@QuantWizard: Lol no, it's just a running gag in this forum now that this hftvol guy suddenly hijacks a thread and takes it completely off-topic, and talks about his passion for C#.
I've used R, Python and Octave (free matlab) for coursework in coursera, and also to perform analysis and build trading systems. I prefer python since it is the best all-rounder languaje between the three, along with numpy, pandas and other packages it offers very complete functionality for pretty much anything except the most time critical tasks. On R, i love that it makes it very easy to get a lot of analysis done on a few lines, and the integration with markdown to build reports straight from the data is priceless... i also like that it has some pretty mature packages for financial analysis (like quantmod). R studio rocks too... on Octave, i like that it makes matrix operations as easy as they can get... but it doesn't have as many libraries...
At my previous job which was with a medium sized fund, we used only matlab. Now at my current position also within a quant fund we use python and R. My thoughts: Matlab is fine, I personally find it to a bit clunky but it's easier to use and debug than R. If you're in a large well funded organization, acquiring licenses for libraries and 3rd party tools (e.g. MOSEK) should be fine. The data acquisition toolbox is great as it hooks up with a number of APIs like bloomberg, I find it to be more refined than R's packages. Bottom line it's expensive and unless you're willing to break some laws, it's out of reach for many people (there's always octave though). I'm a big fan of R, however some complaints I could see are lack of documentation, sometimes a few of the libraries are 'hacked' together, and the way it handles memory. The lack of documentation shouldn't be much of a deterrent to anyone with any sort of development background. I've definitely found issues with pieces of code, but since it's open source I'll go in and do manual adjustments which is fine. There's nothing you can really do to address the memory issue. Now the primary pro I find about R is that it offers a high degree of flexibility which is what you ultimately need for trading. For example if I want to use some kind of filter, there's usually a library and I can get off running fairly quickly. R Studio Server is a godsend, especially if you're working in groups or travel a lot. Python we primarily use as a replacement for shell scripts now and holds together a bit of our infrastructure. I think it's great for webscraping, basic scripting needs, and some data processing. I find it's relatively quick doing line by line iteration. On the data processing side, I think some good modules to work with are Pandas, Numpy, Scipy, Multiprocessing, Statmodels, and Pytables. The fact that pytables works so well with HDF5 is wonderful IMO. If you're not processing a huge amount of data, it's great. As far as scripting languages go, there's been a bit of hype about Julia and its speed. I think if library development picks up on it, it might be a viable option in the future.