True enough, we're all dead, but the mathematics can lead to a bit of depression when it comes to investing once you think about it a bit. Given all portfolio returns are based on the Prime Rate and modeled after random walks, it can be pretty easily seen that the rates of return based on modernish portfolio theory[1] will approach the Risk-Free (Prime) rate.

I wish i could find the actual theorem, but it's been a few years since I took the course and have long lost the textbook[1]. It was pretty clear though when it was taught, or at least that's what I gleaned out of that lesson without the teacher having to say it.

[1] https://en.wikipedia.org/wiki/Capital_asset_pricing_model

From your link:

E(R_m)-R_f is sometimes known as the market premium (the difference between the expected market rate of return and the risk-free rate of return).

Is the CAPM model an accurate description of reality?

Not really: http://www.bfjlaward.com/pdf/25949/61-76_Blitz_JPM_0417.pdf

There's a lot of criticism levelled at the CAPM, including significant methodological challenges (measuring the market portfolio, etc.), but the basic intuition underlying it is sound.

It's what is taught, but I wouldn't say it's entirely accurate though unless the market is fully 'efficient', which we know it's not, or at least not all the time.

It's provably wrong.

The risk free rate of return is reduced because people want to leverage short term cash flows. A 99% chance of gaining 5% is not necessarily worth a 1% chance of losing 5%. EX: Collage tuition is paid before teachers salary's are paid, so collages want somewhere to stuff money for a weeks, but losing money is vastly worse than some minor gains.

If you model the stock market by say buying evenly from all stocks and selling in 50 years repeat. Then some outliers like dell at IPO going up 500x more than makes up for losses. But, you can still lose a lot of money over say 5 or even 20 years and people can't necessarily wait 50 years.