Influencers, Inefficiency and Fraud – The Bitcoin Price Discovery Network Under the Microscope

Abstract

We present a TriSNAR modeling framework for understanding the dynamic inter- actions of multiple markets for Bitcoin trading, including market efficiency, and for identifying influential exchanges in the global trading network. We are particularly in- terested in identifying exchanges that are market leaders. Out of 339 weeks (6.5 years of data), we identify 104 weeks in which TriSNAR provides the best MSFE out of 6 con- testants and significantly outperforms all other models. Among 194 Bitcoin exchanges, we find that exchange Kraken was the leading exchange prior to the market frenzy of 2017, in particular in 2016. In addition, price discovery shows that the Bitcoin exchange networks efficiency decreased from 2015 to 2017, and increased since 2018. We analyse the relation between blockchain fund flows and influential exchanges, and observe that wealthy holders of Bitcoin transact funds to exchanges when influential exchanges arise. We investigate the finite sample and asymptotic properties of TriSNAR. Compared to alternative methods, TriSNAR outperforms in terms of accuracy and ability to discover multi-market network structures.

Keywords: Influencer Identification, Blockchain Network Analysis, Market Efficiency, Structure Detection, Bitcoin Exchanges

JEL Classification: C01, C55, C58, G14