A study on the efficiency of the cryptocurrency markets using quantile regression and other efficiency measures

Abstract

We examine the cryptocurrency market with respect to the efficient market hypothesis by focusing on seven major cryptocurrencies: Bitcoin, Bisthares, Dash, Litecoin, Monero, Ripple, and Stellar. The research uses historical return observations to study whether the return series are formed by a random walk process. To do so, we employ the quantile autoregressive model and other measures to check whether the pricing behavior of cryptocurrencies is predictable. Applying the quantile autoregressive model (QAR) enables us to study the autocorrelation across the whole spectrum of return distribution, which delivers more insightful conditional information on cryptocurrency market dynamics than conventional time series models. The unit root test shows inconsistency with the random walk theory for all cryptocurrencies, whereas the symmetric test shows that returns of the previous period have no symmetrical effect on todays returns. Moreover, the results of the quantile regression claim that some of the cryptocurrencies are inconsistent with the random walk theory at certain quantiles.