...
...
...
This is awesome ultraBTC, thanks for sharing this information and I hope you don't get in trouble for doing it because this is something that users shouldn't know at all.
What is crazy for me is that they simulate 10 billion spins and that's how they estimate the RTP. But I can be sure their simulation was always the same amount, and if the users always bet different amounts then the RTP should change.
The thing is, for extremely high-variance slot games (like this one), simulation goes in billions of spins. (You will notice that in this particular game, the chance of highest payout is 1:4million spins/without buying a free spins feature). So with such a high variance, the sample needs to go in billions to approach the theoretical mean. Less is probably required to be within SD (Standard Deviation). It's the Law of Large Numbers.
For low/middle variance slots, it's in terms of millions of spins or less depending on volatility.
Btw, bet size should not affect RTP/game-play.