My feedAdmissions Advice
48
@matt.kaye
expert
4 answers, 118 votes
 • 2 years ago •
Admissions Advice
[edited]

Fall 2022 Chancing Updates

Hey everyone!

I’m Matt -- I work on data science projects here at CollegeVine and have been around in the past talking about some of the methodology behind our chancing model.

As we do every few months, we're making some great improvements to the chancing engine this week! We do this to keep our chancing as accurate as possible as we get new information and data. Here's what's new:

1. We updated how we forecast acceptance rates in our model. This makes your chances even more accurate for this admissions cycle.

2. We improved the model to better reflect the impact of applying Early Decision. If you're applying ED, you should see your chances go up a bit.

3. We created a separate chancing methodology for test-blind vs. test-aware schools (schools that don't consider standardized tests at all vs. those that do consider them). Our model now does a better job of treating your profile the same as a test-blind school would.

As a result of these updates, you'll likely see your chances change, and that is totally normal. Please leave a comment if you have any questions or reach out to support@collegevine.com. Happy Fall and best of luck with this admissions cycle!

Chancing
48
29
@AimForTheStars2 years ago

Thank you so much! Hope you have a good fall too!

@ADG12 years ago

At Chicago, ED1 and ED2 chances are the same. This is counter intuitive. What do you think?

@ADG12 years ago

Also, has CV ever looked at the correlation of acceptance/rejection between schools? This is obviously critical in assessing how many schools to apply to. Wouldo also be great to know the standard error of your predictions (by school).

@leepalmer2 years ago [edited]

I really appreciate it. I hope you enjoy the fall too! vex 7

Earn karma by helping others:

1 karma for each ⬆️ upvote on your answer, and 20 karma if your answer is marked accepted.

3 answers

18
@Liltrooper
223 answers, 448 votes
 • 2 years ago[edited]

@matt.kaye It's great your team continues to make improvements.

One major improvement your team can do in a few days would be to post accurate acceptance rates for the CV colleges you have in your database. They are 99% INCORRECT, some by a factor of 3 times like NorthEastern which is like 6.8%, not 20%. If someone has 20 or 30 schools on their "my list" and all the known acceptance rates are wrong, even if your chancing percentage is correct within a margin of 5%, it really doesn't give anyone context. So if your CV says your probability to NorthEastern is 10% but the wrong admit rate is posted as 20%, then you immediately think you shouldn't apply because is twice worse. But if you post the real acceptance rate of 6.8%, then 10% is better than the average acceptance rate. Any HS kid can look at either the college common data set, the college website and search freshman profile, or look up acceptance rates from the college's Twitter or into account. I don't understand why this consistently is an important DATA point that is incorrect on all of your colleges in your database. Whoever is in charge of sourcing this data is getting it from the wrong place. Why? because if you go to any of CV competitors they have more accurate numbers and that just makes CV look bad.

https://www.collegekickstart.com/blog/item/class-of-2026-admission-results

https://www.ivywise.com/blog/college-admission-rates/

https://internationalcollegecounselors.com/class-of-2026-admissions-statistics/

https://www.collegetransitions.com/blog/2026-ivy-league-acceptance-rates/

https://collegiategateway.com/regular-admissions-trends-for-the-class-of-2026/

@CamreronBameron posted accurate 3 years' worth of acceptance rates for 2 years on CV but you guys ignore her posts.

https://www.collegevine.com/questions/27562/top-50-latest-class-of-2026-acceptance-rates

https://www.collegevine.com/questions/12065/i-m-keeping-tabs-on-class-of-2025-stats-can-you-help

The same goes with Ivys/Elites/LACS. CV has Cornell at 11%, Dartmouth at 9%, Williams at 15%, Columbia 7% etc. So these are all wrong by almost a factor of 2 times. Emory 19% when it's really 10.7%. Tufts 16% when it's really 9%. The CV chancing engine is using acceptance rates from 5-6 cycles ago or older and if anything thing admits rates continue to push lower each cycle not revert to older rates. Some brought this up last year but nothing was done. What's the logic and thinking for comparing revised upgraded new CV probability chancing rates to non-existent old acceptance rates that haven't been around for in some cases more than 7 cycles? Columbia hasn't had an admit rate of 7% may be back to 2015 when it was 6.9%

Sample of what I'm talking about:

School name- 2016 2017 2018 2019 2020 2021 2022

Caltech 7.89% 7.74% 6.62% 6.42% 6.69% 3.92% 3.03%

Harvard 5.22% 5.20% 4.59% 4.50% 4.89% 3.41% 3.19%

Columbia 6.04% 5.80% 5.51% 5.14% 6.15% 3.89% 3.73%

Stanford 4.69% 4.65% 4.34% 4.34% 5.19% 3.95% 3.86%

MIT 7.81% 7.10% 6.74% 6.62% 7.26% 4.03% 3.96%

Princeton 6.46% 6.09% 5.49% 6.05% 5.55% 4.38% 4.36%

Yale 6.27% 6.91% 6.31% 5.91% 6.54% 4.62% 4.47%

Brown 9.01% 8.32% 7.67% 7.07% 7.67% 5.45% 5.03%

Pomona 9.17% 8.19% 6.96% 6.98% 8.62% 6.64% 6.03%

Vanderbilt 10.36% 8.56% 7.29% 7.90% 9.84% 6.70% 6.13%

UPenn 9.41% 9.15% 8.39% 7.47% 8.07% 5.87% 6.14%

Duke 10.31% 9.23% 8.30% 7.17% 7.67% 5.76% 6.17%

Dartmouth 10.52% 10.44% 8.74% 7.93% 8.79% 6.17% 6.24%

JHU 11.43% 11.79% 9.94% 9.15% 7.05% 6.43% 6.48%

NorthEastern 28.88% 27.44% 19.00% 18.00% 20.47% 18.38% 6.79%

18
@AimForTheStars2 years ago

...big list

@alyssa33
expert
2 years ago [edited]

Hi @Liltrooper - thanks for writing in! We currently display the acceptance rates from the 2020-2021 IPEDS data. As new IPEDS data becomes available, we will update it again! For chancing, we also use the trend of the past few years of IPEDS acceptance rates to predict what the next year's acceptance rate will be. Feel free to check out our similar discussion from the May 2022 chancing updates: https://www.collegevine.com/community/object/comment/14299

@Liltrooper2 years ago [edited]

@alyssa33 There is free, readily available accurate acceptance data provided directly from the colleges themselves. It's far better to use data from college websites, blogs, Instagram, Facebook, Twitter, and college newspapers than IPEDS. And Common Data Sets from 2021-22 are readily available for 99% of the CV colleges covered.

@Liltrooper2 years ago [edited]

@alyssa33 Just type in Class of 2025 or Class of 2026 profile as a google search term for a college. https://www.college.columbia.edu/news/columbia-announces-class-2026-admissions-decisions

1
@GeorgetteKonaduPoku
1 answer, 1 vote
 • a year ago

Thanks!I hope you enjoy the fall too

1
1
@tlund
1 answer, 5 votes
 • a year ago

Hi @matt.kaye there are some people on this blog that have answered my questions and basically say your platform is not accurate at all. Frustrating as I have scheduled college tours based on us having target and safety changes on these schools. How accurate do you really think this platform is? not sure why there are folks on this blog that are so negative on it. Please advise.

thanks

1
[🎤 AUTHOR]@matt.kaye
expert
a year ago

(1/X) Hey @tlund, sorry for just circling back. I saw your other question and the thread -- here are my general thoughts: I think it's extremely naive to say something like "algorithms can't predict college admissions" on the basis of them being decided by a human. Ultimately this is just my opinion, but it's my belief that college admissions is actually a relatively simple machine learning problem (relative to more difficult ones, I mean)

[🎤 AUTHOR]@matt.kaye
expert
a year ago

(2/X) The reason I say that is because of the amount of confidence you could have in a guess about your chances given the data we have relative to (e.g.) trying to make cars drive themselves (which I see as a much more complicated ML problem). The simplest way to think about this is to think about what you would want to know in order to make a confident guess about your chances. The top few things on that list are probably some combination of the school's accept rate and how your academics, ECs

[🎤 AUTHOR]@matt.kaye
expert
a year ago

(3/X) and demographics stack up relative to the average accepted student at a school. Assuming you agree with that assessment, then CV Chancing has the majority of the most important ingredients in the recipe.

[🎤 AUTHOR]@matt.kaye
expert
a year ago

(4/X) For some additional context, it might be helpful to compare how our chancing model does to other things that are often predicted in the world. One we like to reference is that our model does about as well at predicting your chances as the FiveThirtyEight NBA model does at predicting the outcomes of NBA games. Of course, their model is not always right, but it'd also be naive to say that predicting NBA games is entirely random. It's not -- you can make educated predictions about NBA games

What are your chances of acceptance?

Calculate for all schools

Your chance of acceptance
Duke University
UCLA
+ add school
Your chancing factors
Unweighted GPA: 3.7
1.0
4.0
SAT: 720 math
200
800
| 800 verbal
200
800

Extracurriculars

Low accuracy (4 of 18 factors)
Add more factors ›

Community Guidelines

To keep this community safe and supportive:

  1. Be kind and respectful!
  2. Keep posts relevant to college admissions and high school.
  3. Don’t ask “chance-me” questions. Use CollegeVine’s chancing instead!

How karma works