›  Opinion 

The «French Quants» must relearn to code!

In France, many financial engineers feel some aversion for IT. Some, fascinated by models, don’t consider for a second writing thousands of lines of code…

Article also available in : English EN | français FR

With the development of statistical arbitrage, product innovation and the sophistication of financial models, having reliable, robust and efficient computer systems has become a major issue for financial institutions, including banks and hedge funds. The quants, who up until now were happy enough quickly carrying out their mathematical models in “bookstore pricing”, have become part of a gigantic puzzle of race for performance.

And if historically, it’s indisputable that the quants have always been developers, the coming years are likely to accentuate this trend and to permanently democratize the profile of the ‘geek quant‘: an engineer with a high enough level of math to understand the probabilistic theory, but having as much, or more computer programming skills to detect and decipher the optimization pockets of his software programs.

In France, academics don’t seem to have taken this trend into account. Programming courses in the Masters program of Applied Mathematics, which provides the bulk of quants in the job market, typically represents only a small portion in light of all the lessons learned.

In France, for lack of information or simply out of pure idealism, financial engineers out of school have an aversion to computers.

For example, the master of the University of Paris 6 directed by Nicole El Karoui, one of the best Masters programs in Paris, offers only 18 hours of optional introduction to C++, on the sidelines of the official course and as part of a ‘refresher’. The East Marne-la-Vallée Masters University lead by Damien Lamberton, offers a good 3 hours of IT per week during the first trimester, but it’s at the whim of the student volunteers. And in both cases, none of these courses is subject to a final exam which takes into account the calculating of the year-end average.

There are numerous explanations for this. The first is that the Masters in mathematics, which has become the Masters of research, are primarily trained in mathematics and not in quantitative finance. This distinction is relevant because this type of training has historically been intended to guide students to academic research, which means that they’re obliged to fultfill a certain number of criteria in terms of knowledge and theoretical mathematical tools.

Some students are so fascinated by the models and their conceptual beauty that they don’t even think for one second of writing thousands of lines of code and spending endless hours debugging programs.

The second reason is that French academics are relatively conservative and, unlike their Anglo-Saxon colleagues, the transition from university to business or the reverse isn’t the norm for them. The best-known cases, like Peter Carr (PhD in finance, head of research at Bloomberg and Director of the Masters program of the Courant Institute in New York), Emanuel Derman (PhD in physics, director of quantitative research at Goldman Sachs until 2002 before taking the direction of the Financial Engineering Masters program at Columbia University) and Mark Joshi (PhD in mathematics, Head of the risk team at RBS until 2005 and currently a professor at the University of Melbourne), are there to certify that the mixture of genres doesn’t unduly disturb the Anglo-Saxon universities.

This symbiotic relationship between academic research in applied probability and the financial industry has allowed large American institutions to match an increasing and varied student demand (physicists, mathematicians, and IT specialists) and the wish of recruiting financial engineers with polyvalent profiles (able to understand finance, do math and program robust programming tools). The emergence of hedge funds in the U.S. and needs in terms of performance arising from their activities have brought focus on the unavoidable importance of new technology in finance.

In France, for lack of information or simply out of pure idealism, financial engineers out of school have an aversion to computers. Some students are so fascinated by models and their conceptual beauty that they don’t imagine for one second of writing thousands of lines of code and spending endless hours debugging programs. This has an obvious effect on the level of candidates in job interviews. The market reality is however much duller: the academic-style type of quantitative research in the trading room in France is certainly one which offers fewer jobs compared to other financial centers like London or New York.

the quants of tomorrow will be more than ever, without a doubt, "informathematicians" !

And contrary to prevailing thought among future quants, financial models aren’t required, to be considered ‘functional’, to contain all the possible scenarios. Their use is above all an eternal compromise between their ability to quickly give a more or less ‘realistic’ view of the market at the time ‘t’ and the efficiency with which they do it when there’s a lot of information. This efficiency is largly dependent on the robustness of the valuation tools and the power of the machinery they use during calibration and pricing. The effect of a certain standardization trend is that the adjustments and mathematic innovation on the models will be less and less intended to be manufactured.

Implementing a stochastic differential equation and calculating expectancy is undoubtedly within reach of many quantitative engineers, but integrating this body of mathematics in the context of a business the size of a trading desk is undoubtedly more difficult than is often thought once one has their diploma in hand. Because in this context, databases, tools dedicated to pricing and trading systems dominate in the chain of profitability. A place that, in the future, will certainly be much more important than mathematical models.

As a consequence, it won’t be enough for future quants to simply know how to translate algorithms minimizing stochastic spreads in a few lines of code. They’ll now have the recoil of a computing perspective to effectively develop their models on robust, intelligent and efficient pricers, and be able to provide a final rendering of transparent and ‘user friendly’ software for users. The profile of a mathematics researcher with ‘pen and paper in hand’, to borrow a phrase from Daniel Duffy, will inevitably give way to that of the experienced developer capable of indiscriminately taming mathematical algorithms, parallel programming and interface development. The quants of tomorrow will be more than ever, without a doubt, "informathematicians!"

Yann Olivier , February 2010

Article also available in : English EN | français FR

Read also

March 2007

Profile Quantitative Analyst: a dream career for young mathematicians

Each year, more and more young engineers or science graduates are interested in this career. But how does one become a “quant” ?

Share
Send by email Email
Viadeo Viadeo

Comments (0 contribution EN - 12 contributions FR )

Focus

Opinion Psychology and smart beta

‘Smart beta’ sounds like an oxymoron. How smart can it be to continue using the same strategy in such fickle markets? A portfolio manager calling on all his skills (‘alpha’) in analysing market environments (the source of ‘beta’) should be able to outperform an unchanged (...)

© Next Finance 2006 - 2019 - All rights reserved