It's not a gamble, it's an application of the law of large numbers. But yes, changes in the underlying assumptions (e.g. mortality rates) can make the whole calculation untenable.