Primary Market User Behavior Analysis Based On Cryptocurrency Transactions In The Ethereum Blockchain

Ethereum is an open platform based on blockchain,smart contracts can be deployed and run on it.Smart contracts greatly simplify the issuance of digital currency,and directly lead to the explosive growth of ICO projects around the world in 2017.ICO,which is known as initial coin offering,is a way of raising funds for projects based on blockchain in primary market.The founders will provide the white paper of their project,and issue a new kind of digital currency to raise funds in the form of bitcoin or other kinds of digital currencies.Compared with IPO in primary stock market,ICO is lack of regulation and easier to launch.There exists huge speculation in ICO,and some ICO projects are seen as economic crimes in recent years.Transactions of digital currencies are publicly stored on the blockchain,which makes it possible to analyze users' trading behaviors during the period of ICO.In this paper,digital currencies issued on Ethereum are selected and their transactions are used to build transaction network.Participants in the transaction network are identified and the distribution phenomenon of digital currencies are found using this network.Since addresses on Ethereum are anonymous,participants in the transaction network are identified for further analysis.Exchanges can be the hub nodes in the transaction network.Methods based on artificial features and graph neural networks are proposed to identify hub nodes.The experimental results show that both of them can well identify exchanges,wallets and token sales,and can reaches 87.5%on Macro F1.The transaction patterns of different types of hub nodes and the advantages and disadvantages of the two methods are also analyzed.Investment institution is one kind of the roles in ICO.For recognition of investment institutions on Ethereum,public information of investment institutions is collected and networks composed of investment institutions and projects invested by them are constructed,which can be used to analyze the blockchain investment market from a macroscopic perspective.Based on this analysis,heuristic method of inferencing investment institutions on Ethereum is proposed.Addresses found by this method can have high probability of big deals and similar invested projects with investment institutions' public information,and are close to the initial address of digital currency.These characteristics make them look like investment institutions.Distribution is one of the ways to speed up the circulation of digital currency.After determining the scope of ICO,tree structure is used to represent the circulation process of digital currency during the period of ICO,and distribution phenomenon is found using the tree structure.By defining the multi-level distribution index of digital currency,it is found that distribution is common,but the degree of distribution varies greatly,and the significant multi-level distribution phenomenon is rare.By analyzing the position of investment institutions in the circulation process,it is found that the influence of investment institutions on token circulation is relatively small,while in few cases they can act like a direct or indirect distributor.The idea of analyzing trading behavior of this paper can be applied to early warning of economic crimes,which usually have fixed or common transaction patterns,while the patterns are hidden in the transaction data.It's of great significance to the regulation of ICO and the maintenance of a good ecosystem of the digital currencies.