Computational fluid dynamics (CFD) simulations of dense gas-particle flows are extremely challenging due to the formation of clusters. These structures reduce the effective drag on the suspended particles and hence are key for the prediction of the flow in and the performance of fluidized beds. Our group has established models for the effective drag and particle-phase stress via Euler-Euler (EE) simulations . However, EE simulations become cumbersome when treating polydisperse particle systems. Here, we focus on an Euler- Lagrange (EL) approach, with the goal to establish a sophisticated filtered drag model for coarse-grid (i.e., incompletely resolved) EL-based simulations. First, we detail the effect of the grid resolution and the particle-scale (i.e., "microscopic") drag law on our results. Based on our computational data from fully resolved EL-based simulations, we construct a filtered drag model for a later use in coarser EL-based simulations. Our results highlight the significant effect of particle clustering on the average slip velocity between particles and fluid, and indicate how this clustering can be accounted for in incompletely resolved EL-based simulations.