It is interesting that you say that because I had huge variations in my test results, dropping from top of my class to the bottom in an instant, then back to the average.

At one point I met a guy who was working on ways for computers to solve math problems. He told me that early college level problems are actually easier to solve than high school problems for his algorithms. The reason is that while college problems have more steps, and are therefore harder for students, you essentially just need to string theorems together. High school level problems are simpler (less steps) but require significantly more intuition.

And that's what I think happened with my grades: I was never good at math, but I had good intuition. And the switch from using intuition to using actual rigorous mathematical thinking killed me. It was like a completely different subject, and it took me years to catch up.

I wouldn't say if was it was different when it comes to applying methods vs constructing models.