Hi all, The typical "help me" posts here seem to be from the physics/math angle asking how much C++ to know, but rarely from the opposite side.. can you experienced people here give me some advice on how I go about with this switch from PhD in CS to quant finance?I am at a top 10 US university in Comp. Sc., finishing in some months.. I have 5-6 publications and a patent.. but don't find academics appealing any more and looking for a change in career.. My background: For my research, I mostly work with logic, esp temporal logics.. this is pen and paper work, but I have good programming skills, some experience in distributed/parallel algorithm development, decent C++, some STL, (but I don't want to end up being a "mere developer".. Prefer front-office rather than back-office..) I pursued Math semi-seriously before settling for CS, did a fair bit of abstract algebra, complex analysis, etc. but that is many years in the past.. My engineering maths is some years old as well.. though i am quite confident i can brush it up..To start with, I am reading Crack's "heard on the street" as well as his black-scholes book.. I am also working through the Rennie/Baxter book on financial calculus.. I haven't done any courses in finance, do I need to?What else do I go about preparing and how much? statistics? linear algebra? numerical analysis? PDEs? Complex analysis? where do I draw the line .. I guess I need to pick up a fair bit of non-linear optimization and some stochastic calculus.. how much? do i focus on breadth or depth? any recommendations for reading with interviews in mind?Also, .. this is probably putting the cart before the horse, still.. how do I go about applying for these jobs, do I contact the head-hunters posting in the career forum, don't feel quite comfortable sending my CV/resume to complete strangers.. esp after reading some horror stories in these threads.Much thanks for suggestions,--s. Edit: some typos.

Bump. Any advice?

You need to get your calculus right up to speed, and you're right about stoch calc.C++ is where the action is, add Boost to STL to get it down pat.don't feel quite comfortable sending my CV/resume to complete strangers.. esp after reading some horror stories in these threads.We're not strangers, think of us as friends you haven't met yet We don't want your CV anyway, papers you've written would be far more interesting, and yes one of the C++ screening questions may includewhat 42;Actually does.Are you in the UK ?

to brush up on topics like non-linear optimization or stochastic calculus...you might conisder taking some distance learning classes from Columbia University's CVN...good program...some classes to check out are "Stochastic Processes for Financial Engineering" or "Optimization Models and Methods for Financial Engineers". You can still sign up for Fall 2005 semester classes as a "CVN Special"...i.e. you just need to sign up and pay, you are not required to be part of any degreee program... ...you could also consider Paul Wilmott's CQF program...which is compact and will introduce you to all the maths/concepts you need in terms of derivatives pricing...I've done both the CQF as well as CVN classes...the latter more for CS than FE...

QuoteOriginally posted by: DCFC C++ is where the action is, add Boost to STL to get it down pat.DCFC, any particular reason why you suggested the Boost libraries in particular? I've worked with ACE and Dinkum on UNIX and Windows before, and Boost looks like the same kind of thing. Also, is it really true that one has to know STL down pat to get work as a quant?

> one has to know STL down pat to get work as a quant? What do you for lists, maps, iterators?Do you roll your own, hope not.

> add Boost to STL to get it down pat.Wha da?

come on Angelique, Boost and ACE are as similar as Python and a Snake (a real one)! You seem to have worked with ACE, that's a major issue, one that you should brag about, since unfortunately there aren't that many real-time live trading apps based thereupon - you can always just custom built your stuff straight from the Frameworks can't you? However, Boost is a completely different ball-game, it works at a much lower level, bridging the gap between the standardised features of C++ and those which unfortunately can't still be adopted, take the existing 6 smart-pointers implementation as an example. In order to be able to compare those libs as lightly as that, one should really know them by heart. Anyway, to become a full-fledged quant-developer, and if you're willing to establish a somewhat subjective hierarchy, I would humbly suggest the following incremental approach: STL, Boost (smart-pointers), Loki (patterns), back to Boost (internals) and then ACE (frameworks). Cheers, Serp.

Thanks all for your replies. Let me get one thing straight: are you suggesting I sharpen C++/STL/Boost for any Quant position or for Quant developer position? I am more interested in the non-developer positions, because I can get developer positions in my area of specialization in startups etc., and I am possibly switching as way of reducing the coding and increasing the Math-ing.DCFC: Having read a couple hundred of your posts, I think you don't fall into the category of HHs that do various evil stuff mentioned in other threads. As for location, I am in the States, but willing to relocate. I'll PM you a link to my research etc.DC: Thanks for the suggestions on possible long distance courses. At the very least, I have a syllabus for self-preparation.

what does 42; do actually? what's the assembly code look like?

QuoteOriginally posted by: Cuchulainn> What do you for lists, maps, iterators? Do you roll your own, hope not.OK, so the collection classes are what I consider the basics. If you know just those in and out, I don't think you can claim you have STL down pat. STL is a huge beast.SierpinskyJanitor -- Thanks you for the tips. Clearly I don't know much about Boost, hence my questions. I'll follow your advice and look into Boost smart pointers and internals. (Although ACE also has smart pointer implementations).

> OK, so the collection classes are what I consider the basics. If you know just those in and out, I don't think > you can claim you have STL down pat. STL is a huge beast.Was a starter.STL is part of standard C++ and you don't know C++ until you know STL in my opinion. Leave ACE, Boost (and Loki!) on the sidelines until the fundamentals are right. Then by all means, start with these, if there is time. It's the 80/20 rule.

if you want to be a quant, i'd advise concentrating on the maths side, you need decent C++ but it doesn't have to be great. Read the standard basic texts and really learn them so well you would get an A if you were sitting an exam on them. (i won't do the usual plug for my book )

ArthurDent if you PM me, I would be happy to pass on the contact for a job in NYC, requiringexpert C++. I couldn't pass the screening exam, apart from the maths and some C++. There wereno financial maths questions, just debug some C, class, inheritance etc. I tripped up on templates... It sounds like you on the other hand would do well