Yup. Rule of large numbers basically says that even if something is unlikely to happen, a large enough dataset will essentially ensure it happens. As the number of samples increases, the real-world outcomes will approach the expected values. Two die rolls on a d6 won’t always average to 3.5, but a thousand rolls will probably be pretty damned close to the 3.5 average. Something that happens based on a bell curve will approach that theoretical bell curve as more samples are collected.

Imagine a die that rolls a random number from one to a million. Your odds of rolling a 1 are pretty low. Quite literally one in a million. But now imagine that the die is rolled ten trillion times. Chances are very good that a 1 was rolled several times. Even though it only had a 0.0001% chance of happening, it still happened eventually. Because it had a chance of happening, a large enough dataset will essentially ensure it does happen.

This is pretty much how the lotto works. Each individual player only has a minuscule chance of winning. But when there are millions of players, the odds of someone winning are actually pretty good.