Wash-trading Address Detection
Wash trading refers to the practice of artificially inflating trading volume by executing trades with oneself or coordinated entities. The Wash-Trading Addresses Detection task aims to identify addresses engaged in wash-trading activities within the Ethereum network. The EX-Graph dataset provides a graph representation specifically designed for this purpose, enabling researchers to detect suspicious addresses involved in wash-trading market manipulation.
The diagram above illustrates four typical wash trading patterns. Nodes A, B, C and D represent Ethereum addresses involved in a cycle of transfer activities.
Key Features
-
Comprehensive Data: The EX-Graph dataset includes a X graph and an Ethereum graph, providing a holistic view of the Ethereum ecosystem and its connection with social media.
-
Wash-Trading Address Graph: The dataset provides a graph representation for detecting wash-trading addresses within the Ethereum network. This feature helps researchers identify suspicious addresses engaged in market manipulation activities.
-
Anonymized X Matching: An anonymized X accounts file matched with Ethereum addresses is included, allowing for the analysis of numerical ID associations without compromising personal information.
Usage
Researchers can utilize the provided graph representation and anonymized X matching to detect wash-trading addresses within the Ethereum network. By leveraging the comprehensive data, researchers can gain insights into market manipulation activities and enhance the integrity of the Ethereum network.
For more details on how to use the EX-Graph dataset for wash-trading addresses detection, refer to the documentation and code examples provided.
Leaderboard
Without X Features
Rank | Method | Test AUC-ROC | Test Precision | Test Recall | Test F1 | Contact | References |
---|---|---|---|---|---|---|---|
1 | GAT | 0.8 | 0.8 | 0.59 | 0.68 | EX-Graph | Link |
2 | GATv2 | 0.8 | 0.76 | 0.68 | 0.72 | EX-Graph | Link |
3 | TAGCN | 0.8 | 0.76 | 0.67 | 0.71 | EX-Graph | Link |
4 | Cluster-GCN | 0.8 | 0.74 | 0.69 | 0.71 | EX-Graph | Link |
5 | DAGNN | 0.8 | 0.74 | 0.71 | 0.72 | EX-Graph | Link |
6 | APPNP | 0.8 | 0.74 | 0.7 | 0.72 | EX-Graph | Link |
7 | GCN | 0.79 | 0.81 | 0.56 | 0.66 | EX-Graph | Link |
8 | GGNN | 0.79 | 0.75 | 0.69 | 0.72 | EX-Graph | Link |
9 | GraphSage | 0.78 | 0.81 | 0.52 | 0.63 | EX-Graph | Link |
10 | DeepWalk | 0.54 | 0.54 | 0.59 | 0.56 | EX-Graph | Link |
11 | Node2Vec | 0.54 | 0.54 | 0.59 | 0.56 | EX-Graph | Link |
With X Features
Rank | Method | Test AUC-ROC | Test Precision | Test Recall | Test F1 | Contact | References |
---|---|---|---|---|---|---|---|
1 | GATv2 | 0.81 | 0.74 | 0.74 | 0.74 | EX-Graph | Link |
2 | Cluster-GCN | 0.81 | 0.75 | 0.72 | 0.74 | EX-Graph | Link |
3 | DAGNN | 0.81 | 0.75 | 0.73 | 0.74 | EX-Graph | Link |
4 | APPNP | 0.81 | 0.75 | 0.73 | 0.74 | EX-Graph | Link |
5 | GAT | 0.8 | 0.74 | 0.72 | 0.73 | EX-Graph | Link |
6 | GraphSage | 0.8 | 0.75 | 0.7 | 0.72 | EX-Graph | Link |
7 | TAGCN | 0.8 | 0.74 | 0.73 | 0.73 | EX-Graph | Link |
8 | GGNN | 0.8 | 0.75 | 0.72 | 0.73 | EX-Graph | Link |
9 | GCN | 0.79 | 0.82 | 0.56 | 0.66 | EX-Graph | Link |
10 | DeepWalk | 0.55 | 0.55 | 0.58 | 0.57 | EX-Graph | Link |
11 | Node2Vec | 0.55 | 0.55 | 0.58 | 0.57 | EX-Graph | Link |