Mean Variance Optimization

This is an implementation of [Mean Variance Optimization] (https://en.wikipedia.org/wiki/Modern_portfolio_theory) and was built with much inspiration from the one found at [quantandfinancial] (http://www.quantandfinancial.com/2013/07/mean-variance-portfolio-optimization.html).

This implementation consists of a simpler interface that exposes two top-level methods optimize_sharpe_ratio which optimizes the portfolio in regards to the Sharpe Ratio, and optimize_return_amount which optimizes the portfolio with the minimum variance for a requested return on the portfolio.

Sample Usage

stocks = ['BA', 'WFC', 'CSCO', 'GOOG', 'FL']
start = 2013
rf_return = .015

optimized = optimize_sharpe_ratio(stocks, start, rf_return)
# => {'CSCO': 50.0, 'BA': 0.0, 'GOOG': 0.0, 'WFC': 50.0, 'FL': 0.0} 

Which means you should invest 50% into CSCO and 50% into WFC. If you don't want to invest that much into just two stocks change the max_weight parameter to a lower value, e.g.

optimized = optimize_sharpe_ratio(stocks, start, rf_return, max_weight=0.25)
# => {'CSCO': 25.0, 'BA': 25.0, 'GOOG': 0.0, 'WFC': 25.0, 'FL': 25.0}

Dependencies

• scipy 0.16.0
• numpy 1.11.1
• pandas 0.18.1
• requests 2.9.1