This thesis focuses on the risk measure in the Markowitz algorithm. We discuss why assuming normality is unrealistic, and why the unconditional sample covariance matrix is an inappropriate input for the algorithm. We compare ...