Anomaly bias Stock Trades and Order Dataset

The datasets used herein (trade.csv, orders.csv) are representative of such resources, originating from the detailed logs generated by an exchange's electronic trading system, capturing order lifecycle events and trade executions with high-fidelity timestamps and unique identifiers. While obtained through these open channels, the data adheres to strict privacy protocols, employing anonymized identifiers for participants. Therefore, throughout this study, we treat all identifiers (MEMBER_CODE, CLIENT_ID, TRADER_ID) as unique labels representing distinct entities within the dataset's ecosystem. This approach allows for rigorous analysis of trading patterns and network structures while respecting the confidentiality requirements and ethical usage guidelines.

The primary value of this dataset lies in its granularity and the combination of both order and trade information. The orders.csv file provides insight into the intentions of market participants – the prices, quantities, and timing of their desired trades before execution. The trade.csv file provides the outcomes – which orders actually matched, at what price, quantity, and time.

This data is especially valuable for researchers and surveillance professionals seeking to build fraud detection algorithms using real-world complexities rather than synthetic approximations. It supports a wide range of analytical methodologies, including graph-based cycle detection, participant network analysis, latency profiling, and behavioral clustering. With clean formatting, high fidelity, and clear attribute separation, the dataset is ideally suited for exploring how manipulative intent might emerge in high-frequency trading environments and offers a solid foundation for both academic inquiry and regulatory innovation.