Legacy trade surveillance systems frequently suffer from several inefficiencies, e.g., a high number of false positive alerts, low alert quality and insufficient surveillance coverage. As a result, legacy systems do not only impose high cost related to manual alert handling and maintenance, but do also open windows for market abuse practices remaining undetected.
Following regulators’ call for improved market monitoring practices, Machine Learning and AI-based techniques offer a path towards less costly and more accurate trade surveillance. Therefore, we discuss in our whitepaper the following use cases for the application of Machine Learning in trade surveillance:
Model calibration: The process of calibrating surveillance models such that they effectively detect actual or attempted market abuse while reducing the number of false positives.
Alert Scoring: The application of statistical methods to automatically classify generated alerts for improved alert handling and investigations by compliance officers.
Alert Generation: The system-based process of identifying and reporting potential cases of market abuse based on the available data.
Integration of unstructured News Data: The system-based transformation and integration of unstructured news data into trade surveillance systems to improve alert outcomes.
Data Augmentation: The process of creating synthetic data by modifying existing data to improve the capabilities of surveillance models.
Moreover, our whitepaper sheds light on the risks, challenges, and benefits of applying Analytics and Machine Learning in the above use cases. By this means it provides the outline for the transition to modern AI-based technologies in trade surveillance.
Leveraging its front-to-back knowledge of products, markets and state-of-the-art technologies, LPA offers a comprehensive trade surveillance advisory suite with individual modules that can be flexibly selected and combined, depending on the individual needs of your institution.
In advising our clients, we offer our expertise as part of a trade surveillance health check, assist you during implementation, or support you as an AI/Machine Learning think tank to take your surveillance setup to the next level.