My guess is that some of the seemingly strange omissions have to do with size. The top NESCACs may not be large enough to show up on the lists of some of the Ivies despite the fact that a high percentage of the former also apply to the latter. To put it numerically, Williams had 9,560 applicants in 2018. Harvard had 42,749. Stanford had 47,452. If 21% of Stanford applicants also applied to Harvard they would make up a larger overlap than if every Williams applicant also applied to Harvard. (Or 10.5%/50%, 5.25%/25%…)