Dear community,
We're a team of researchers interested in infometrics, chainalysis, or colloquially analysis of blockchain data. I've made this account to post about our recent research on analysis of NFT trading networks and our findings in detecting potential illicit activity in NFT trading networks.
This post is primarily aimed at providing educational material regarding NFT's and the authors are not blaming any particular NFT, creator, or project.
Traditional financial markets enjoy a significant presence of regulators enforcing a sea of illegal activity. However, most enforcing is trivial as user information is disclosed to regulating bodies linking each account to a real identity. Algorithms for detecting such patterns are not directly applicable to decentralized markets where only a pseudonymous identity is available. In other words, a user can have many wallets, and unless a centralized KYC platform is used, linking wallets to users is difficult. Prior research concentrated on examining NFT trading graphs to find recurring patterns that would indicate illegal activity. However, the ability to create wallets at fiat makes these efforts inadequate.
Our method takes a different approach. Here's how we did it.
Our Approach: Linkability Network
The core of our method lies in the integration of NFT trading graphs with the entire Ethereum Transaction Network, creating a Linkability Network. Here’s how it works:
- NFT Ownership Traces: We track the historical transfers of ownership for each NFT, capturing every instance of trading or transfer.
- Ethereum Transaction Network: We extract and map the entire transaction history of Ethereum accounts involved in NFT trading, focusing on direct transfers of ETH between accounts.
The result is a directed graph that highlights potential collaborations or shared ownership across wallets that engage in wash trading. These connections, revealed through normal Ether transactions, indicate a high likelihood of coordination between wallets, even if no direct NFT trade is observed.
As commonly said, a picture tells a thousand words.
https://preview.redd.it/2ak5wznkhltd1.png?width=794&format=png&auto=webp&s=c9555a299e406754f015e8bb8f516f40ac4b3091
The images above are examples of NFT trading graphs that our algorithm computes. The trades where the NFT was sold are edges colored red; the green edges are transfers between two accounts, and finally, the purple dotted lines are accounts historically linked on the Ethereum network. The proposition is simple: two or more accounts buying/selling/transferring the same NFT between them while having had transactions between in ETH is hardly a coincidence.
This usually happens when two or more wallets owned by the same entity/person continuously buy the same token from one-another. However, they can only do this if they have enough liquidity, or they send the ETH back and forth to fund the next buy.
The left graph shows how the first 9 accounts are all interlinked on the Ethereum transfers. These wallets were all participating in trading the token, resulting in a significant price increase. Finally, the last 3 trades are likely honest buyers that speculated on the continuation of the price action and are now holding the token, to this day, unable to sell due to lack of buyers.
Similarly, the graph on the right illustrates the same pattern amplified both in terms of value as well as trades.
The likability network was computed by following all transactions of each participating wallet until depth 4. This took a lot of time and significant computation power. For those interested, the data of cca. 135k NFT's and their graphs is publicly available, as is the paper that goes into more detail.
In conclusion, it is unfortunate that many of the market participants were unaware about these practices. Hopefully, this post will help raise awareness about these practices for those of you trading NFT's.
