Regression, multicollinearity and Markowitz

This paper shows that the usual drawbacks of the Markowitz model (high optimal weights, high volatility and low out-of-sample performance) can be overcome by correcting for the multicollinearity of individual assets that directly affect the estimation of portfolio weights. That improves the stability, predictability and out-of-sample performance of the Markowitz model, allowing it to provide better results than the 1/n rule.