Deep Semi-supervised Fraud Detection in Derivatives Market

In the previous projects with TMX some unsupervised techniques have been explored to detect anomalies in trading for the derivative market. All have their pros and cons but they all have provided promising results. The first objective of this collaboration with TMX is to treat all these techniques that have been explored in previous projects; contrast them and integrate them in order to develop an improved pipeline for anomaly detection. The second goal is to add the use of labeled data to this pipeline. TMX disposes of a large set of labeled manipulative examples which will be very useful to develop a semi-supervised methodology.

Faculty Supervisor:

Manuel Morales

Student:

Partner:

Bourse de Montréal

Discipline:

Mathematics

Sector:

Finance and Insurance

University:

Université de Montréal

Program: