Dynamic portfolio optimization with ambiguity aversion

•We investigate portfolio selection with transaction costs and parameter ambiguity.•By distinguishing between ambiguity aversion to returns and to return predictors, we derive the optimal dynamic trading rule using the robust optimization method.•The robust portfolio gives less weight to highly volatile return-predicting factors.•The robust portfolio loads less on the securities with large and costly existing positions.•The robust strategy outperforms the corresponding non-robust strategy in out-of-sample tests.