MEVA

Introduction

Meva contains two mean-variance portfolio optimizers: an analytical optimizer and a numerical optimizer. Both are single-period optimizers.

Analytical optimizer

The analytical optimizer aopt() handles linear equality constraints and soft linear equality constraints (a penalty proportional to the squared deviation from equality is subtracted from the objective function).

Numerical optimizer

The numerical optimizer nopt() is a long-short optimizer that allows you to separately specify the sum of the negative portfolio weights and the sum of the positive weights. It handles soft linear equality constraints, inequality constraints (implemented as iterative soft constraints), turnover constraints (iteratively multiplying linear transaction costs), and (linear and quadratic) transaction costs.

Covariance estimation

Both portfolio optimizers need an estimate of the covariance matrix of asset returns. Meva contains two algorithms to estimate the covariance matrix: cov_pca() which is base on principal component analysis and cov_fa() which is based on factor analysis.

Install

Requirements:

meva	python, numpy
speed	accelerated BLAS such as ATLAS
unit tests	nose

Meva is a pure Python package. To install, all you have to do is to make sure Python can find the meva directory. Or you can install the traditional way:

$ python setup.py build
$ sudo python setup.py install

After you have installed meva, run the unit test suite:

>>> import meva
>>> meva.test()
<snip>
Ran 33 tests in 7.038s
OK (KNOWNFAIL=1)
<nose.result.TextTestResult run=33 errors=0 failures=0>