This paper examines price discovery and liquidity provision in the secondary market for bitcoin — an asset that has no observable fundamentals and is associated with a high level of speculative trading. Based on a comprehensive dataset of the full limit order book of BTC-e over the 2013-2014 period, we find that order informativeness generally increases with order aggressiveness within the first 10 tiers, but that this pattern reverses in the outer layers of the book. In a high volatility environment, aggressive orders seem to be more attractive to informed agents, as reflected by the increased information content of such orders, although market liquidity appears to migrate outward in response to the information asymmetry. We also find that learning in the market does not seem to depend on the history of the limit order book, which lends some initial support to the Markovian learning assumption often made in theoretical models of limit order markets.
Note: Research papers posted on SSRN, including any findings, may differ from the final version chosen for publication in academic journals.