Methodology · Pre-registered
How we measure ourselves
Recruiting guidance is easy to give. Showing what happened afterward is harder. Prospecta will publish both.
This page was written before the first athlete in this calibration cohort was scored. The rules are being set now so they cannot move later to make the results look better.
What we're testing
Prospecta gives athletes an evidence-based read on the level of college programs that currently fit their profile.
We're testing a simple question:
How closely does that read match where athletes ultimately appear on a college roster?
We're starting with women's lacrosse.
For each cohort assessment, Prospecta preserves the inputs, prediction, confidence level, scoring state, benchmark set and tier mapping in effect when the assessment was made.
What we'll report
Exact hit. Prospecta's predicted tier matches the athlete's verified outcome tier.
Within one tier. Prospecta's prediction lands either on the athlete's verified tier or immediately adjacent to it.
We will report both. The tier mapping used for the cohort is established before scoring begins and the scoring state used for each prediction is preserved with that cohort record.
The bar, set in advance
Our bar is 70% within one tier across eligible profiles.
We chose that number ourselves; it's a judgment, not a figure derived from published research. We're stating it in advance anyway, because a standard set after the fact isn't a standard.
Every eligible athlete counts regardless of Prospecta's confidence level.
Confidence is evaluated separately. Higher-confidence predictions should prove more reliable than lower-confidence predictions. If they don't, we'll say so.
We publish the misses too
A miss remains a miss.
Afterward, we may classify what contributed to it, such as a model judgment, incomplete or incorrect athlete input, a materially changed recruiting trajectory, or an unusual circumstance.
That explanation does not remove the athlete from the result.
A case can only be excluded under rules established before scoring:
- Incomplete required inputs.
- Outside the defined cohort scope.
- A required historical input cannot be reconstructed with reasonable confidence.
Optional inputs do not determine eligibility.
We will not exclude an athlete later because the result was awkward or because Prospecta got the prediction wrong.
Small samples stay small samples
Until at least 25 athletes have been graded, we will publish the raw results as an initial calibration set:
- Exact hits.
- Within-one-tier results.
- Misses.
- Sample size.
We will not publish an accuracy percentage below 25 graded profiles.
Once the sample is large enough to report percentages, every percentage will include its sample size.
Not all evidence is equal
We use three levels of evidence.
1. Prospective
Prospecta makes its assessment before the recruiting outcome is known.
This is the strongest evidence.
2. Retrospective with verified historical inputs
The outcome is already known, but the historical information used to reconstruct the athlete's earlier profile can be independently verified.
3. Retrospective with self-reported historical inputs
The outcome is already known and some historical inputs depend on the athlete's or family's recollection.
When a retrospective assessment materially combines verified and self-reported historical inputs, we label it as mixed rather than treating it as fully verified.
We keep these evidence types separate.
When Prospecta eventually publishes a headline performance figure, the prospective cohort will be the primary evidence.
Retrospective results are calibration and supporting evidence. We are committing to that now so we cannot later lead with whichever result looks better.
How outcomes are verified
A family telling us where an athlete committed is not enough.
The primary outcome is the athlete's first verified appearance on a college roster for the entering class being evaluated, using public college roster information collected through Prospecta's roster data system.
This verification is independent of the family's self-reported outcome.
If an athlete initially commits somewhere and later enrolls and rosters elsewhere, the roster placement is the outcome.
A redshirt still counts as rostered.
A later transfer does not change the original recorded placement. The first roster placement stands.
For prospective participants who do not appear on a qualifying roster by the predefined verification cutoff, we report no verified roster placement by cutoff rather than removing the athlete after the fact.
What this does not tell you
The retrospective portion of this work begins with athletes who ultimately reached a college roster.
That creates an important limitation.
It evaluates Prospecta's ability to read recruiting level among athletes who were ultimately placed.
It does not measure an athlete's overall probability of being recruited or making a college roster.
That limitation will accompany any performance result we publish from the retrospective cohort.
Prospective cohorts provide stronger evidence because Prospecta's assessment is recorded before the final outcome is known.
Who grades the first cohort
The first cohort will be graded by Prospecta's founder, and that will be disclosed.
The grader applies the predefined tier mapping.
The underlying placement outcome itself is not determined by the grader. It comes from the verified roster record.
The commitment
We are doing this because Prospecta should be accountable to the same standard we ask families to use:
Look at the evidence. Show your work. Change your mind when the evidence says you should.
If the results are strong, we'll publish them.
If they're weak, we'll publish that too, explain what we learned, and show what changed.
Published August 7, 2026, before scoring began for this cohort.