How our admit odds work
We compete on transparency, not false confidence. Here is exactly what goes into an AcceptanceAtlas admit-odds estimate, what it can and cannot see, and how to read the number without mistaking it for a promise.
The short version
What goes into the estimate
Every estimate is built from a small set of inputs we can actually measure. Nothing here is secret:
- The school's overall admit rate (U.S. Department of Education College Scorecard)
- The school's SAT / ACT mid-50% range (U.S. Department of Education College Scorecard)
- Your GPA
- Your test scores
- Your course rigor
- Which round you apply in (Early Decision, Early Action, or Regular Decision)
We compare your profile against the school's published numbers and translate that comparison into a probability band. That is the whole machine — there is no hidden scoring of you as a person.
What we do NOT claim to measure
Admissions is holistic, and the parts that often decide a real outcome are exactly the parts a public dataset cannot see. Our number does not account for:
- Your essays
- Your recommendations
- Your interviews
- Demonstrated interest, where a school does not publish how it weighs it
- Institutional priorities — the year's specific needs in majors, programs, geography, or class shaping
Any one of these can move a real decision in either direction. Our estimate cannot see them, so it cannot price them in. A strong essay or a needed major can carry an application past what the numbers suggest; a thin recommendation can sink one that looks safe. Read the odds knowing this whole layer is invisible to them.
Test-optional and test-blind
How a school treats testing changes what your estimate leans on.
- Test-blind (for example, the University of California): we ignore test scores entirely and lean on your GPA and course rigor versus admitted students. Submitting a score will not change the estimate, because the school will not consider it.
- Test-optional: published mid-50% ranges typically reflect only the students who chose to submit, who tend to be higher scorers. That can skew the range high. We surface this so you do not read a submitter-only band as the bar for everyone.
How to read a probability band
An estimate is shown as a range — for example, 34-42% — never as a single confident number. The width of the band reflects how much uncertainty there is: wider when the data is thinner or your profile sits near a threshold, tighter when the comparison is clean.
The most important thing to remember: one applicant is one outcome, not a percentage. A 40% estimate does not mean "mostly no" any more than 60% means "mostly yes." It means that, across many profiles like yours, roughly that share are admitted — and you will land on exactly one side of the line.
The tiers
To make a long college list readable at a glance, we group estimates into named tiers. The percentages below are the estimate bands behind each label:
| Tier | Estimated odds |
|---|---|
| Likely | ~78-92% |
| Target | ~52-68% |
| Borderline | ~34-46% |
| Reach | ~12-22% |
| Long shot | ~3-9% |
These are estimate bands, not guarantees. A school in your "Likely" tier can still say no, and a "Reach" can still say yes — the tiers help you balance a list, not predict a result.
Where the data comes from
Our facts come from public, authoritative sources:
- U.S. Department of Education College Scorecard — our primary source for admit rates and test-score ranges.
- IPEDS and the Common Data Set — used to cross-check figures and catch discrepancies.
We refresh data on the annual admissions cycle, and each figure is stamped with its source year so you always know how current it is. When sources disagree, we prefer the most recent College Scorecard figure and flag the gap rather than quietly averaging it away.
How we tag every figure's confidence
Behind the scenes, every value we store carries a source, a cycle year, and a confidence level, so nothing is shown without a provenance trail. The six levels:
| Confidence level | What it means |
|---|---|
| Official, current | From the institution's own current-cycle source (e.g. this year's admissions page). |
| Official, historical | From an official source for a prior or aggregate cycle (e.g. a College Scorecard admit rate). |
| Official, self-reported | Reported by the institution to an official program (e.g. Common Data Set C7 factors). |
| Manually verified | A person confirmed the figure against the official source. |
| Estimated | Modeled by us (your admission-fit band, a merit-aid range) and always labeled as an estimate. |
| Unknown | Not yet sourced — we show nothing rather than a guess. |
The rule underneath all of it: a blank beats a fabrication. If we can't source a number, we leave it out and point you to the college's own page.
Sources we deliberately do not use
We build only on free, official, primary sources — the Common Data Set, the College Scorecard, IPEDS, the Urban Institute's harmonized federal data, and each college's own admissions and financial-aid pages. We do not scrape, republish, or derive our numbers from ranking aggregators or "chance-me" engines such as U.S. News, Niche, Princeton Review, CollegeVine, Appily, or College Board's BigFuture. Their figures are paywalled, user-submitted, or the output of an undisclosed model — none of which we can verify, and none of which you should have to trust on faith.
Our honesty rules
The whole product rests on a few rules we will not break:
- Facts are sourced and dated. Admit rates and test ranges are attributed to the U.S. Department of Education College Scorecard and stamped with their year.
- Estimates are labeled. Anything we model — odds, tiers, merit-aid ranges — is shown as a range and marked as an estimate.
- We never present odds as a guarantee. Our number is directional guidance, not a guarantee of admission, and we will not dress it up as one.
What we have verified so far
We would rather be narrow and right than broad and wrong. Our catalog covers a growing set of colleges with official, sourced figures, and we add schools only as we verify each one against its primary source. Some deeper layers are intentionally incomplete: per-major admit rates appear only where a school actually publishes them, and demonstrated-interest weighting appears only where we've confirmed it against that school's Common Data Set. Where we don't yet have a figure, you'll see a prompt to verify on the college's own site — never an invented number to fill the gap.
Corrections
If a figure looks wrong or out of date, tell us at support@acceptanceatlas.com. We'll re-check it against the primary source and correct it. Accuracy is the whole product, so we treat error reports as a priority, not a nuisance.
Want to see your own list sorted into honest tiers, with every figure sourced and every estimate labeled? Get my admissions map — free to start.