Search results

Indeed, the post was mostly directed at specific quantitative trading positions (where a PhD is essentially a must), rather than general quant jobs. IB hiring has picked up quite a lot recently in London, although funds still seem to be "consolidating"!

If you are keen to become a quant analyst or quant dev I would choose the maths over the stats. The real analysis and linear algebra will allow you to do other courses such as stochastic calculus and numerical analysis. You can always pick up stats later, as it is much easier to learn via...

That question really needs to be answered in the context of the type of job you're applying for. Unfortunately, the derivatives pricing quant jobs have declined significantly in the last few years, and the word "quant" has become attached to any sort of task related to mathematical programming...

Well, it depends what you mean by "data structures". There is the simple array structure, which can handle matrices (tensors of rank 2) and above. If you are referring to banded, block or sparse matrices, for instance, then you also have these sub-libraries...

Numerical Analysis is also extremely useful, if only to help you understand how computers store numbers and thus the limitations on numerical algorithms. All good knowledge for a quant career!

Yes, it should be fine. But you have to bear in mind that those binaries are built against a specific version of NumPy, and new releases of NumPy are made often. The pre-built binaries should be absolutely fine for nearly everything you want to do though! Out of interest, what are you...

If you are after a specific version of NumPy (say, for a particular new feature) then you have to be careful about using the pre-built binaries such as python-numpy. Otherwise, it is probably the easiest way. You might also want to look into pip/virtualenv, as they will help you setup separate...

I thought this was from the Daily Mash for a second...

NLA underpins many areas of numerical methods. Matrix decomposition methods (such as LU, QR and Cholesky) are generally used to solve the matrix equation Ax=b, with A (matrix) and b (vector) known. Matrix iterative methods (such as Jacobi, Gauss-Seidel and SOR) are also used substantially. NLA...

So uBlas still seems like a good contender? I haven't actually utilised CUDA much beyond a few "toy" problems from some introductory books. Although, Alain, I'd be interested in why you thought it was a waste of time? Do you mean in terms of productivity or just that the speed-up wasn't worth it?

Mark, don't worry about the quantity of books in that list. Andy has compiled an extremely comprehensive list, but it is intimidating! The best thing is to start slow with some of the introductory texts and work up from there.

Firstly, I'll provide some advice for programming. Your best bet is to learn a language like Python first. Python is extremely versatile and will allow you to gain skills in basic programming concepts like branch/flow logic, variables, looping etc, while avoiding the trickier concepts of...

I made use of Python extensively in a former role. I was interacting a lot with the web to obtain data. There are some great libraries for that sort of thing, including Beautiful Soup and Python Requests (used instead of the horrible urllib2 library). It is very straightforward to interact...

Coming from a C++ background (i.e. not specifically C), I always prefer the former approach. As Polter says, it puts emphasis on the type.

I thought it would be interesting to start a thread discussing the issues around implementing a Matrix class in C++/CUDA/C#/Python. Although this topic has been talked about before, I'd like to see how people have approached the problem recently, now that we're heading into 2013. Some topics...

I'm struggling to see what the killer app benefits are to the consumer who may already have an iPhone and a MacBook Pro. It's not something you can carry in your pocket!

R is an interpreted language, so will naturally be slower than a compiled language like C/C++. I do not have any experience as to how it compares to MatLab though.

Dive Into Python is the best way for a programmer to get to grips with Python. While I'm at it, Dive into HTML5 is also extremely good! I think I mentioned Mark Lutz's book elsewhere (Learning Python, 4th Edition). Extremely comprehensive.

Although I probably sound like a broken record by now, if speed is not the primary concern, then other languages provide a much more rapid development cycle and libraries that "just work". C++ is sometimes a sledgehammer to crack a wallnut.

I'm looking to trade London right now. I'll be chatting to some Investment Banks then!

Sounds good, I'll let you know how I get on.

Does anyone know of any lists of NYC quant/finance events? Thanks in advance. QS. ---------- Post added at 12:46 PM ---------- Previous post was at 12:40 PM ---------- Silly me...I didn't see the sub-forum labelled "Events" :-)

I agree that FOSS software can sometimes have horrendous documentation, but there are projects which do possess incredible docs. These are predominently web projects though where the software and the communication processes between developers are tightly coupled. In finance, the culture is not...

I'd take a 3D screen over ETFs any day!

That would make sense. What is generally included in an Advanced Calculus course over in the states?

Changes from 2.6 to 3.0 This is because in Python 3.x, the print statement has been turned into a print function. This alone is enough to break a lot of legacy code! Bear in mind that Python 3.x will become a lot faster with the inclusion of Unladen Swallow, Google's Python speed-up project...

My biggest problem is probably that I'm UK based. Does anybody know of any good brokerages over here that accept automated trades, for low fees?

One issue with choosing proprietry software over open source software is that there are generally more people around to employ who are versed in proprietry tech. The prevalence of Visual Studio means that there are far more C# devs around than Python devs, even though the languages...

Roni, I guess that would work if the student had the Calculus basics grasped before SC or PDEs. I'm London based so am not sure what level a student in NYC would have at that stage. Over here, people rarely use the term "Calculus" and instead focus on "Analysis" at undergraduate mathematics...

Python is great for getting stuff built very quickly. It has a lot of "batteries included". C++ is great for optimising code to be extremely fast. However, Python is quickly catching up on the speed front! Try downloading IPython, it's an interactive Python shell - extremely useful for all...

Well, I'd be interested in the June course (I'm London based). Could you PM me your contact details/website? Thanks, QS.

Ahhh yes, I've had of Ernie Chen, he has written a book on the topic. Is it worth picking up? Thanks for the thread list, Andy. Joy - as it happens I don't own MatLab, but I do have Python. I don't suppose there are any UK brokerages that have a REST style API? If so, I could work through that...

Your best bet is to read through the (brief) tutorial, this will give you an idea of what it is capable of. An extremely powerful bit of kit!

Hi everyone, I'm keen on getting into algorithmic trading at home (primarily paper trading right now). I'm a capable software dev, so that's not an issue, but I would like a heads up on what protocols are out there, what reference texts are worth looking at, which blogs to read etc. Where does...

Yves, I'm surprised you're not using Django (Django | The Web framework for perfectionists with deadlines) for your web front end, instead of PHP. It would allow you far greater ability to tie it into your underlying valuation suite.

Are there any other sources of good information on algo trading besides the protocol implementation?

Given that I do a lot of web development, the Firebug debug/development tool is crucial. There is nothing like that on Chrome just yet (that I am aware of). This is the beauty of the Firefox plugin system.

I would imagine that GPU computing makes PS3 supercomputers redundant. Fermi is designed for this sort of work, whereas a PS3 is not.

Visual Studio also has GUI tools "out of the box" with Windows Forms. If you're looking for graphing tools as well (which you probably will be if you're doing financial applications), then try this: Main Page - ZedGraphWiki

I find it difficult to see how one could attempt courses on Stochastic Processes or PDEs without a basic grasp of Real Analysis. For example, Ito's Lemma is analogous to the Chain Rule for functions and would require the knowledge of Taylor Series. These concepts would only be taught in a Real...

How do they work? I presume it is a case of two frames with separate polarisation?

I think what you mean is programming an application that sits on top of a database layer. For that, Python would be a good start as it has a fantastic MySQL library (MySQL-Python). However, C++.NET with SQL Server might another good way to go. I believe there are free versions of Visual Studio...

Don't forget Hacker News, it's great for seeing bleeding edge tech (although, mostly a Web 2.0 -esque crowd).

It seems I forgot to introduce myself - I'm Mike or "QuantStart" and have just recently finished a PhD in Numerical Methods/Fluid Dynamics at Imperial College London. Looking forward to some (more) interesting conversations!

So true :-) I had fun reading the Web Design Programmer snippet. I especially appreciate how it is "borrowed" from elsewhere!

Learning Python by Mark Lutz, 4th Edition is a fantastic book and will help you get familiar with the language. His other book, "Programming Python" is more application specific. I also recommend starting C++ first though, as it will teach you about memory management (VERY important for...

I'm still worried about the future of MySQL. In time, for web projects (which I work on quite a lot), it may be prudent to switch to PostgreSQL.

Windows is the dominant OS in most banks, so you will need to get to grips with Visual Studio C#/C++. You can download free versions from Microsoft's website to start learning. They come with nearly everything that you'll need as a lone developer. The Professional Editions cater more for team...

I was aware of the volatility models (Jump-Diffusion and, obviously, GBM), but the mean-reversion sections are new to me. Very interesting! Thanks Sanket.

Since I am quite new to GPU architecture, and you mention the overhead, can you elaborate on how much overhead there is with tranfer of data (i.e. I/O bottlenecking) to/from the card? For instance, in the PDE situation, can all the initial data be transferred at the start of the calculation...

Did anybody post any reviews of the course? I'd love a first hand account.

I imagine the academic community will be leading the way, particularly with CFD, as suggested by joyryde. I know that a lot of my previous codes would have benefitted immensely from a GPU setup, as they were easily parallelised. I'll be saving up for my Fermi. :-)

So with job quantity in decline and a saturation of new "talent" on the hunt, it would seem that now is a suboptimal time to hunt for a quant job? Unless one is exceptionally good of course...!

Looks like some interesting topics in the courses. I'd be very interested in having a play around with more of the modules in SciPy, I tend to stick with NumPy only at the moment.

Thank you both for the recommendations. I'll be checking out your FDM book as well as Thomas'. The explicit, stable schemes you mention - they may be interesting candidates for GPU speedup if they are easily parallelised?

A good thread. Thanks. I wonder how things are going to improve now that Fermi is on the scene.

Indeed...I am also a big fan of open source. When I previously worked on a large web application with Python/Django (the web framework) there were times when we needed to patch the Django code for our own use. This would not have been possible using a technology such as ASP.NET MVC.

Have you had a look at this page on Wikipedia? It talks about different volatility models utilised in the Black-Scholes model: Stochastic volatility - Wikipedia, the free encyclopedia

I'm always wondering about the relative supply/demand for quant careers. Are all these courses turning out "too many" quants to fill the roles?

Has anyone used CUDA/OpenCL in their applications? I imagine for Monte Carlo type scenarios it provides an incredible speedup ("embarassingly parallel") but what about PDE methods?

What are you favourite FDM books? I'm interested in more advanced texts primarily. I enjoyed Duffy's "Financial Instrument Pricing using C++" but haven't had a look at his FDM book yet.

Has anyone had any trouble upgrading from 8.10 to 9.10 via apt-get dist-upgrade? I'm hesitant as I have heard horror stories before (although this was Feisty).

As a CS major, I would assume that you have learnt a lot of discrete mathematics. The area that you will probably be lacking in (assuming you haven't done your own reading or you haven't taken courses on the subject) is continuous mathematics/calculus/analysis as well as probability theory. In...

What I'm really asking is whether anyone is considering replacing some of their large C++/Java code bases with Python, in order to obtain the advertised benefits. These benefits would include ease of maintenance, rapid development time, less code to debug, no compilation steps etc. I can see...

May we ask what your university background is (i.e. what subject you are majoring in)? It might help us to be able to point you in the right direction.

A natural hedge is to start a passive income business on the topic of jobseeking.

Why is Python being neglected from the discussion? It has some incredible libraries, is immensely fast to develop in and code maintenance is straightforward due to excellent readability. One may question the performance of Python code against C++, but NumPy/SciPy optimisations (such as those...

This is a problem I am having too. I have Vols 1 and 2 of Shreve and am wondering whether I should just concentrate on Crack and the questions in there. Where do people see their time being optimally spent in terms of a first career in Quant?

How many people are using optimised Python, i.e. NumPy and SciPy, in production financial applications? Is it "fast enough"?