Testing the Weak Form Efficiency of the Bitcoin Market

This paper examines the weak-form efficiency of bitcoin market. The return dynamics is consistent with the random walk model on all five examined exchanges. The random walk model was examined by four independent tests, including: serial correlation test, runs test, variance ratio and ADF tests. All of these were performed under 1% significance level. Only ADF test results were to some extent contradictory. Nevertheless, the aggregate share of unit root rejections was greater than 80% in each year for all exchanges. Moreover, the results obtained from the serial correlation, variance ratio and runs tests showed evidence for a random walk pattern. Finally, the test of the day-of-the-week effect showed that there is a small percentage of calendar anomaly on three exchanges. However, the rejection of null hypothesis, that average returns are equal in each day of the week, happens only in the particular years ( 2013 for BFNX, 2012 for BSTP and 2013 for IBIT), not on the whole sample period for each exchange. In this regard, it is most likely that in the monthly interval, informational weak form efficiency is observed. The test results are very close in all five exchanges, so there are no significant mismatches across different trading venues.