The majority of electronic markets worldwide employ limit order books, and the recently emerging exchanges for cryptocurrencies pose no exception. With this work, we empirically analyze whether commonly observed empirical properties from established limit order exchanges transfer to the cryptocurrency domain. Based on the literature, we establish a structured methodological framework to conduct analyses in a systematic and comprehensive way. We then present results from a unique and extensive limit order data set acquired from major cryptocurrency exchanges for the currency pair Bitcoin to US Dollar. We recover many observations from mature markets, such as a symmetry between the average ask and the average bid side of the order book, autocorrelation in returns on the smallest time scales only, volatility clustering and the timing of large trades. We also observe some idiosyncrasies: The distributions of trade size and limit order prices deviate from commonly observed patterns. Also, we find limit order books to be relatively shallow and liquidity costs to be relatively high when compared to established markets.

Recently, the persistence-based decomposition (PBD) model has been introduced to the scientific community by Rende et al. (2019). It decomposes a spread time series between two securities into three components capturing infinite, finite, and no shock persistence. The authors provide empirical evidence that the model adopts well to noisy high-frequency data in terms of model fitting and prediction. We put the PBD model to test on a large-scale high-frequency pairs trading application, using SP 500 minute-by-minute data from 1998 to 2016. After accounting for execution limitations (waiting rule, volume constraints, and short-selling fees) the PBD model yields statistically significant and economically meaningful annual returns after transaction costs of 9.16 percent. These returns can only partially be explained by the exposure to common risk. In addition, the model is superior in terms of risk-return metrics. The model performs very well in bear markets. We quantify the impact of execution limitations on risk and return measures by relaxing backtesting restrictions step-by-step. If no restrictions are imposed, we find annual returns after costs of 138.6 percent.