Loading Events
  • This event has passed.

Anomaly Detection – What to Do without Enough Data – Oct 31, 2024

Session Description

October 31 2024 @ 12:00 pm - 1:00 pm

Abstract: Detecting anomalies in managed investment products is crucial for making timely investment decisions, yet limited data often hinders traditional detection methods. In this talk, “Anomaly Detection—What to Do Without Enough Data,” we explore innovative strategies to accurately identify significant changes in fund performance, operations, and portfolio composition despite data constraints.

We discuss various methodologies, including supervised and unsupervised machine learning, ensemble models, and a statistically driven rules-based approach. By evaluating the effectiveness of each method in low-data scenarios, we highlight how selective application or combination of techniques enhances detection accuracy. The project culminates in a robust model and process of identifying anomalies that promptly alerts analysts to noteworthy fund changes, enabling swift and appropriate adjustments to fund ratings. Future work involves integrating a human-in-the-loop (HITL) machine learning process to leverage analyst feedback for continual improvement in anomaly detection.

Speaker Bio: Quentin Bartram is a Quantitative Research Team Lead at Morningstar, where he collaborates with members of the Quantitative Track of the Morningstar Development Program (MDP) and Morningstar Internship Program to develop new and rigorous methodologies that empower investor success. He graduated from Monmouth College with a double major in Data Science and Mathematics and a minor in Computer Science.

During his tenure at Morningstar, Quentin has worked on a variety of impactful projects, ranging from spearheading anomaly detection for the fund research team to developing AI-powered notebooks for the sales team to enhance efficiency. He has also aided in the creation of an executive-level dashboard that leverages real-time data to provide insights into the company’s holistic coverage metrics. Quentin’s diverse experience highlights his commitment to leveraging technology and quantitative analysis to drive innovation in the financial services industry.

All are welcome to attend. Online Meeting.

Link to join the meeting:

https://utoronto.zoom.us/j/86194070409

Details

Date:
October 31 2024
Time:
12:00 pm - 1:00 pm
Registration Website:
https://carte.utoronto.ca/event/anomaly-detection-what-to-do-without-enough-data/
Scroll to Top