Bitcoin Risk Modeling with Blockchain Graphs

Abstract: A key challenge for Bitcoin cryptocurrency holders, such as startups using ICOs to raise funding, is managing their FX risk. Specifically, a misinformed decision to convert Bitcoin to fiat currency could, by itself, cost USD millions. In contrast to financial exchanges, Blockchain based crypto-currencies expose the entire transaction history to the public. By processing all transactions, we model the network with a high fidelity graph so that it is possible to characterize how the flow of information in the network evolves over time. We demonstrate how this data representation permits a new form of microstructure modeling - with the emphasis on the topological network structures to study the role of users, entities and their interactions in formation and dynamics of crypto-currency investment risk. In particular, we identify certain sub-graphs ('chainlets') that exhibit predictive influence on Bitcoin price and volatility, and characterize the types of chainlets that signify extreme losses.