Methodology & Prototype
How Our Admissions Model Works
A structured account of the quantitative approach behind Nadia Moore’s admissions decision-support system: what it takes as input, how it treats a college list as a portfolio, and what stage of development it is actually at.
This page describes methodology rather than results. The system is a research-based prototype; it has not been calibrated or validated against institutional outcome data.
Six layers of the model
The model is deliberately legible. Each layer below can be inspected, questioned, and disagreed with — which matters more than apparent precision when the underlying system being modeled is a human committee. The published admissions research these layers draw on is collected in Admissions Intelligence.
Layer One
Student Profile Factors
The model begins with the attributes an application actually carries: academic record and course rigor relative to what the student's own high school offers, testing where submitted, intended field of study, geographic context, and the depth and duration of a student's substantive work outside the classroom. These are represented as structured inputs rather than a single composite score, so their individual influence stays visible and can be interrogated.
Layer Two
School-Specific Factors
A published acceptance rate is an average across enormously different applicants. The model instead represents institution-level structure: differences between application rounds, residency and in-state enrollment structure where a public university operates under one, variation by intended major, and reported yield behavior. Sources are public — institutional disclosures, common data sets, and published admissions research.
Layer Three
Portfolio Construction
Applications from one student are not independent events; the same profile drives all of them, so their outcomes are correlated. The model therefore treats the college list as the unit of analysis and examines the distribution of outcomes across the whole set — the likelihood of at least one admit the student would enroll at, and how concentrated the list is in a single tier or institutional type.
Layer Four
Early Decision Considerations
A binding Early Decision application is a one-time, non-transferable commitment, and published early-round admit rates reflect a different applicant pool than regular-round rates rather than a straightforward bonus. The model examines Early Decision as a placement question at the portfolio level: what committing that single application to a given institution does to the shape of the entire list.
Layer Five
Scenario Analysis & Simulation
The prototype's practical output is comparative, not oracular. It simulates a specific list configuration, then re-simulates it with a school added, removed, or moved between rounds, so the difference between two candidate strategies can be examined directly. The useful signal is the change between scenarios under stated assumptions — not a precise probability attached to any one school.
Layer Six
Institutional Calibration — The Production Roadmap
Today the model is fitted to public data and research-based priors, which describes a national picture rather than any particular school's population. The production roadmap is to calibrate against anonymized historical application and outcome data contributed by participating institutions, and to backtest estimates against realized outcomes for comparable prior-cycle students. That work has not yet been performed.
What We Do Not Claim
The limits of the current model, stated plainly
Admissions committees make holistic human judgments under institutional priorities that are not fully public and that shift between cycles. No model can render that deterministic, and we do not claim ours predicts committee decisions. What a well-specified quantitative model can do is make assumptions explicit, keep the comparison between two strategies consistent, and surface portfolio-level risk that per-school intuition routinely misses.
The current implementation is a research-based prototype built on public admissions data and research-based priors. It has not been calibrated on institutional outcome data, and its estimates have not been backtested against realized results. Moving from that state to institution-specific calibration is the entire subject of the work we are proposing to counselors and schools.
See the prototype run against your own cohort.
Demonstrations are conducted directly by the founder for independent educational consultants and school counseling offices, using a real or representative portfolio from your practice. More about the product and its current stage is on the counselors and schools page, and the company overview is on our homepage.
Families seeking individual guidance rather than software can learn about our four-year advisory practice, which serves a capped cohort of twelve families per cycle.