It may be worth clarifying the sort of "speed" we're talking about here. An IMO takes place over two days. On each day the candidates get a question paper with

It may be worth clarifying the sort of "speed" we're talking about here. An IMO takes place over two days. On each day the candidates get a question paper with three questions on it. They have 4.5 hours to attack the questions. The threshold for getting a gold medal might be roughly equivalent to getting five of the six problems more or less completely solved. The questions are always of the form "prove X" -- maybe occasionally "prove or disprove X, whichever is possible" or "find quantity X, with proof"; but it's always a matter of finding proofs -- where X is something no more than a few lines long. So it's "speed math" in comparison with actual academic mathematical research, which typically takes place on a timescale from weeks to years and where there's no guarantee that any human-accessible solution exists. But it's a whole lot slower than pretty much anything else these (roughly) 16-19-year-olds are being asked to do in mathematics. And, empirically, it does turn out that people who do well in IMOs tend to do pretty well in mathematical research, and in other mathematical things they put their minds to. (But yes, of course there are kinds of genius that have nothing to do with mathematics, and kinds of impressiveness that have nothing to do with genius; and there are plenty of first-rate researchers who never took to IMO-like problem-solving.) (Random example of an IMO question, from this year's IMO, to give an idea of what a typical X looks like: "For each positive integer n, the Bank of Cape Town issues coins of denomination 1/n. Given a finite collection of such coins (of not necessarily different denominations) with total value at most 99+1/2, prove that it is possible to split this collection into 100 or fewer groups, such that each group has total value at most 1.")