The case of Python vs R vs Matlab vs an incredibly fast HFT platform vs the FPGA option

(Last Updated On: April 5, 2013)

Once again, this is the continue dialogue with highly intelligent Youtube Channel (youtube.com/quantlabs). This one is about the performance of Python and what I plan to do!

For those that want to learn more, I would join my FREE newsletter to be part of the neat dialogue

Here it is:


Also, I saw your video on r vs matlab. It doesn’t really apply to python, because python is far more flexible. You can develop python from spyder or eclipse (IDEs capable of debugging).


So, if python people ask you about why matlab or open-source, it is still not going to be a convincing argument unless you focus it on simulink and integrated fpga development. Python really doesn’t have anything comparable to a system like that. That should be the main reason, and it is really a good enough answer


Jython for integrating python with java. Jython can use eclipse.


IronPython for integrating python with .net. Ironpython does in fact have Visual Studio integration.


You can generate c/c++ code using Cython. The c/c++ code can call the optimized cython libraries (much faster than the standard python libraries).


Plus you can use SciPy, NumPy, Matplotlib. You can call c/c++ code using cython or swig.


Examples for interfacing the ta-lib libaries with cython:


Interfacing ta-lib with Python using Cython : moving average function example




Python also works with quantlib and quickfix. For speed, the basic python is on par with matlab.


You can run some basic code for FPGA:


Programming FPGAs with Python




Although, it is nowhere near the capabilities of Simulink.


Python is used extensively by JP Morgan and man hedge funds, etc. Scientific Agencies like CERN and NASA also use Python.


You can generate guis with pyqt.


Spyder is similar to Matlab’s IDE in many respects:




Here is a nice portable distribution of python with pyqt, numpy, matplotlib, scipy, and an integrated IDE (spyder), etc.





You can try-out the bundle I linked to you in your spare time if you want to, you don’t need to, it’s just to inform you what python is capable of doing.


You also have this:




Capable of GARCH, ARIMA, VAR, etc.


Many different free tools like this.


The point is that it is possible to replicate a lot of the functionality you have in matlab and in the case of integrating your algorithms into c#, java c/c++ applications, or for using it on the web or building ground-up desktop applications with pyqt, python has matlab beat.


The real advantages of matlab is that it was designed to be an easy-to-use pseudo-language for non-programmers (mainly engineers) and of course simulink. So, it is easier to use matlab for non-programmers than python. Simulink provides a separate field that python does not specialize in and cannot compete.


You have been looking into various different platforms over the past few years, but it looks like now, that this is as good as it gets. It may be possible to use a cracked version of deltix for personal use, but in an enterprise, it is basically too risky. Tradelink is sort of redundant when you already have matlab. I think a lot of the research has been done, it is now more about implementation of the whole end-to-end system from strategy design to execution from within the program (simulink/matlab). Provides execution from within matlab, c++ code-generation and FPGA code generation and deployment all under the one banner.


What do you think?


–> I know about this Python debate. The CEO of the proposed  HFT system  feels there is more potential with it vs R or Matlab. From my perspective, Matlab is still the top pick just due to its extensive options available including the ability to extend M scripts to so many environments. The performance is all the same but as I said in my video, I am not going to risk millions of dollars on unproven technologies. I don’t have the resources of a JP Morgan for army of coders.

With what I Just saw in this HFT platform, your jaw will be left in the dinosaur’s age compared to what I saw. Honestly nothing compares to it. Anyone who wants to argue it down would be purely stupid and a moron of another dimension. Between Matlab/SImulunk and this platform, you are looking at the ultimate trading machine that will do circles around anything. It makes all those other technologies look so dated and obsolete. In terms of FPGA, the Simulink is a nice option but I am really startng to think this HFT platform still beats it hands down. I am of course referring to its relative ease of use and rapid development capabilities.


NOTE I now post my TRADING ALERTS into my personal FACEBOOK ACCOUNT and TWITTER. Don't worry as I don't post stupid cat videos or what I eat!

Subscribe For Latest Updates

Sign up to best of business news, informed analysis and opinions on what matters to you.
Invalid email address
We promise not to spam you. You can unsubscribe at any time.


Check NEW site on stock forex and ETF analysis and automation

Scroll to Top