Return on capital isn't a super meaningful metric for capacity constrained trades. If a fund earns 80% returns, but has no means to compound the resulting profits through the same mechanism, whoever receives them naturally puts them into something with worse returns, so their wealth still grows slowly over time.
I think Medallion is somewhere between HFT and stat arb, probably a mix of multiple strategies along those time frames. The faster you trade independent opportunities, the more you recycle capital, and prime brokers extend tons of leverage. When I worked in HFT, our profits were bound by other factors way before cash, and my desk's ROC was far higher than this when doing well (even when doing poorly, ROC was quite high. It was the expenses of finding those returns that killed us).
If all the money is employees', Medallion is basically just a prop firm. The employees paid out of the fund are essentially partners/owners, and the rest earn discretionary payouts of management/performance fees.
Even within the fund structure, most profit is return on labor, not capital. I'm sure if you compare margins paid to partners in professional services firms like law or consulting vs. typical publicly traded companies, they're also far higher, but Cravath, McKinsey, etc. won't let you buy in as a passive investor; you have to work for it.
ETA: If you're wondering how it's possible to earn 80% returns, or even more: there are myriad tiny inefficiencies you can trade on given the right research and infrastructure. I'm sure 80% is simply the point at which Medallion makes the optimal $ per year relative to risk. They could probably throttle back, make less $ on smaller capital, but far higher percentage return, if they wanted.