I'm a 4th year PhD student in Computer Science. My research interests are machine learning and data mining. I've interned at Microsoft research and Google and found it no more interesting doing research in IT. I have a little bit background on Finance. I've taken courses including financial engineering, stochastic processes and other finance courses. I'm wondering how should I start my career as a quant researcher in major IBs? What knowledge should I prepare (say courses) before interviews? I know there are lots of expert here. So any inputs are welcome. Thanks.

QuoteOriginally posted by: msresearchI'm a 4th year PhD student in Computer Science. My research interests are machine learning and data mining. I've interned at Microsoft research and Google and found it no more interesting doing research in IT. I have a little bit background on Finance. I've taken courses including financial engineering, stochastic processes and other finance courses. I'm wondering how should I start my career as a quant researcher in major IBs? What knowledge should I prepare (say courses) before interviews? I know there are lots of expert here. So any inputs are welcome. Thanks.Get an internship if possible. Even though you are a CS PhD, I would strongly advise you to study C++ (from MJ's books). People would also expect you to be good at algorithms, statistics etc. Revise all your courses before interviewing. A big component of Quant work is simply IT, so if you don't like it, you might be better off in academia.

You have some of the basics for algorithmic trading. Personally, I'd see if you could get transferred to MS research in Cambridge who have good links to some large banks.I'd look at signal processing, and of course you will need to learn how to program in C++.Sadly, it is the case that I've had to say to some CS PhDs that they actually just need to learn to program, and then move on to C++.I'd steer clear of TA, but Market Microstructure ought to be on your reading list.Be aware that the quant ends of several large banks don't count CS as a relevant PhD, at you will need to dodge those who want to suck you into a code monkey role.But you know all this stuff anyway don't you ?

QuoteOriginally posted by: veeruthakurQuoteOriginally posted by: msresearchI'm a 4th year PhD student in Computer Science. My research interests are machine learning and data mining. I've interned at Microsoft research and Google and found it no more interesting doing research in IT. I have a little bit background on Finance. I've taken courses including financial engineering, stochastic processes and other finance courses. I'm wondering how should I start my career as a quant researcher in major IBs? What knowledge should I prepare (say courses) before interviews? I know there are lots of expert here. So any inputs are welcome. Thanks.Get an internship if possible. Even though you are a CS PhD, I would strongly advise you to study C++ (from MJ's books). People would also expect you to be good at algorithms, statistics etc. Revise all your courses before interviewing. A big component of Quant work is simply IT, so if you don't like it, you might be better off in academia.Thanks but who is MJ anyway?

QuoteOriginally posted by: DCFCYou have some of the basics for algorithmic trading. Personally, I'd see if you could get transferred to MS research in Cambridge who have good links to some large banks.I'd look at signal processing, and of course you will need to learn how to program in C++.Sadly, it is the case that I've had to say to some CS PhDs that they actually just need to learn to program, and then move on to C++.I'd steer clear of TA, but Market Microstructure ought to be on your reading list.Be aware that the quant ends of several large banks don't count CS as a relevant PhD, at you will need to dodge those who want to suck you into a code monkey role.But you know all this stuff anyway don't you ?Thank you DCFC. With all due respect, I would like to ask this question anyway, why do most of you guys (if not all) think that CS PhD cannot program in C++? In my perspective,Computer Science department is the only place where you could learn systematic programming skills. Most of us mastered C++ in Junior year and wrote at least 10,000 lines for course projects. When I was young, I used to laugh at kids who were not in CS and tried to mimic programming.So there must be a reason, say the requirement of quant programming is different from other fields? In terms of design patterns or other stuff? Anyway, I would like to listento your experiences and make it very clear before I move on.Thanks. I appreciate it.

MJ is Mark Joshi (http://www.markjoshi.com). Yes, read his books, and do the exercises and projects there. Oh and btw, he's also active in the forums here.QuoteOriginally posted by: msresearchThanks but who is MJ anyway?

QuoteOriginally posted by: msresearchQuoteOriginally posted by: DCFCYou have some of the basics for algorithmic trading. Personally, I'd see if you could get transferred to MS research in Cambridge who have good links to some large banks.I'd look at signal processing, and of course you will need to learn how to program in C++.Sadly, it is the case that I've had to say to some CS PhDs that they actually just need to learn to program, and then move on to C++.I'd steer clear of TA, but Market Microstructure ought to be on your reading list.Be aware that the quant ends of several large banks don't count CS as a relevant PhD, at you will need to dodge those who want to suck you into a code monkey role.But you know all this stuff anyway don't you ?Thank you DCFC. With all due respect, I would like to ask this question anyway, why do most of you guys (if not all) think that CS PhD cannot program in C++? In my perspective,Computer Science department is the only place where you could learn systematic programming skills. Most of us mastered C++ in Junior year and wrote at least 10,000 lines for course projects. When I was young, I used to laugh at kids who were not in CS and tried to mimic programming.So there must be a reason, say the requirement of quant programming is different from other fields? In terms of design patterns or other stuff? Anyway, I would like to listento your experiences and make it very clear before I move on.Thanks. I appreciate it.Ideally, I would want to tell you to ignore what DCFC thinks. However, that is the general perception in the field. Consequently, you will end up getting very few relevant interviews and your only hope is to have a high conversion rate. just concentrate on getting your C++ right, and upgrade your math skills. You can find a decent position with your CS PhD degree (I did)You probably will need about 6months of preparation to get on track.

QuoteOriginally posted by: msresearchThank you DCFC. With all due respect, I would like to ask this question anyway, why do most of you guys (if not all) think that CS PhD cannot program in C++?I don't think that it is the case that CS Ph.D's can't program in C++, merely that the likelihood of finding a CS Ph.D. who can program applications in C++ is roughly the same as finding a mathematics Ph.D. that can. The thing that CS Ph.D.'s focus on (like compiler design) is useful, but it's a separate set of skills that you need for application programming.QuoteMost of us mastered C++ in Junior yearGood for you. I've been programming C++ for close to 20 years, and I haven't come close to mastering it.QuoteWrote at least 10,000 lines for course projects.Most industrial projects involve hundreds of programmers involving code bases that have millions of lines of code. Also the problem with writing code for classes is that they get thrown away at the end of the class. Writing things that other people use that you can't throw away is a whole different level of complexity, as is trying to use code that should be thrown away but can't be.I suppose part of what happens is that if you are a mathematician or a physicist writing code, you are writing it to do something, and since your main priority is to get something to work, you care only about the parts that get something useful done, and the other parts you don't worry about. When a CS Ph.D. writes code, it often doesn't have to do anything, and certainly doesn't have to do anything outside the specific environment that it was created in.QuoteSo there must be a reason, say the requirement of quant programming is different from other fields?It's actually not that different from other fields. There is more algorithmic code, and there is much more time pressure, but there is no particularly reason to think that a CS Ph.D. would make a good applications programmer any more than a Literature major would make a good poet.

QuoteMost of us mastered C++ in Junior year and wrote at least 10,000 lines for course projects. When I was young, I used to laugh at kids who were not in CS and tried to mimic programming.All university students think they have mastered something. It's when you enter industry you realise how little you know, as we all here have experienced.University projects start off with no unknowns. The solution is kind of known beforehand. But not in real life.And the best programmers are not CS types, in general. The best tend to be musicians, language people, ... dare I say creative people, no idea, it's just what I see.In the 1970's/80's in TCD Dublin the CS department was lead by an engineer and there was very hands-on approach, certainly when it came to getting industrial contracts, ergo ==> good programmers.Have a look at this about Michael Jackson (not the singer), a real guru of System Development IMO. Yours truly was a certified JSP programmer, but I did not put that on my CV because it's a bit like saying one knows Fortran, Cobol and paper tape.

I appreciate all that replied and clarified my question.Though it seems weird to me, I'll accept the fact thatCS students need to learn C++ again. At least for now.

QuoteThough it seems weird to me, I'll accept the fact that CS students need to learn C++ again. At least for now. Why does it seem weird? What are your criteria? I think DCFC would agree; there is a mismatch between theory and real life. And the chasm is widening from what I can see BTW my own background is not CS.

QuoteOriginally posted by: CuchulainnQuoteThough it seems weird to me, I'll accept the fact that CS students need to learn C++ again. At least for now. Why does it seem weird? What are your criteria? I think DCFC would agree; there is a mismatch between theory and real life. And the chasm is widening from what I can see BTW my own background is not CS.

Thank you DCFC. With all due respect, I would like to ask this question anyway, why do most of you guys (if not all) think that CS PhD cannot program in C++?Because I've met them and read their PhD theses. Actually, my position was that the majority of CS PhDs cannot program and that they haven't even tried to do C++.That is not to say that a non-programming PhD is not valid, or even that a CS PhD should know C++ of course. But I worry about the general competence of someone who regards an O/S kernel as scary black magic.I don't think that it is the case that CS Ph.D's can't program in C++ merely that the likelihood of finding a CS Ph.D. who can program applications in C++ is roughly the same as finding a mathematics Ph.D. that can.That's just wrong.The thing that CS Ph.D.'s focus on (like compiler design) is useful, Also very rare these days. CD is not very fashionable, partly because a lot of what is easy to discover has been done, and also because it's hard which scares off many CS PhD candidates.but it's a separate set of skills that you need for application programming.That's just wrong 2fish, you should know better.Most of us mastered C++ in Junior yearI'm trying to work out whether that's sarcasm or stupidity ?You didn't "master" C++ in your first yearm anybe you did a little program. Did you master calculus completely from scratch in a year flat ?No you didn't. Wrote at least 10,000 lines for course projects.Good.Most industrial projects involve hundreds of programmers involving code bases that have millions of lines of code. Also the problem with writing code for classes is that they get thrown away at the end of the class. Writing things that other people use that you can't throw away is a whole different level of complexity, as is trying to use code that should be thrown away but can't be.True, which is why I think CS students should do some work on large codebases, plenty available in open source. So there must be a reason, say the requirement of quant programming is different from other fields?Different functions of time to get it to work and time to execute.

I want to disagree with most of replies here. There is not need to improve your C++. Work on math, brainteasers and finance.

I cannot believe that a PhD in computer science (that's eight years of programming, right?) needs to brush up on his or her C++ programming skills. I'd hire you in my bank!