Originally Posted by JellyWoW

See you are correct, but there is actually a small issue with that. Actually there are 2 and I can go through and explain them but I will save you the math.

The first issue, which is true for statistics in general, is that the 50% chance is only true over very large data sets. Thus everyone's individual rolls are not going to be true for that, by a long shot. Especially for the fact that rainbows are a 0.5% chance. You would need to pull gacha 1057 times to even start to get a good approximation (just trust the math, I rather not explain how you get the number).

The other issue is how computers get random numbers, idk if this actually applies to fkg, but numbers are actually only psudeo-random. If FKG actually uses truly random methods, power to them, but because of how expensive these methods can be do to enact I highly doubt it. In essence the number that precedes a value can actually influence the number that comes after. The effect is minor and would require you to have a detailed understanding of how RNG tables are actually generated to understand. In the short term, we do not notice this because the effect is minor as I noted, but does help explain how people can go over 400 pulls without a rainbow when some people get them almost every other pull. As a hint, if this is true, pulling directly after you see someone get a rainbow, would probably not be the best idea, although we are talking an effect that is really only a .05-0.1% deviation over the course of 1000's of pulls.

I'm 100% certain they're pseudo-randoms. Not only that, but it also seems the randomizer is occasionally "slow" with updating its seed (it's a function of the internal clock of the hardware). Ever noticed how often your girls make a beeline on maps with random nodes, especially when they all happen to cross the same nodes very shortly (on the same move) after another? There's your answer.