Order book modeling, Market Making applications.

Summary This thesis addresses different aspects of market microstructure modeling and market making problems, with a particular focus on the practitioner's point of view. The order book, at the heart of the financial market, is a complex high-dimensional queueing system. We aim to improve the knowledge of the LOB for the research community, propose new modeling ideas and develop applications for Market Makers. In particular, we thank the Automated Market Making team for providing the high-quality high-frequency database and a powerful computational grid, without which this research would not have been possible. Chapter 1 presents the motivation for this research and summarizes the main results of the different works. Chapter 2 focuses entirely on the LOB and aims at proposing a new model that better reproduces some stylized facts. Through this research, not only do we confirm the influence of historical order flow on the arrival of new ones, but a new model is also provided that replicates much better the dynamics of the LOB, in particular the volatility realized in high and low frequency. In Chapter 3, the objective is to study market making strategies in a more realistic context. This research contributes to two aspects: on the one hand the new proposed model is more realistic but still simple to apply for strategy design, on the other hand the practical Market Making strategy is much improved compared to a naive strategy and is promising for practical application. High frequency prediction with deep learning method is studied in Chapter 4. Numerous results of the 1-step and multi-step prediction found the non-linearity, stationarity and universality of the relationship between the microstructure indicators and the price change, as well as the limitation of this approach in practice.