REU 2004 – Alternate Methods for Covariance Estimation in Portfolio Analysis

With the support of New Frontier Advisors, our group at WPI is working on a problem within the finance world; to optimize returns while minimizing risk, as measured by the asset covariance matrix. The ultimate goal would be to find the true covariance matrix, from which the efficient frontier can be derived. However, since the true covariance matrix is impossible to completely determine, several models have been proposed to estimate it. Unfortunately, no single model estimates the covariance matrix in a superior way. Thus, we will combine models in order to minimize the error involved within the final model. From this new covariance matrix, we can come up with an efficient frontier which represents the portfolios which will yield optimal points of return versusrisk.