First prize

• Hannah Maidment
  Thesis: "Hawkes Process-Driven Models for Limit Order Book Dynamics"
  Master : MSc in Mathematical and Computational Finance, University of Oxford
  
  "In my thesis, I consider the use of Hawkes processes to model the arrival of orders in a limit order book and show that they effectively capture the clustering behaviour observed in limit order book dynamics. In a world where electronic, order-driven markets are becoming ubiquitous, the advantages of developing a meaningful model for the dynamics of a limit order book are extremely broad. These range from gaining clearer insight into the role of supply and demand in price dynamics to informing the design of electronic trading algorithms and optimal execution strategies." 

Second prize ex æquo 

• Rémy Vuillet
  Thesis: "Rough Volatility: Simulation and Calibration", realized at BP London
  Master : MSc in Mathematical and Computational Finance, University of Oxford
  
  "My thesis focuses on rough volatility models, which provide more realistic volatility dynamics for stock prices. First, I look at the simulation of the Rough Heston model and I show the tension between arbitrarily rough volatility and unbiased option prices. Secondly, I improve the benchmark for the calibration of these models with a neural network. Finally, I propose a straightforward method to estimate the regularity of time series using the signature transform from rough path theory."

• Thomas Hengstberger
  Thesis: "Increasing Venture Capital Investment Success Rates Through Machine Learning"
  Master : MSc in Mathematics and Finance, Imperial College
  
  "The thesis explores the use of machine learning models to identify successful startup investments for the purposes of Venture Capital (VC) investing. The process of how VC investments generate returns is formalised and then translated into a binary target variable, which hinges on a future investment entry point and a subsequent investment exit point.  To imitate how predictive models might be used in a commercial setting, three machine learning models (Logistic Regression, Random Forests and Extreme Gradient Boosting) are trained using historical data from Crunchbase, while their performance is evaluated based on the predictive accuracy on unseen data. The resulting models are able to correctly identify successful startup investments in 46-52% of cases within a 4-year outcome period, which compares favourably to the typical VC 30% success rate."

2020 Laureates

 

From left to right: Jean Cheval, President of the Natixis Foundation for Research & Innovation; Gilles Pagès, Director of the Probabilities and Finance Master's Degree at the Université Pierre et Marie Curie Paris 6 and École Polytechnique; Olivier Pironneau (on screen), Académie des Sciences and member of the scientific committee of the Natixis Foundation; Mathieu Rosenbaum, professor at the École Polytechnique and winner of the 2020 Louis Bachelier Prize; Nicole El Karoui, founder of the Probabilities and Finance Master's Degree at Pierre et Marie Curie University Paris 6 and École Polytechnique; Pierre Gasnier (on the screen at the top right), winner of the 2020 Best Master's Thesis Prize, Université Pierre et Marie Curie Paris 6 and École Polytechnique; Charles-Albert Lehalle (on the screen at the bottom right), Capital Fund Management and member of the scientific committee of the Natixis Foundation; Lukas-Benedikt Fiechtner, winner of the 2020 Best Master's Thesis Prize, University of Oxford; Michel Crouhy, President of the scientific committee of the Natixis Foundation.

• Pierre Gasnier
  Thesis: "Mean Field Game Theory for Gas Storage Valuation", realized at BP London
  Master: M2 “Probabilities and Finance”, Pierre and Marie Curie - Paris VI University and École Polytechnique
  
  "The subject of my work was to solve an optimal trading problem where the price dynamic depends of the actions, buying or selling, of the market agents. Mean field game theory simplifies the problem considering there is an infinite number of agents, but they have individually an infinitesimal impact. In my thesis, I show that the state of equilibrium of this game is not only the solution of a partial differential equation but also is the limit of a series of games where the market agents adapt naively their strategies to the previous games, giving us therefore two different numerical methods to solve the problem."

• Lukas-Benedikt Fiechtner
  Thesis: "Risk Management with Generative Adversarial Networks"
  Master : MSc in Mathematical and Computational Finance, University of Oxford
  
  "In my work I apply Generative Adversarial Networks, a novel machine learning technique, to generate artificial stock price trajectories whose statistical properties closely match those of real financial time series. Leveraging this generative power one can greatly enlarge the dataset available to estimate risk measure such as Value-at-Risk. An out-of-sample backtest shows that this methodology is competitive with traditional Value-at-Risk estimation methods such as historical simulation. The work also paves the way for future applications of GANs in financial mathematics. For instance, one could backtest trading strategies using artificially generated data."
  
  

Previous laureates by year