An important part of Twoin18’s comment in #808 is that the math test used by his UK company had questions that were more complex than the SAT-2 (presumably Level 2), and not simply given in shorter time.
I would guess that most professors at Harvard or MIT would tell you that there is a wide range of student effectiveness in their classes. I won’t call it ability, because it is more like developed ability. But differences in developed ability cannot be entirely compensated by self-advocacy, resilience, grit, etc. within a semester. I doubt that all of them can be compensated entirely within a four-year period. I don’t believe that developed ability correlates especially strongly with any demographic characteristics.
For that matter, Richard Feynman observed significant differences in developed ability among Caltech students, who don’t tend to be selected holistically, and who are relatively few in number.