I think it would be reasonable to assume that the percentage of coins given a particular grade when submitted to NGC would be a close representation of the percentage of your likelihood to receive that same grade. For example, let's take the Roosevelt dime from the above 1981 proof set. Let's assume it is a Type 1, it has no visible marks or hairlines, and it has enough frost on both sides to qualify for the Ultra Cameo designation. (As Coinbuf mentioned, if it does not qualify for the UC designation, it is probably not worth the cost. And if it has any visible defects, it certainly wouldn't grade high enough to be worth it. ) Of the 1,696 coins designated by NGC as PFUC, 7 of them received a grade of PF67, 288 got a 68, 1176 got a 69, and 225 received a grade of PF70. Judging by that, your chances of receiving a 69 are 5 times greater than your chances of receiving a 70. That is obviously not scientific, but it should give you a fair idea of what grades you could expect to receive.