"Hidden Variable" is a loaded term, so I can't say much about it without a context. What I am referring to is a bit different than the "hidden variable" concept normally applied to QM. In the case I am talking about there is no "hidden" function per se, rather we are viewing the function directly but are modeling it conceptually with stochastic methods rather than discrete deterministic ones. Actually, this is what usually happens if you model any closed finite state process by sampling its state, so this is not surprising. Unfortunately, our measuring/sampling tools are pretty coarse which affects the resolution of the analysis, though this has been getting better with time. With much higher resolution data we might be able to infer or properly guess the deterministic function, but until then we are stuck with stochastic functions and sampling.
I would add that the fact it can be modeled this way at all suggests that the function is very, very simple. These type of analysis doesn't scale well at all with function complexity.