This only means that the model is an oversimplified, abstract bullshit, which is, basically, what all the statistics-based models of complex stochastic, partially observable systems are.
The fundamental principle is that one single major factor missed or overlooked makes a model fundamentally flawed, but still useful to gain attention and funding. One single flaw in logic makes a philosophical system worthless, but it does not stop publishers and teachers.
Estimated probabilities could never be explanatory by definition. Only for simplest, fully-observable, deterministic systems like a dice, could be adequately modelled with probabilistic models, and even those are nothing more than a disconnected from reality simulation, which in principle could tell nothing about what exactly the next outcome would be.
Probabilistic modeling is nothing but a new metaphysics. A way of secure a decent living as an astrologer in a court of some government.