Topological Data Analysis of Tranaction Data and Twitter Data for Crypto-Assets
| Reference No. | 2023a010 |
|---|---|
| Type/Category | Grant for Project Research-Short-term Joint Research |
| Title of Research Project | Topological Data Analysis of Tranaction Data and Twitter Data for Crypto-Assets |
| Principal Investigator | Yuichi Ikeda（Graduate School of Advanced Integrated Studies in Human Survivability, Kyoto University・Professor） |
| Research Period |
May 25,2023. ～
May 26,2023. August 9,2023. ～ August 10,2023. |
| Keyword(s) of Research Fields | Blockchain, Crypto-asset, topological data analysis, network science, transaction data, twitter data |
| Abstract for Research Report |
(1) As the next technology after web2, web3, which positions blockchain as the core technology, has been proposed. Blockchain technology is essential for the realization of a new digital economy, and distributed ledgers and crypto-assets based on blockchain technology have the potential to revolutionize the conventional economic system. However, due to anomalies in crypto asset transactions, such as money laundering and fraud, there still exists a negative view of the new economy based on blockchain technology. (2) In this study, we will investigate the fundamentals for detecting anomalies such as money laundering and fraud and predicting price spikes by making full use of network science, topology, machine learning, and other techniques. (2) We will show that the problems of crypto-asset transactions can be suppressed by the power of mathematics in order to create a "new economic system" by fully utilizing the advantages of distributed ledgers based on blockchain technology. (3) We establish the following three working hypotheses regarding crypto-asset transactions. (Hypothesis 1) Anomalies such as "money laundering" are hidden in the time trends of regular transactions, high-value transactions, and transaction chains through multiple exchanges from the trading network of crypto assets Kondor (2014). (Hypothesis 2) Anomalies such as "insider trading" are hidden in the relationship between price fluctuations and information during the conversion of crypto assets into legal tender and inflows into USD-pegged stablecoins such as USDT and USDC Balcilar (2017). (Hypothesis 3) Anomalies such as "price spikes" are hidden in the relationship between the polarity and other properties of information such as news and Twitter and the price movements and trading volume of crypto assets before that information flows Rognone (2020), Kim (2016), Abraham (2018). (4) With respect to these three working hypotheses, after conducting a literature survey of various methods for detecting anomalous transactions, we have used the following methods for crypto assets such as Bitcoin, ETH, and XRP: network analysis [1] and Hodge decomposition [4] for the time variation of transactions, random matrix theory [3] for the time variation of prices, and correlation tensor analysis of embedded vectors [2], and Topological Data Analysis using tf-idf for textual information such as news and Twitter for crypto assets to investigate the relationship between market transactions and prices of crypto assets. Finally, we examine anomaly indicators for the crypto asset market by integrating the validated detection methods through machine learning [5]. (5) The actual data of crypto-asset transactions are registered and published on the blockchain. However, information on market participants and small transactions that are not registered on the blockchain is not public and is held only by the companies that operate the exchanges. In order to improve the accuracy of anomaly detection, it is necessary to consider the use of such information. We will discuss the possibility of utilizing actual data from exchanges in the future, as well as the ripple effects on CBDC (Central Bank Digital Currency) and local currency transactions, while considering the actual conditions of exchanges, with the participation of relevant parties who are familiar with practical business operations. |
| Organizing Committee Members (Workshop) Participants (Short-term Joint Usage) |
Akihiro Fujihara（Chiba Institute of Technology・Professor） Yoshimasa Hidaka（High Energy Accelerator Research Organization (KEK), Institute of Particle and Nuclear Studies・Professor） Abhijit Chakraborty（Kyoto University・Assistant Professor） Yuichi Ike（The University of Tokyo・Assistant Professor） Yasushi Nakayama（SBI Financial and Economic Research Institute・Senior Researcher） Tomoyuki Shirai（Kyushu University・Professor） |