Cryptocurrencies have rapidly gained prominence on the global market and their effects can have significant ramifications on emerging economies.
This study uses machine learning (ML) techniques to predict and build trading strategies for cryptocurrencies. Data used includes daily returns of bitcoin, ethereum and litecoin on various exchanges as well as daily first-order autocorrelations, trading volume and market capitalization data obtained from CoinMarketCap website.
Market Characteristics
The cryptocurrency market is still relatively young and complex. Therefore, it is vitally important to understand its features. Different approaches have been devised to classify cryptocurrencies by their regulatory aspects, level of decentralization, supply issuances and economic incentives; but these approaches only cover part of the total market.
Cross-quantilogram heatmaps show weak (blue) predictability between ETH, LTC, BCH and BTC with one day’s lag under normal crypto market conditions; at larger lags these predictability patterns tend to disappear altogether.
K-means Clustering results are linked with Market Cap, Volume and Beta variables. Cluster 3 exhibits moderate behavior and includes most popular cryptocurrencies; its prototype boasts the least pronounced negative mean return and lowest standard residual when compared with the other two clusters; additionally it is associated with Market Cap variable having the highest cardinality of intersections, suggesting its cryptocurrency members possessing the greatest dominance on other markets.
Market Trends
Cryptocurrency prices tend to be highly correlated and respond similarly to market dynamics. This makes cryptocurrencies vulnerable to spillovers from traditional financial assets that respond similarly, especially given their growing mainstream use (Antonakakis et al. 2021). Contagion risks remain real (Antonakakis et al. 2021).
The cryptocurrency market’s volatility is higher than other asset markets due to lack of liquidity and high price volatility of certain cryptocurrencies, as well as being unbacked by any physical commodity which provides safe haven.
This study investigated the dynamics of 16 virtual currencies based on their market capitalization: ADA, BCH, BNB, DASH, EOS, ETH, LTC, NEO USDT TRX XEM & XRP. BTC played an influential role in price correlation network; its high-order moments, kurtosis, and skewness were studied to investigate structural changes within them.
Market Volatility
Recent developments surrounding cryptocurrency have garnered widespread interest from both academics and market practitioners, particularly its price fluctuations. While existing literature on this topic mainly covers Bitcoin-specific material.
In this study, we investigate the relationship between various cryptocurrencies and uncertainty indices of stock, oil and gold markets as well as dynamic spillover size and direction – as well as their interconnection – as well as traditional financial markets and traditional cryptocurrencies.
We discover that cryptocurrencies exhibit strong positive correlations with traditional markets, as well as among themselves by way of volatility. USD Tether’s (USDT), an ETP tied to US dollars, has weaker relations; yet its return volatility doesn’t differ significantly from that of the other cryptocurrencies; Ripple differs slightly in that its volatility pattern begins with rapid increase around December 2017 before becoming steadier over time with less extreme jumps than BTC and ETH.
Market Cap
Cryptocurrency market growth has been remarkable. Within less than a decade, cryptocurrency use has increased from seven in 2013 to 10,748 by November 2023 according to Coin Market Cap. Bitcoin currently dominates this space with an estimated market cap of over $200 billion.
Rolling window wavelet correlation coefficients show that cryptocurrencies exhibit time-varying correlations, with positive associations at relatively higher wavelet scales. Furthermore, this evidence points toward volatility connectedness and contagion among them in some wavelet scales; yet an interim period with low correlations between early and mid 2018 can be observed.
Ripple markets itself as an international payment solution utilizing blockchain technology for large financial institutions, providing lower cost and time savings compared to traditional methods such as wire transfers. Furthermore, it is expected to become a significant competitor to Bitcoin.
