Top French quant: Best mathematicians are in France, or Russia
Raphael Douady has seen a lot of young quants. Currently a research professor in Paris (at University of Paris I: Panthéon-Sorbonne), he previously spent four years as a professor of quantitative finance at Stony Brook University in New York and as a visiting professor at New York University. He's worked in quantitative finance on both sides of the Atlantic, and after a career spanning nearly three decades, he's reached a notable conclusion: the mathematicians coming out of France are of a far higher standard than almost all their international counterparts.
"The general level of education and mathematics in France is extremely high," says Douady. "Stony Brook is one of the best universities in the U.S. and yet I was teaching my students there notions that you will learn in high school in France. The only place I have seen the same level of mathematical education is in Russia."
We already remarked upon the prevalence of Russian quants in banking when war broke out in Ukraine earlier this year. Douady confirms that when you look at the population of quants on Wall Street, 33% tend to come from America, 33% from France, 15% from Russia and the remainder from places like China. The UK has a "few good universities" in the form of Oxford, Cambridge, Imperial and Warwick, says Douady, but the general standard of mathematical education in Great Britain is simply not the same.
While the standard of overall mathematics in France is superior, Douady says there's one area in which France falls behind: statistics, and by implication data science. "France sees statistics as a branch of mathematics, when in fact it is a science in the same way that physics is a science that uses mathematics to work," says Douady. "There is a strong interaction between math and physics, but physics is not a branch of math."
French quants' weakness at statistics matters, but not that much. French mathematicians have such a broader grounding in math that they can adapt very fast, says Douady: "When you start looking at things like machine learning, the French kids who have learned high level functional analysis and geometry can pick it up at a speed that no American can copy."
Douady himself specializes in machine learning and data science. Working with his PhD student, Thomas Barrau (currently a quant researcher at Axa IM in Hong Kong), he's just released a book on Artificial Intelligence for Financial Markets which proposes a new AI technique based upon multiple non-linear univariate models instead of traditional multivariate regressions. Known as Polymodels, Douady says the technique is particularly suited to highly uncertain markets: in a traditional approach, quants would look for correlations between their investments or portfolio with a limited number of risk factors (typically five to 10). Under the Polymodel approach, they're able to take anything from 100 to 1,000 such factors into account: the system generates multiple models, which can be referenced according to the situation. A machine learning layer helps identify the most relevant signal.
Douady has been refining this approach for years. In the early 2000s, he says he was asked by a hedge fund asset allocator to find a way of predicting fund risk from historic returns. He attended various meetings with the French guy who was "a pioneer in the hedge fund industry" and noticed that most of his time was spent discussing one-off events. "He spent five minutes talking about Sharpe Ratios and volatility correlation, and 55 minutes talking about what happened on a specific date when peers lost money. - Which investments did the fund liquidate first etc.?" Douady says classical risk models were too restricted to encompass the necessary breadth of information.
He says current markets are at a "critical moment' for quantitative risk managers, but that the quant risk profession is always wrought with contradictions: "Even if I am telling you there's a 25% risk of a market crash as big as 2008, there's still a 75% chance that this won't happen."
Quants in France and globally need to familiarize themselves with machine learning techniques, says Douady: "Machine learning is invading the entire space." Equally, whether you want to work on the sell-side (in a bank) or on the buy-side (in a fund), he says it will help to understand the kinds of complex mathematical techniques like differential equations, differential geometry, algorithms, control theory, and harmonic analysis that French quants have such a good grounding in.
Hedge funds are increasingly using quants to optimize trade execution and simply make markets, he claims: "The big hedge funds have two main functions: one is to find the right bet to make money in the markets, the other is simply to create liquidity and to be a market maker. They will never tell you the details, but I can tell you absolutely that they are market making too."
Have a confidential story, tip, or comment you’d like to share? Contact: firstname.lastname@example.org in the first instance. Whatsapp/Signal/Telegram also available (Telegram: @SarahButcher)
Bear with us if you leave a comment at the bottom of this article: all our comments are moderated by human beings. Sometimes these humans might be asleep, or away from their desks, so it may take a while for your comment to appear. Eventually it will – unless it’s offensive or libelous (in which case it won’t.)