The American press awarded you with the nickname of "dissident Wall Street" while your excellent book "Fooled by Randomness" was released. You defend the theory that chance plays a role in the performance of financial players that is more important that one could figure, and that returns are not always the result of the trader or manager’s talent. What are the main shortcomings that may be encountered in the practice of financial engineering in its broadest sense, as practiced today in investment banks or hedge-funds?

In general, the main encountered problems are: the use of portfolio theory, the trust we give to the laws of Gauss/Poisson, and reducing the concept of uncertainty in the single parameter that is variance. If you do a test in this framework, you will see that these methods can be successfully used to describe past events, but that they are useless on the overall - what is called "over fitting." They fail "out of the sample."

If you try to replicate what happens in real life using portfolio theory, and you adjust the settings to... say, 1999 and then you look at the risk and return between 1999 and 2000, then you adjust the settings again until 2000, and look at the error between 1999 and 2000, and so, you may be unpleasantly surprised. This does not work at all. It’s a bit for us what astrology was to our ancestors: it sounded true, it seemed to make sense, but it did not apply in any case in real life, apart from “narrative field of calibration”. This so-called "science" of economics, is really only crudely unscientific and non-empirical.

I truly believe that the practice of finance should work as medicine, that is to say based on empirical evidence, and that it should be the least prone to excessive theorizing. This apparently craft approach, would certainly be "more scientific".

Nobody seems to have been intrigued by these issues, except perhaps the famous Benoit Mandelbrot and myself, despite our independent intellectual origins.

In your book you tell the story of a young trader who for several years, has earned his bank millions of dollars, before losing in one day more millions than he had made in his career (a bit like Amaranth Fund). And you said that this is the exact archetype of bad traders. What do you think is the typical profile of a successful trader? By definition, is not a trader someone who earns his employer money by sometimes taking big risks, including precisely the risk of losing more than he has ever won in his whole life?

Bingo! In my point of view, a good trader is someone who is not always "short" on bearish positions. The track of his P & L would thus be more revealing of his true skills, especially if he is "long" on volatility. There are two types of traders: the "concave" (or "short" on large deviations) and the "convex" (or "long" on large deviations).

There is also a category of strong traders: a "market-maker" shows his skills more quickly because it is subject to natural limitations related to his business, with a high frequency of transactions, and a strict discipline in managing non- mathematical risks.

Regarding your personal career, what lessons have you learned from your years on the trading desks, beyond the significant presence of chance and luck as described in your book?

I think I understood that we have not yet pinpointed the world around us and that we are still very vulnerable to so called sciences. I learned so much from the time I spent on the desks that I have generalized it to the scale of the whole history. I have thus developed a theory called "the ludic fallacy", or the playful error, showing that the chance that we have in our everyday life is not the one described in books as it has nothing to do with a game. In real life indeed, the rules are particularly blurred, and way too imprecise to be well mathematically modeled. That is why we have errors that can greatly exceed those that would be expected by our models. In addition, games are Gaussian, while economy is “Mandelbroty”.

Finally, how do you see the development of the financial industry, particularly derivatives and what do you think should be the next lines of research to which practitioners will have to move?

I think we should turn to the Mandelbrot fractals as a new framework for thinking. They cannot provide precise answers to everything, but that’s true science, as true as possible. We already are almost 50 years late: we must catch up!