So you're hypothesis is: after a certain point it's better to use other stats besides focusing on making sure your accuracy is capped?

I'm I correct in understanding what you wrote?