I think this article has been discussed before here. The data is still out of date, based on a 2014 article in Psychology Today. At the latest, the data would have come from the 2013-14 CDS, high school class of 2013/college class of 2017, now six admissions seasons ago.
In case any students might think to use this list for the purpose of determining reaches, matches, and safeties, it is important to note that these SAT numbers refer to the Old SAT, not the current test.
Let me first state that I’m highly skeptical of any rankings. Almost all of them take a bunch of questionable data, then add arbitrary weightings to them to come up with single numerical score. Rankings would change if these weightings change and everyone would, or at least should, weight these factors differently.
I also want to comment on Caltech’s relatively low yield rate. Its graduation rate is also not among the highest. They’re not unexpected, however, because of the size of the school and the intensity of its science curriculum. If a student there struggles in her/his chosen field, s/he wouldn’t have nearly as many alternatives as, say a Harvard or even MIT student. When Caltech accepts an applicant and deems that student “can do the work”, it truly means something different than nearly all its peers. If all Caltech students were transferred to Harvard, almost all of them would likely succeed. However, the reverse is clearly not true. The vast majority of Harvard students are not likely to succeed at Caltech.