Saturday, May 9, 2009

Learning Matlab: Installation

After reading Ernie Chan's "Quantitative Trading" I became interested in MatLab as a way to develop new trading systems to diversify my trading approaches. Today I have a java application that reads in InteractiveBroker statements and does simple MonteCarlo tests. I would like to move all of these simple risk measurement functions to MatLab.

Priced at $1000 MatLab seemed to need more consideration. An open source alternative called Octave looked like it would do everything I want and cost nothing. The Octave developers have worked to make

Octave completely equilivent to MatLab. It seems, from reading forum posts, that Octave is almost compatible and that most script files developed in MatLab will run on Octave. But, not all Octave scripts will run on MatLab because Octave has many additional commands that do not exist in MatLab. Octave has also been reported to be much slower than MatLab, for me this is not important.

I will use Ubuntu as all new software development is free of anything related to Microsoft. This is mostly because of an intense dislike for Windows anything. The decision is made easier as Ubuntu runs my InteractiveBroker and custom trading apps with no problem.

Step 1: I downloaded "Octave", "gnuplot" and 'QtOctave' using the Ubuntu Package Manager and had them running within minutes without a single configuration change. Very nice.

Step 2: Study - I ordered 'MatLab Dymistified' from Amazon (because it was the cheapest MatLab book...). Then discovered the very complete Octave tutorial at: http://wiki.aims.ac.za/mediawiki/index.php/Octave:Getting_started

Save yourself some money - the free tutorial is better than the book....

Step 3: Script Compatibility - try some of the scripts from Chan's book and in the public domain....