I'd bet that a significant majority of prediction market revenue comes from people who bet big.

So I expect your solution would fix all of it, as a second order effect, in that running one would stop being a viable business model.