I dont see the issue with using the expected accuracy of people in your current score range to assign priorities. At a higher score the expected accuracy will be higher for every question type, so I don't see why the priorites would change much if at all.

For example lets say in your highest priority question type you are 10% less accurate than others in your score range, you would be even less accurate in that question type compared to others in a higher score range. Now, Im sure accuracy on each question type does not increase uniformly with score, but I dont think slight variations would change the priorities.