An event with only a 1 percent chance of occurring is not guaranteed to eventually occur. But empirically, an event with a 1 percent probability will occur an average of approximately 1 time in every 100 tests.

So if you run 1000 tests, empirically, we would expect to see 10 occurrences. Experimentally, this turns out to be the case. Even though results for each set may vary above and below what the probability predicts should occur, as the number of sets increases, the average of all the sets converges on the probability.