How the Forecast works.

Pipeline

1. 1

   Structured input

2. 2

   University dataset

3. 3

   Profile analysis

4. 4

   Rules, scoring and pathways

5. 5

   AI interpretation

Scores and positions are decided entirely in steps one to four, by deterministic rules. The same answers always produce the same result. Model version: ivy-forecast-rules-0.2.

Weights

The overall Ivy Profile Score is a weighted average of the dimensions provided. A missing dimension is removed and the remaining weights are rescaled, so a student who has not tested is not penalised for it.

Ratings

Each dimension is shown as a rating first and a score second, so a few points of difference never read as more precise than they are.

Positions

Each university is assigned a selectivity band. A profile's position depends on its overall score and, for the top two positions, minimum academic and rigor scores.

Below the developing threshold, the position is significant opportunity.

Pathways

Pathways read the same evidence differently: which routes it supports, not how competitive it is. Signals include activity types, self-assessed skills, areas of curiosity and stated ambitions. Each pathway is labelled strong fit, emerging fit, worth exploring, not indicated yet, and always shown with the evidence behind it. A hybrid appears when a strong academic profile sits alongside an emerging independent path.

• Selective university

  Grades and rigor, then Testing, then Depth and leadership, then Positioning, then Selective university

• Research

  A real question, then Mentor or lab, then Paper or presentation, then University research program, then Advanced research

• Builder / entrepreneur

  Technical project, then Prototype, then First users, then Iteration, then Startup or engineering role

• Creative / interdisciplinary

  Body of work, then Audience, then Recognition, then Program or studio, then Creative career

• Hybrid

  Independent project, then Research or building, then Selective university, then Venture or advanced study

Pathways are possibilities. They are never predictions, and never guarantees.

Data

Every result is labelled as fact, interpretation or possibility. Facts are what you entered and university data, each shown with its source, date and confidence. University records are maintained separately from the model so they can be updated without changing how profiles are scored. Selectivity bands are an editorial classification. Institution-specific admit rates and test ranges are added only from each university's Common Data Set, with the source and year recorded. ACT scores are converted using the official ACT/SAT concordance tables.

AI

After scoring, an AI model may write a short plain-language interpretation. It receives only the finished assessment, never raw answers, and is instructed not to add statistics or odds. Its output is checked automatically; anything containing a number not in the assessment, or any language about chances or probability, is discarded and the rules-based summary is shown instead.

Limits

• The Forecast cannot read essays, recommendations, interviews or the quality of an activity, only its shape.
• It does not know a university's institutional priorities in a given year, which can outweigh any profile.
• Thresholds are set by Ivy's judgement and will be recalibrated as outcome data from Ivy students is collected, with consent.
• It is an analytical estimate, not a prediction or guarantee of admission.