Bitcoin Transaction Graph Analysis
Michael S. Kester
January 3, 2014
Bitcoins have recently become an increasingly popular cryptocurrency through which users trade el ec tr on i -
cally and more anonymously than via traditional electronic transfers. Bitcoin’s design keeps all transactions
in a public ledger. The sender an d receiver for each transaction are identiﬁed only by cryptographic pub li c -
key ids. This leads to a common miscon ce pt i on that it inherently provi d es anony m ous use. While Bitcoin’s
presumed anonymity o↵ers new avenues for commerce, several recent studies raise user-privacy concerns.
We explore the level of anonymity in the Bit coi n system. Our approach is two-fold: (i) We annotate the
public transaction graph by linking bitcoin public keys to real people - either deﬁnitively or stat i st i c all y. (ii)
We run th e annotated graph through our graph-an aly s is framework to ﬁnd and summariz e activity of both
known and unknown u se rs .
We present a bitcoin transacti on-gr ap h- ann ot at ion s y st em i n two parts. First, we deve l oped a system for
scraping bitcoin addresses from public forums. Second, we include a mechanism for matching users to
transactions using incomplete transaction information. For example, suppose we hear Bob say to Al ic e:
“I sent you $100 in bitcoins yesterday at noon”; t hou gh we don’t kn ow the exact ti m e of the transaction
(since “at noon” could easily mean 11:50 or 12:10) or the exact amount in bi tc oi ns (exchange rates ﬂuctuate
signiﬁcantly), we can generate candidate transac t ion matches and associated matching pr obab i l it i es .
We also present a graph-analysis framework capable of tracing and clustering user activity. For example,
our framework suggested the FBI seizure of Silk Road assets as “interesting” activ i ty on 10/25/2013 without
prior knowledge of the FBI or Silk Road public keys. Furthermore, ou r sy st e m found clos e li n k s between
Silk Road and real users identiﬁed with our annotation s y st em .
Recently, several research studies [3, 2, 4] have suggested the potential privacy limitat i on s with bitcoi n
transactions.  investigates an alleged theft by leveraging external sources of information and combining
them with techniques such as context discovery and ﬂow analysis. [ 4] , on the other hand analyzes statistical
propertie s of the transaction graph to answer questions about ty pi cal user behavior, spen di n g/acq u ir i ng
habits, and ﬂow of bitcoins between multiple account s belonging t o the same user. Realiz i ng the need for
stricter privacy in the bitcoin graph, the authors in  suggest an exten si on to bit coi n that augm ents th e
protocol to all ow for fully anonymou s cu r re nc y t r ans act i ons .
4 Threat Model
4.1 Attacker Goal: Tie “real” nam es to transacti ons
The “real” name her e may be a person’s true name or username from an online publi c forum (or any other
public data source). The goal is to associate nu me rou s u nr el at ed cryptographic IDs with an act u al u ser .