The article introduces new tests for predicting returns in panel data that account for realistic conditions like interdependence between units. These tests can handle various scenarios like different predictive abilities, lasting predictors, and complex error patterns.