Cover crops are increasingly used for weed management. But selecting the most suitable species of cover crop to be associated with a main crop requires long-term trials. We present a model-based method that uses a reduced number of parameters to help select cover crops in the context of banana cover-cropping systems. We developed the SIMBA-CC model to focus on radiation interception. The model was calibrated for 11 cover crop species by measuring their growth in 4m2 plots with three levels of shade (0, 50, and 75%). The SIMBA-CC model served to predict the long term growth potential of the 11 cover crop species in function of the radiation under the banana crop canopy. The model was validated using three species in association with banana plants. We defined three indicators based on outputs of the model to assess the ability of each of the 11 species (i) to compete with weeds and (ii) to be maintained in the long-term under the canopy of the main crop, and (iii) to evaluate competition with the main crop for nitrogen resource. This ex ante evaluation revealed the most promising species to be intercropped with banana. Finally, the SIMBA-CC model was used to define the light interception traits of a virtual cover crop that satisfy the three indicators in the case of intercropping with banana. We showed that to satisfy the three criteria, cover crops with low values of optimal photosynthetically active radiation (PARopti) should have moderate maximal biomass productivity, while crops with higher PARopti values should have a higher maximal productivity. The use of functional traits and modeling appears effective to disentangle the relations between intrinsic traits of cover crops and effect traits that affect the performances of the intercropping system.