Analytics
Computed against everyone, then scoped to you.
Our analytics are built on a global ecosystem, not one school's spreadsheet. Every metric is computed across the full population and then narrowed, which is the reason a number means what it says it means.
Why these numbers hold up
A three touch EPA per touch of 1.94 tops any leaderboard and means nothing. Most products ship it anyway, because the row looks the same as every other row. We mark it instead.
Scoped to a comparable population for the season being asked about, not pooled across every division that happens to be in the table.
Snaps taken in decided games are removed before anything is computed, so production is measured where it was contested.
The sample travels with the figure. A rate is never shown without what it was taken over.
Thin evidence is marked as thin rather than rounded into a ranking. We would rather show nothing than show a leader who had three touches.
What is global, what is yours
Every figure on this page sits in one of three scopes, and the scope is decided before anything is computed. Global facts are shared: they describe the league and are the same number no matter who is asking. Org facts are private to a school and never leave it. The interesting numbers are the third kind, where a global benchmark is read against your own economics.
Computed across shared league data, so the same figure is usable by every school.
Stars, class ranks and recruit history are shared market facts about a player.
The benchmark population is league-wide, which is what makes a percentile mean anything.
Where players went and what happened next is shared ecosystem data.
A player belongs to a specific school's roster in a specific season.
Private school economics. It is yours, and it is not part of any shared population.
School-specific evaluation and workflow, not a view of somebody else's board.
Who was offered, who decided and who owns the relationship stays inside the program.
School-generated proprietary data. Your grades on a player are not anyone else's input.
Derived entirely from that school's own activity.
Global player performance combined with private org economics.
A global production benchmark read against that school's own pay bands.
Your roster and spend measured with global performance primitives.
Global player outcomes narrowed to that school's historical recruiting classes.
The six families
An org output. Pedigree, production, positional scarcity, positional market premium and eligibility horizon combine into a value score, a dollar figure, a cap percentage and a tier. This is NIL cap allocation.
Global. Four grades, all percentile ranked within position group, and NULL rather than zero when evidence is thin. S-Grade for production volume, P-Grade for efficiency per opportunity, G-Grade for year over year change.
An org output, and the most distinctly ours. A signed number: a player's composite minus the median composite of their own pay band. Plus 40 means outperforming everyone paid like them.
Global. Twenty player metrics across roughly 181k rows plus team metrics. The conventional ones, and six that are ours: leverage usage share, team dependency, deployment breadth, explosive dependency, situational EPA delta and usage trend.
The discipline above, treated as a feature rather than a footnote. It is applied before computation, not filtered afterwards.
Every figure traces to its source and its denominator. A number whose lineage cannot be reconstructed is a number nobody should act on.
Data model
Five layers, one schema.
League data and your own work are the same kind of record, held in one place instead of stitched together by export. Every entry carries an owner, a status, the evidence behind it and who approved it, which is what lets a question cross layers without anything being copied out.
- 05Agents
The unified data used directly inside Claude, ChatGPT, our own agent framework, or anything else you connect. Each one inherits exactly what its holder's role opens and nothing beyond it, so delegating work can never widen access, and every figure comes back with its source and its denominator attached.
- 04Runtimes
Every way you might actually use it. Models your staff builds and can inspect, jobs that recompute when Saturday's charting lands, reports and visuals, and the same records reachable over the API from any third party tool you already work in: a notebook, Cursor, your own application. The store does not care which one you reach for.
- 03Integrations, custom data
Read and write, both directions. We pull your evaluations, spend history, contact logs and practice load out of the products they are stuck inside, and write back to those same products: a board exports to the sheet your staff already keeps, an agent's message goes out from the coach's own inbox. Your data becomes flexible rather than locked to whichever tool holds it.
- 02Team model
The same population narrowed to your program, with your history inside it rather than beside it. What your last three classes actually produced by position, which of your own evaluations held up, and where your trends sit against the field instead of against last year.
- 01Global model
The world model: every program, every roster, twelve seasons of play by play, and the portal as it moves. It carries the context of the sport itself, how each team is trending and what a level of production is actually worth. A back at 5.7 a carry means nothing until you know the other seventy in his position group.
