Yes, it is the main way I determine which types of questions to drill. My only question is how expected accuracy is determined (and whether it shifts over time based on drill/PT performance)