What Makes a Good Statistical Model?

Dec 7, 2012

In this paper, we investigate whether the new Barra Europe Stochastic Factor Model (EURS1) performs differently in portfolio construction when compared to other statistical models commonly used by investors.  In order to explore this, we built a typical principal component analysis (PCA) model and used both the PCA and EURS1 models to track two popular benchmarks using the Barra Open Optimizer.  In all cases, we found EURS1 to have the lowest tracking error, and the EURS1 slow model to have the lowest turnover.