AI: Regime Switching Models in Finance

5-th year project 2019-2020 • ESILV • PI²5 ►https://bit.ly/2L323US
?

Deep Learning is a field of study of artificial intelligence that allows automatic learning based on an extensive database. It has only recently (2010) been democratised in more and more areas, namely market finance. It helps to evaluate the risks that a company would take before making an important decision. Or the prediction of a company’s share price, which was the subject of our project. The issue of our project is how to build a long term strategy on regime-switching models, such as signal timing, using deep learning.

In our organisation, task distribution and planning (Deadline) were crucial. Furthermore, we used several platforms to communicate and exchange (Messenger, Github). Moreover, we imposed a meeting every Tuesday to give feedback on our progress. We thought it was essential to work in parallel between the progress of our code and the whole theoretical part of our project. That’s why we preferred lengthy analyses before acting.

Thus, our project started with a work of learning and storage. To fill this gap, we did a lot of research in the field of Machine and Deep Learning, which would enable us to propose mathematical concepts.

Our research has mainly covered the field of statistics as well as probabilities. We were able to apply these new concepts during our tests, primarily concerning moving averages, as well as the Markov regime in the American stock market. Once we completed the mathematical research for our project, we also had to improve our skills in the field of programming (Python). So, we had to learn how to use the PyTorch library. 

The study of this open-source deep learning library required a lot of effort from us to research and understand predictive modelling with deep learning in finance. We were able to optimise our algorithms and future Machine Learning prediction programs.

More about ESILV:

Facebook : http://facebook.com/esilvparis
Twitter : http://twitter.com/esilvparis
Linkedin : http://bit.ly/25WVOCa
Pinterest : http://pinterest.com/pleuniversitair
Google+ : https://plus.google.com/+esilv

http://www.esilv.fr