Laboratory: GIPSA-lab and CERAG
Dates: 5 months starting in February or March 2020
Internship location: Gipsa Lab
Salary: monthly stipend of 550 euros net (in accordance with french regulation)
Funding: Cross Disciplinary Program (CDP) Risk, Initiative of Excellence (IDEX Univ. Grenoble Alpes)
Name and status of the supervisors:
Isabelle Girerd-Potin (Professor of Finance, CERAG, team Anticipation et gestion des risques)
isabelle.girerd-potin@univ-grenoble-alpes.fr
Didier Georges (Professeur of Automatic Control, GIPSA-lab, team Infinity)
didier.georges@univ-grenoble-alpes.fr
 

Title of the project:

Discovering Mutual Funds Portfolio Holdings to increase market transparency and reduce market systemic risk
 

Objectives:

The project aims at improving the transparency and the supervision of mutual funds, using an objective and easily available information, namely mutual funds returns. Mutual funds are portfolios managed on behalf of investors by professionals in banks or asset management companies. The regulations partially contribute to transparency. For example, in Europe, the directive 2009/65 compels the disclosure of the funds holdings every 6 months with a delay of 2 to 4 months.
The aim of the project is to remedy an insufficient frequency of disclosure of the portfolios asset composition and to extract from the returns, information about the portfolio and the manager’s behavior. From the view point of system and information sciences, the Master 2 student will have to manage financial data and to implement new algorithms devoted to dynamically estimate the portfolios compositions. We already proposed a non-linear observer, similar to a Kalman filter (see Georges and Girerd-Potin, 2017). However, this approach has been tested and validated on a small simulated portfolio. The objective is now to enlarge the asset universe working on real mutual fund portfolios and to compare the approach in (Georges and Girerd-Potin, 2017) to a machine learning approach (use of neural networks with supervised learning).
 

Références

[Byrd et al, 2019] Byrd, D., Bajaj S., Balch T. H.  Fund Asset Inference Using Machine Learning Methods: What’s in That Portfolio?  The Journal of Financial Data Science, Summer 2019
[Georges & Girerd-Potin, 2017] D. Georges, and I. Girerd-Potin, A Discrete-Time State Observer Approach to Discovering Portfolio Holdings, IFAC World Congress, Toulouse, France, IFAC PapersOnLine 50-1 (2017) 946–951.
 [Georges, 2020] D. Georges, Machine Learning for Receding Horizon Observer Design: Application to Traffic Density Estimation, submitted to IFAC World Congress 2020, Berlin, Germany, July 2020.
 

Requested domains of expertise:

At least two of the three fields: computer science, optimization including machine learning, finance. Examples of expected curriculum: Master 2 in Finance with a background in applied mathematics/computer science; Master 2 in EEA interested in finance; double degree in Quantitative Finance (Management and Engineering).

How to apply?

Send a cover letter, a CV (highlighting your level in English) and the list of Master 2 lectures followed with grades/rankings when available to isabelle.girerd-potin@univ-grenoble-alpes.fr and didier.georges@univ-grenoble-alpes.fr