Quote Originally Posted by unutbu View Post
I've written a suite of perl programs whose logic could be streamlined and improved. The perl programs are my attempt at testing the Efficient Market Hypothesis. I can no longer wrap my mind around the logic of the system, even though I wrote it.

So while I am having fun learning python, I have in mind a complete rewrite of the perl programs.
R and numpy both look like they could be useful, though the math I'm using is not very high-powered.

Two pieces of perl that I'll need to replace with python modules are Math::Brent (http://search.cpan.org/dist/Math-Brent/Brent.pm)
for nonlinear function minimization (for the calculation of rates of return, solving of high-order polynomial equations) and a statistical linear fit package
(http://search.cpan.org/~randerson/St...ics/LineFit.pm)

A python package (or R?) than can do multivariable statistical regression would be of great interest to me too.
well, in that case, definitely go with R, because R is built for statistics - it's simply the right tool for the job you have in mind. specifically you might look at function lm() for linear fits and multivariate statistical analysis, and function optimize() for brent's optimization algorithm (both part of the core R install).