This paper proposes the Generalized Information Ratio (GIR) to evaluate the performance of active portfolios under misspecified models. Motivated by the theoretical link between abnormal returns and residual covariance matrix, GIR is derived as alphas scaled by the inverse square root of the residual covariance matrix (GIR = {\Sigma}^(-1/2) {\alpha}). GIR translates residual covariance information into adjusted alphas and nests alphas and Information Ratio as special cases. Simulation results show that GIR is robust to model misspecification and produces stable out-of-sample returns. The gain of using GIR is substantial when the fund pool is large; however, a longer history of fund returns is required for accurate estimation of high-dimensional residual covariance matrix.
↧