MM, I think you make a very useful distinction, but let me restate it to you so that I'm sure I understand your view.
I think you are saying that
[1] no indicator or system of indicators (or whatever we wish to label that bunch of stuff on which we trade) can be perfect in the sense of producing a "win" or a correct prediction every time; but,
[2] it is possible to construct a system which produces results which are internally consistent for a long enough period so that one may safely extrapolate that said system will produce at the same level of consistency into the future.