Methodology & model card

How our estimates work — and where they stop.

We believe a number is only useful if you understand how it was made. This page explains, in plain language, how our models work, what they can tell you, and — just as importantly — what they cannot.

Acceptance model accept-v0.2 · maintained by the AdmitQuant research team · last reviewed June 2026

The core model

The acceptance-rate estimate

When you build a student profile, we produce an for each school. Think of it as a statistical estimate: a structured, evidence-based read of how a student’s record compares to what a given institution has historically admitted.

The model considers broad categories of input rather than any single score. Those categories are:

Academics

Grades and the strength of an academic record over time, read in the context of a student's school.

Test scores

Standardized scores where submitted, interpreted against published score ranges for each institution.

Rigor

The difficulty of the courses taken — honors, AP, IB, dual enrollment — relative to what was available.

Context

Circumstances that shape how an application is read, such as school setting and opportunities available.

Soft factors

Qualitative signals like activities and demonstrated interests, captured at a high level — never a measure of a person's worth.

We do not publish the specific weights the model places on each category. Those weights are proprietary, and — more to the point — exact weights would imply a precision that admissions simply does not have. What matters is that the output is an estimate, not a measurement, and that it is built from the categories above.

Honesty first

What this is not

We would rather under-promise than mislead. A few things we want to be unambiguous about:

We are not affiliated with any college and cannot know or influence admissions decisions. No part of our software communicates with admissions offices.
Probabilities are themselves uncertain. An estimate is our best read of the available evidence, not a count of a fixed reality — treat a number as a range, not a verdict.
We deliberately avoid false precision. Admissions involves human judgment and factors no model can observe, so we present estimates plainly rather than implying more certainty than exists.
Selective admissions are noisy by nature. Two similar applicants can see different outcomes, and our estimates describe tendencies across many cases, not the destiny of any one student.

Portfolio view

Monte Carlo simulation

Your college list is a portfolio, and any single admit probability only tells part of the story. runs your full list through many simulated admissions seasons — thousands of “what if” runs — to show the range of likely outcomes rather than a single guess.

The result helps answer practical questions: How likely is at least one admit from this list? Is the list balanced between ambitious and safer choices? What does a typical season look like, and what does a disappointing one look like?

What it does not do is predict your actual result. It cannot tell you which schools will admit you, and because it builds on the estimates, it inherits their uncertainty. It is a tool for understanding risk and balance, not a forecast of your specific spring.

Knowing what you want

Conjoint analysis

Families often say they want one thing and choose another. is a well-established research technique that reveals preference trade-offs from the choices you actually make — comparing options that vary across like cost, location, size, and academic fit.

The output is a clearer picture of how you weigh those attributes against one another, so your list reflects your real priorities rather than first impressions. It is not a personality test, and it makes no claim about who anyone is. It simply describes the trade-offs .

Where the data comes from

Data sources

Our foundation is public, authoritative data — including the U.S. Department of Education’s College Scorecard and IPEDS (the Integrated Postsecondary Education Data System), supplemented by each institution’s published Common Data Set.

On top of that, our research agents continuously gather and verify information from primary sources, such as institutions’ own published figures. Every data point is source-tracked, so the information behind an estimate can be traced back to where it came from. Published figures change over time, and there can be lags between an institution’s update and ours.

Keep exploring

Put the methodology to work

Estimates, not guarantees.

Everything our software produces is a statistical estimate built from imperfect, changing data about an inherently uncertain process. We are not affiliated with any college and cannot know or influence admissions decisions. Please use our outputs as one input among many — alongside your own research, your student’s counselor, and your family’s judgment.