Optimal Factor Composition Under Sampling Error of Parameters - Diversification to Factor Portfolios

Existence of premiums for factors is known to quantitative equity fund managers. They use factors to get premium of them when they construct (factor tilted) portfolio from a lot of stocks.They try to use many factors in order to get the premiums effectively by calculating covariance of each factor. Increasing number of factors is expected to enhance efficiency of portfolio by theory. But in practice, when they construct portfolio from large number of factors, sampling error in estimated parameters sometimes makes trouble to factor tilted portfolio. In this study, we will show case studies of factor tilted portfolio from a lot of factors in three classes, and will propose robust method to deal with sampling error