That prediction is waaaaay off, if you invested then you should know that
No its not. It's called variance and happens in the short term. If you know daily betting volume and the houses edge, you can easily calculate what your expected profit should be. In the long run you will be right on, in the short run, you'll show variance. It's called math and it's absolute.
No, it's just wrong. He's saying you can divide the returns the bankroll should earn over a year by the bankroll amount to get an estimate of your expected earnings on an annual percentage basis.
Suppose the bankroll is 2000 BTC and the expected earnings are 1000 BTC per year. Using QS's formula that gives 1000/2000 = 0.5 as his annual percentage expected earnings whereas in reality earning 1000 from 2000 is more like 50%. I think he forgot a factor of 100 somewhere.