> I had someone arguing against me that it's not a good indicator since only approx 20% of competitors get a PhD in Math

As lordnacho says, that argument doesn't hold up at all. 20% seems really high, to me. A PhD in math isn't like scoring highly on an IQ test, it's not purely the result of being mathematically capable, it's also the result of undertaking a PhD in math. Plenty of smart people apply their abilities elsewhere. Perhaps in finance, for instance.

Also, I'd guess that the Olympiad winners who started a math PhD were more likely to succeed than the average math PhD student, and more likely than average to eventually get tenure.

> not being food at the competitions

I rather like this typo ;-P