Estimated acceptance rate to top 20 grad school?

[phun]

It's 5 am and I can't seem to fall asleep.
This random thought just hit me:
Say each of the top 100 undergraduate programs generate on average 20 graduates with physics degree (To me it seems like this is an overestimation, but let's say it's the upper limit) every year. If you assume that 15 of those 20 graduates apply to grad school, that will make a national applicant pool of 1500 people.
If 15 domestic applicants on average enter each of the top 20 departments, This would mean 1 out of 5 applicants enter top 20 departments.
Does this seem at all like a reasonable estimate?

 

[weichi]

If you go to www.gradschoolshopper.com, you can find pdf files with lots of info on all physics grad school programs. Usually they they you what % of applicants were accepted for the most recent year.

From my investigation, most schools, not just the "top 20", accept 10-20% of their applicants. Closer to 10% for most of the "name" schools, but not all. I found 11% for MIT, 10% Illinois, 20% Cornell, 20% Chicago, etc.

 

[leright]

What are the chances of a guy with a dual degree in engineering physics and electrical engineering with a summer of an engineering internship and (hopefully) a summer with an REU getting into a (hopefully semiconductor physics/solid state technology research) top grad program (MIT, cornell, UM ann arbor, UIUC, etc). I have a while before I finish, but I think I am looking at a 3.6 GPA or so. Hopefully I have good GRE scores and good letters of recommendation.