A recruit has not played a college snap. So his rating is a projection, and the band is the point.
Recruiting is the purest bet the engine makes. A high-school prospect is run on the College engine in its high-school configuration, a projection onto the college scale and never a locked present-tense number, on the thinnest data the engine works with. It trusts the tools and the translatable athletic traits most and the production least, it produces the widest bands the engine makes, and it treats the recruiting-service stars as an outside input to reconcile against, never as its own answer. Because the confidence is the whole product here, a projection without its band is a lie, and where a recruit commits re-prices the bet.
Project onto the college scale, trust the tools, and band it wide.
A composite prospect who has not played a college snap. He runs on the College engine in its high-school configuration, and the output is a projection onto the college scale, never a locked present-tense rating.
The projection never produces a locked OVERALL on its own, and the moment real college production lands the evidence resolves the prior and the projection is retired (replaced, not averaged). A recruiting rating and a college rating are different objects even at the same number. The honest form of the projection is its band.
Illustrative on the real HS configuration (the College engine in high-school config, the projection onto the college scale, the tools-trust extreme, the widest-bands rule, the projection-as-prior). Composite prospect, demonstration figures.
Reason from the data to the trait, never backward from a ranking.
A composite prospect the recruiting services have already ranked. The engine does not read the ranking and reason backward from it. It reasons from the data to the trait, and then it reconciles against the board.
The divergence is where a recruiting operation makes its money, and the engine surfaces the agreement score and the divergence findings rather than laundering the star rating into its number. The overlay can raise a trait's confidence, but it can never move the projection. The stars tell you what the market thinks, the projection what the data says, and the space between is the edge.
Illustrative on the real Scout Overlay mechanics (competitor ratings as inputs never the output, the no-pedigree rule, the reason-from-data-to-trait discipline, the divergence-as-finding). Composite prospect, demonstration figures.
Frame upside makes the widest bets, and the program bends them.
A composite recruit resting on Frame Upside, the projection the engine is least certain about and most interested in. Frame Upside is set for a young prospect with room to add mass or physical scores below the bar with room to grow.
The developability tiers do the sorting the eye cannot, a board that knows which is which allocates its offers and money accordingly, and none of it is a promise, all of it is a distribution. The engine makes the odds. The staff makes the offer.
Illustrative on the real projection-shaping layer (the Frame Upside physical projection, the developability sort, the program re-pricing). Composite prospects and programs, demonstration figures.
A recruit has not played a snap on the scale you are rating him on, so his number is not a measurement of what he is, it is a projection of what he might become, built on the thinnest data the engine ever works with. That is not a weakness to hide behind a confident number, it is the truth the engine reports: it runs the prospect on the college engine in its high-school configuration, trusts the tools and the translatable athletic traits over production that came against uneven competition, and hands back the widest band it produces, because a projection this far on data this thin is meaningless without it. It refuses to let the recruiting stars become its answer, treating them as an outside input to reconcile against and reasoning from the data to the trait rather than backward from a ranking, so that the space between its projection and the consensus board is an edge rather than an echo. It bands the frame-upside prospect as the honest long shot he is, and it re-prices every projection by the program the recruit commits to, because the same player is a different bet in a different place. Report the projection, report the band that is most of the truth in it, and let the staff make the wager the engine will not.
Project honestly, band it wide. Let the staff make the offer.
Recruiting Intelligence projects a high-school prospect onto the college scale as a banded bet, treats the recruiting stars as an input and not an answer, sorts the boom from the safe by developability, and re-prices the projection by the program, never handing back a false-precision number.