In the last two decades, the increasing number of shocks and financial crises has been a major issue for the financial risk management teams.
Among the wide range of exercises in this field, Stress tests have become a main guideline for the regulator in order to assess the banking system resilience against the realizations of various categories of risk (market, credit, operational, climate, etc). The main challenge is to simulate unfavorable extreme (but plausible) negative returns similar to a historical dataset.
This is an unsupervised learning problem: Given real data from stock market indexes that will act as a train dataset, the task is to learn a generative model that simulates synthetic stock market indexes.
The difficulty mainly lies in two points : a) find an appropriate transformation to generate extreme values for some given input noise, b) learning a generative model based on a dataset which contains very few extreme data.
The competition will begin the 25th of October and will end the 22nd of November.
There will be a Kick-off presentation on the 15th of October with the different stakeholders from BNP Paribas and Ecole Polytechnique.
Competitors will be organized in teams of 3-5 students (ideally from different universities and different countries).
The teams will be coached by 5 mentors: from Phds/Postdocs to experienced Data Scientists from BNPP.
There will be a one week delay between the end of the competition and the award ceremony.
After a presentation in front of the jury of Ecole Polytechnique and BNP Paribas,
The winner will receive the jury prize, worth several thousand euros
As well as get a chance to discover BNP Paribas’s "métiers"during an exclusive event, and look out for their dream job.
English will be the official language during the whole competition.
This competition is open to students only. The competition winners will need to prove their enrollment to a university program.
Final teams of 3-5 people. They can be built throughout the first two weeks of competition thanks to the dedicated Discord forum. Once a submission is made, members cannot leave and form a new team.
Every week a new private dataset will be used as evaluation.
Once a week, competitors may submit their model/outputs to update the leaderboard on a validation set.
Moderators may check the code of participants after each submission. If the code doesn't run or if the results are not reproducible, the score will not be taken into account.
On the last day of the competition, a final ranking will be established on a test set. The top 5 teams will be invited to present their work through a 15 minutes oral presentation at the award ceremony. The award ceremony will be either in the Paris region or fully remote depending on the health situation.
Using models from a public source is allowed, as long as the reference is cited.
Prizes: 3 Jury prizes, to be announced later.
For students with ECTS (must be validated by their program director): at the end of the competition, each team must submit both its code and a pdf report to present its work (1-3 pages).