Hybrid AMX₃ perovskites (A=Cs, CH₃NH₃; M=Sn, Pb; X=halide) have in the last years revolutionized the scenario of photovoltaic technologies. Despite the extremely fast progress, the materials electronic properties which are key to the performance are relatively little understood. We developed an effective GW method incorporating spin-orbit coupling [1] which allows us to accurately model the electronic, optical and transport properties of halide perovskites, opening the way to new materials design. In parallel, a series of computational simulation carried out using Car-Parrinello molecular dynamics have been performed investigating the nature of the perovskites/TiO₂ interface, the role of moisture in the perovskite degradation process and the effect of the defect on the device working mechanism. Finally, a series of different strategies will be reported to increase the device stability and efficiency.[2] While instability in aqueous environment has long impeded employment of metal halide perovskites for heterogeneous photocatalysis, recent reports have shown that some particular tin halide perovskites (THPs) can be water-stable and active in photocatalytic hydrogen production. To unravel the mechanistic details underlying the photocatalytic activity of THPs, we compare the reactivity of the water-stable and active DMASnBr₃ (DMA = dimethylammonium) perovskite against prototypical MASnI₃ and MASnBr₃ compounds (MA = methylammonium), employing advanced electronic–structure calculations. We find that the binding energy of electron polarons at the surface of THPs, driven by the conduction band energetics, is cardinal for photocatalytic hydrogen reduction.[3] In this framework, the interplay between the A-site cation and halogen is found to play a key role in defining the photoreactivity of the material by tuning the perovskite electronic energy levels. Our study, by elucidating the key steps of the reaction, may assist in development of more stable and efficient materials for photocatalytic hydrogen reduction. The overall picture of our theoretical investigations underlines a crucial role of computational investigation, casting the possibility of performing predictive modeling simulations, in which the properties of a given system are simulated even before the materials laboratory synthesis and characterization. At the same time, computer simulations are shown to offer the required atomistic insight into hitherto inaccessible experimental observables.