Parameter uncertainty in portfolio selection: Shrinking the inverse covariance matrix

► We directly apply shrinkage to the inverse covariance matrix for portfolio choice. ► We optimise our estimators under portfolio objectives using non-parametric methods. ► The proposed shrinkage estimators lead to a set of new portfolio strategies. ► The new strategies offer low levels of variances and high risk-adjusted returns. ► Our framework can easily accommodate a small sample and short-sale constraints.