I found it completely impossible (and I did try) to pass university level maths courses without carefully doing every tutorial.

By "homework" i mean specifically comparing cohorts of students in high-school that do and do not do the homework they're set, and connecting that with long-term grade trajectories.

I dont have the sources to hand, but i recall an economist talking about large studies in this area.

I think some of this might be that high school tends to be quite easy for the brightest students. I did close to none of my homework all the way through high school because I didn't have to (paying attention in class was sufficient for me to graduate with mostly A's and get into the university of my choice).

Well, yeah... sure, in high school the material is so easy and the pace is so slow that someone reasonably "smart" can do very well without doing homework.

Just because that's the case it doesn't follow that "homework doesn't improve grades."