Well, it's hardly "colloquial" usage, more like conversational, dictionary, typical, etc. The comment didn't seem like it was grounded in maths/stats terminology.

I read the comment as being about an individual rather than a population, which illustrates the problem: predicting individual success is (nearly?) impossible even if population data is consulted. Probably due to chaotic processes and the impossibility of understanding all relevant variables, exacerbated by the confusion of correlation with causation i.e. post hoc ergo propter hoc fallacies.