With their dipole-forbidden 4f transitions, lanthanides doped in nanoparticles promise high excited state lifetimes and quantum yields that are required for applications such as composite lasers or nanoscale quantum memories. Quenching at the nanoparticle surface, however, severely reduces the lifetime and quantum yield and requires resource-consuming experimental optimization that could not be replaced by simulations due to the limitations of existing approaches until now. Here, a versatile approach is presented that fully accounts for spatiotemporal dynamics and reliably predicts the lifetimes and quantum yields of lanthanide nanoparticles. LiYF4:Pr3+nanoparticles are synthesized as a model system, and the lifetimes of a concentration series (≈10 nm, 0.7−1.47 at%) are used to match the model parameters to the experimental conditions. Employing these parameters, the lifetimes and quantum yields of a size series (≈5 at%, 12−21 nm) are predicted with a maximum uncertainty of 12.6%. To demonstrate the potential of the model, a neutral shell is added around the core particles in the model which extends the lifetime by up to 44%. Furthermore, spatiotemporal analysis of single nanoparticles points toward a new type of energy trapping in lanthanide nanoparticles. Consequently, the numerical optimization brings applications such as efficient nanoparticle lasers or quantum memories within reach.