Probabilistic approaches have changed the maxim of categoricity often assumed in phonological analyses. With more statistical power, we are now able to perform more comprehensive and detailed analyses, which in turn help us adjust our own theoretical assumptions. Weight and stress have benefited greatly from such approaches, and we now know that the computation of weight is not as categorical as we thought. Instead, more intricate, subtle patterns can emerge once we examine enough data. In this paper, I show that this is in fact the case for Portuguese.