PhD in Math/Computer Science in Europe -> Quant?

[Novak]

Hello, everybody!

This is my first post on this forum, and I realize this is a forum of a NYC-based quant community, but I still believe some of you could give me a piece of advice.

The question is, of course, if my education so far gives me a chance of becoming a quant.

Let me try to summarize my education: I was born in Belgrade, Serbia, Europe, where I obtained a university diploma at the Faculty of Mathematics as one of the top 5 students. Then, I joined a European Union - funded MSc program in Computational Logic (basically, math+AI). The program and the funding was highly competitive, and some 20 people from all over the world got accepted, so I obtained a MSc in Computer Science from Technical Universities of Vienna (Austria) and Dresden (Germany). I finished the program with honors, and I was awarded a Best Thesis Award 2007. In the meantime, due to the educational reforms in my home country, my university diploma in mathematics from Belgrade University is now acknowledged as a MSc diploma.

I continued my education in France, in a famous INRIA institute, where I do a PhD in Computer Science / Mathematics. I apply certain mathematical theories (category theory) to formal mathematical objects - formal proofs in classical logics, which is, in turn, connected to a termination of certain logical programs in computer science. (Oddly enough, a full-professor at the Faculty of Economics in Belgrade did his PhD in almost identical topic).

I am on my first year of my PhD, and I already have a record of 3 papers accepted to some important international conferences, I have world leading experts in the field of AI as coauthors in some of the papers.

Now, why would I think of becoming a quant, in the first place?

Well, I find the finance to be extremely interesting domain. I come from a family where both of my parents and my brother - they are all economists. Even my best friend. I have good knowledge and some experience of various mathematics, probabilities / stochastic processes, numerical methods, mathematical modeling, and of course, good programing skills. I lack background in finance, but I am thinking of reading Hull, Baxter & Renie, Wilmott, Joshi, Williams, and mastering the contents of their books before I finish with the PhD.

Again, the question is if I should still think of a career as a quant, and if so, should I get some sort of formal education in Finance? Also, can you tell me more about quant jobs in Europe (Britain, France, Germany, ...)? Do you know if the employment practice there differs form the one in the States?


Sorry for the long post.

 

[DominiConnor]

(Elsewhere I'm known as DCFC)

AI is a pretty broad field, some of which is directly applicable to finance, some isn't.

Classical logic is rarely applied to finance, indeed most quants are so wholly ignorant of even basic logic that on the CQF I have a remedial lecture on it, else C++ will bite them. Higher order predicate calculus won't help you much, and again the depth of ignorance is so great that to some will imply that you can solve PDEs.

Our guide has a reading list, it's long...

As I say to all CS PhD, you need to learn C++

Fuzzy logic has been applied to finance, indeed if you look hard you will see echoes of the garbage I did in the 1980s. But you should not just pity my folly, but try to find ways of using your skills on time series.
You will fail of course, but the journey is the reward. To even get to the point where you can judge that you've succeeded or failed, you have to understand a lot of things.
Also you need some way of adapting your ideas to a data stream that is so random many useful models assume there is no signal there at all.

 

[Novak]

@Chipolopolo:
hehe, of course, you are right, finance does pay lot more than a AI career. It is relatively easy, though, to get money if you do something related to life sciences, but that is boring for me.

@DCFC:

AI is indeed a broad field. I studied a lot of different stuff in the area, but the area where I had my own contributions is the area of Description Logics (DLs). It is a family of formalisms that underly reasoning on ontologies, semistructured data, semantical web, ... I have heard of a recent applications of DLs to Business Process Modeling (BPM), which is somewhat expected, since DLs can be employed to handle diagrammatic and hierarchical structures. I am not sure if a quant should be interested in that.

I have done some work, and I studied fuzzy logics intensively. I applied fuzzy reasoning when I created an agent (a program) that is able to play an arbitrary (sinlge/multiplayer, turn-taking/simultaneous) game given a game specification in a defined syntax. The game must be a complete information game, and the entire area is called General Game Playing. We used also some game theory there. However, one should not get too excited, there are serious limitation of this approach, mostly coming from requirement of the generality.

Now I also do a lot of category theory, but I am unaware of it's applications to finance (even though I have heard of papers being published on that).

There is a lot of other stuff that might be useful from what I'm doing, but I'd rather not go any further without knowing more of the problems that quants face on a daily basis.

The question remains - is my background suitable for a quant? I am quite experienced in programing in C++ (in fact, that was my language of choice for quite a while), and I can do intensive programing, but I wouldn't be too excited to have my job reduced to programing. After all, I think of my self as a mathematician. I can solve PDEs, but not with a Higher order predicate calculus! Also, is some sort of formal education in finance required for someone of my profile?

 

[DominiConnor]

Game theory has uses in quant finance. It's not a primary skill in what banks have on their job specification,s but the conversations I have with hiring managers put it as a useful skill.
Your subset will serve you well in interviews, but of course the practical applications are usually stochastic.
Pretty much all quants do some programming, the average seems to be 60% but be aware that you will have periods of 100%, and the distribution includes a significant chance that these periods are the norm.

Your career is a random traversal of a directed acyclic graph. Your halting states include being some sort of a developer, maybe a quant a developer, maybe a mainstream developer.

Agent based market microstructure is congruent with some of the work you might end up doing, certainly it is worth you reading some of this stuff, see if it fits.

Education will help of course, it's optimality depends on what sort of quant work you actually want to do.