Beschreibung:
We mine the leaked history of trades on Mt. Gox, the dominant Bitcoin exchange from 2011 to early 2014, to detect the triangular arbitrage conducted within the platform. We exploit user identifiers per trade to identify and describe the individual trading patterns of 440 arbitrageurs. We introduce proxies for expertise and show that the non sophisticated users conduct on average non-profitable arbitrage when transaction costs are accounted for, while the expert ones obtain a positive and statistically significant premium. Most importantly, by including user fixed effects, we document that expert users make profits on arbitrage by reacting quickly to plausible exogenous variations on the official exchange rates. A small number of arbitrageurs conduct the vast majority of the arbitrage actions and systematically yield higher profits: our results provide empirical evidence that arbitrageurs are few and sophisticated users, characterized by the ability to incorporate information and quick reaction to exogenous shocks within tiny time scale intervals