Makes sense if you run it once per year and use the results to determine the admissions list. This assumes your goal is to maximize the number of doctors trained each year given a limited number of slots available. If you can take it whenever you want and it is just another hurdle to get into med school, then I'm not sure what value your score relative to your cohort is. Maybe it assumes all classes should have equally capable applicants, and since the content of a test would necessarily change to prevent cheating, they can't guarantee that all tests will be of equal difficulty. So to even that out, they just take the top performers from each cohort.