@Thomas Shey is right: I know we all think we are great at spotting patterns, but we are not. We are terrible at it. Saying "I can always tell when a GM is changing some math" is a useless statement -- you cannot notice that you do not notice it! It would require an active test. And you would fail it, almost certainly.
I did actually do this as a test once. I ran four D&D 4E sessions where I modified every d20 roll to round down to the nearest multiple of 5 (so I just used results of 0, 5, 10, 15, 20). I also modified all NPC stats to be multiples of 5.
No one noticed. Not even slightly. Not even when I asked about it. It is very hard for people to spot statistical patterns -- we are not designed for it. Our reward system makes us see patterns where there aren't any (penalty for thinking that pattern in the bush is a tiger when it isn't -- you look silly as you scream and run. Penalty for thinking a tiger isn't actually a tiger -- you do not have offspring) so trying to evaluate fairness is exceptionally hard for us.
Practical application: If you want to be a professional gambler, bet on events where you betting against other people whoa re evaluating odds. They are going to be terrible at it, so you will have an edge. I had a friend who did this for at least a decade, betting on unusual events occurring more often than people expect them to.