Internship – M2 Applied Mathematics
As part of a POC, we want to study the feasibility of applying AI to trading, in the form of decision-support software, with a risk assessment.
As part of the Signal Processing team, in a stimulating and demanding environment, you will familiarise yourself with financial markets and contribute to the development of a demonstrator.
You’ll be working in a rewarding environment, as part of dynamic technical teams ready to share their knowledge (and learn).
The course will be run in partnership with Francesco Costantino, professor at the Institut Mathématique de Toulouse.
The aim of the internship is to contribute to a feasibility study of a trading decision support software on the price of financial assets, with risk control.
The proposed methodology is based on careful preprocessing of the training data (a priori using PCA or functional PCA + clustering), so as to reduce the size of the training data while retaining as much useful information as possible. This pre-processed data is then used to train a predictive AI, ideally returning a probability distribution of the share price trend. Patterns are then detected for which the probability of correct prediction is high. These patterns can then be interpreted (investor run-ups, corrections, collapses, etc.).
The feasibility study for a demonstrator in Python will be carried out in collaboration with the team.
- Review the state of the art in equity price modelling.
- Select and parameterise a stochastic market model to reproduce its statistical properties as closely as possible (for example, fractional Brownian processes or certain multifractal processes).
- Validate the non-(always) Markovian behaviour of the market.
- Select free data sources (preferably with volumes, and the order book if possible.) Build a dataset from these sources.
- Test the PCA + AI and functional PCA + AI approach on these data (define the right architecture), in order to predict the near future of the share price from the past.
- Any additional ideas are welcome.
You have a good level in mathematics and have completed 5 years of higher education (university, business school or engineering course in IT/market finance/statistics/applied mathematics) with an option in artificial intelligence. You are looking for an end-of-studies placement.
You have a good command of the Python language and have already carried out Deep Learning/Machine Learning projects. Experience in developing neural network architectures would be a plus.
- Good knowledge of proba/stat and the theory of stochastic processes.
- Good level of programming in Python.
- Dynamic and motivated, you are keen to learn and enjoy working in a team.
Working conditions and compensation
6 month internship. Salary to be determined.
Join our team
Top scientists, who are the soul of our company, are looking to expand their ranks.