CMStatistics 2018: Start Registration
View Submission - CFE
A0524
Title: Discover regional and size effects in global bitcoin blockchain via sparse-group network autoregressive modeling Authors:  Simon Trimborn - National University of Singapore (Singapore) [presenting]
Abstract: Bitcoin blockchain has been continuously growing to a global network with millions of accounts since its creation in 2009. The dynamics of the transaction interactions reflects virtual funds movements of the Bitcoin economy. It also provides insight into the inherent risk in the Bitcoin network at a global level. We propose a Sparse-Group Network AutoRegressive (SGNAR) model to describe the essential dynamic dependence structure of Bitcoin. Our study considers up-to-date Bitcoin blockchain, from February 2012 to July 2017, with all the transactions classified into 60 groups according to region and transaction size. We develop a regularized estimator for large-dimensional dynamic network with two-layer sparsity, which enables discovering active groups with influential impact on the global Bitcoin transactions and demonstrate dynamic evolution phases of the Bitcoin network. In particular, large investors from North America and medium sized users from Europe influence the network in the last year, while previously no network connectivity was observed. It follows the inherent risk, defined as the risk of the Bitcoin network to fail, shrank lately compared to all years up to 2015.