelde katsayısı negatif olan rejim 1 daralma rejimi dönemini, katsayıları pozitif olan rejim 2 geçiş ve rejim 3 ise genişleme rejimi dönemini göstermektedir. Bitcoin getiri serisinin bir rejimdeyken bir sonraki dönemde aynı rejimde kalma olasılıkları yüksek iken bir sonraki dönemde özellikle rejim 1’den diğer rejimlere, diğer rejimlerden ise rejim 1’e geçiş olasılıklarının düşük olduğu tespit edilmiştir. Tartışma – Çalışma kapsamında ele alınan dönem için Bitcoin getirilerinin rejim kalıcılığının yüksek olduğu tespit edilmiştir. Bu bağlamda yatırımcıların, Bitcoin getirilerinin incelenen dönemde hangi rejimde olduğunu bilmesi durumunda, bir sonraki dönemde bu rejimde kalma olasılığını tahminini yaparak yatırım kararını buna göre verme imkanı bulunmaktadır. Bununla birlikte ortalama rejimlerde kalma sürelerinin düşük olduğu göz önüne alındığında özellikle Bitcoin’i portföylerinde bulunduran aktif yatırımcıların sürekli olarak bu aracın rejim değişimlerini takip etmesi durumunda portföylerinin faydasını arttırma imkanı yakalayacağı görülmektedir.\n
This paper examines interlinkages and hedging opportunities between nine major cryptocurrencies for the period between 30 September 2015 and 4 June 2020, which notably includes the coronavirus disease 2019 (COVID-19) outbreak lasting from early 2020 through the end of the sample period. The results of dynamic conditional correlation (DCC) analysis using a minimum connectedness approach show a high degree of correlation between cryptocurrencies throughout the sample period. However, the correlations reach their minimum values during the COVID-19 pandemic, which indicates that cryptocurrencies acted as a hedge or safe haven during the stressful period of the COVID-19 pandemic. The weight of cryptocurrencies was significantly reduced and their hedging effectiveness varied greatly during the pandemic, which indicates that investors&rsquo; preferences changed during the COVID-19 period.
This article sets forth a framework for deep reinforcement learning as applied to trading cryptocurrencies. Specifically, the authors adopt Q-Learning, which is a model-free reinforcement learning algorithm, to implement a deep neural network to approximate the best possible states and actions to take in the cryptocurrency market. Bitcoin, Ethereum, and Litecoin were selected as representatives to test the model. The Deep Q trading agent generated an average portfolio return of 65.98%, although it showed extreme volatility over the 2,000 runs. Despite the high volatility of deep reinforcement learning, the experiment demonstrates that it has exceptionally high potential to be employed and provides a solid foundation on which to build further research. <b>TOPICS:</b>Currency, big data/machine learning, performance measurement <b>Key Findings</b> ▪ The authors use deep neural networks to create a Deep Q-Learning trading agent that approximates the best actions to take based on rewards to maximize returns from trading the three cryptocurrencies with the largest market capitalization. ▪ The Deep Q-Learning agent generates a return of 65.98% on average over the course of 2,000 episodes; however, the returns do exhibit a large standard deviation given the highly volatile nature of the cryptocurrencies. ▪ The authors introduce a framework on which future deep reinforcement learning and rewards-based trading agents can be built and improved.
Christoph J. Börner, Ingo Hoffmann, Lars M. Kürzinger, Tim Schmitz
This paper evaluates and assesses the risk associated with capital allocation in cryptocurrencies (CCs). In this regard, we take a basket of 27 CCs and the CC index EWCI$^-$ into account. After considering a series of statistical tests we find the stable distribution (SDI) to be the most appropriate to model the body of CCs returns. However, as we find the SDI to possess less favorable properties in the tail area for high quantiles, the generalized Pareto distribution is adapted for a more precise risk assessment. We use a combination of both distributions to calculate the Value at Risk and the Conditional Value at Risk, indicating two subgroups of CCs with differing risk characteristics.