Your rule of models is flawed. Newton's mechanical model of the universe is good enough for all practical purposes except where it isn't. Then you have to go relativistic or quantum. Probabilities only apply to future events. Once events have been observed, the probability changes.