A decade of recruiting in hindsight: what the stars actually deliver
We joined ten years of recruits to how they really performed in college. Five-stars hit 64% of the time; three-stars, 25%. The gap — and the noise around it — is the whole story.
SC
The Scout
Players & projections ·
3 min read
Recruiting coverage lives entirely in the present tense: who's rising, who flipped, who's a five-star. The one thing it almost never does is go back, years later, and check who was right. We did. Every offense-skill recruit from 2010 to 2021, joined to whether they became a real college contributor. Here's what a decade of hindsight actually shows.
64%
5-star hit rate
became contributors
25%
3-star hit rate
the bulk of every class
0.72
Model skill (AUC)
vs 0.71 stars-only
8,850
Recruits tracked
offense-skill, 2010-21
Stars are real — and noisier than the hype
Contributor rate by star tier (2010-2021)
5-star64%
4-star44%
3-star25%
2-star13%
Stars clearly predict. They also leave most outcomes uncertain.
The gradient is real and steep: a five-star is roughly five times likelier to pan out than a two-star. But look at the ceiling — even five-stars become contributors only 64% of the time, and three-stars, who make up most of every class, hit at 25%. That's why our model tops out around 0.72 AUC: the stars carry most of the signal, and the rest is genuinely hard to call. Anyone selling certainty on an individual recruit is selling noise.
Some positions are far more projectable than others
Position
Contributor rate
Recruits
Quarterback
29%
1825
Running Back
28%
2195
Receiver
22%
3505
Tight End
17%
1325
Offense-skill contributor rate by position (2010-2021).
Quarterbacks and running backs convert at the highest rates (29% and 28%) — roles where one recruit can win a job outright. Receiver and tight end are tougher (22% and 17%): deeper rooms, slower developmental curves, more ways to get buried. A four-star receiver and a four-star quarterback are not the same bet.
Which programs actually develop
Control for the ratings of who each school signs, and some programs consistently get MORE out of their recruits than the stars implied. Over the past decade Georgia State, Buffalo and Miami led the way — their offense-skill recruits became contributors well above their rating-implied rate. That's the closest thing in this data to a measurable 'development' edge.
Contributor rate above rating-implied expectation
Georgia State+15
Buffalo+14
Miami+13
Fresno State+13
Michigan+13
Duke+13
Iowa State+12
Boston College+11
Percentage points above expected (controls for recruit ratings).
Geography tilts the board too, if only slightly. Over the decade, NJ (+5.8), IL (+4.5), FL (+2.3), AZ (+2.1) over-produced their ratings, while TN (-6.4), NC (-3.5), MI (-2.0), AL (-1.8) came up short. The effects are small — a couple of points — and we found they don't reliably improve a projection for any individual recruit. But as a description of where talent has converted, the map is real.
Stars are the best single predictor we have. They're also wrong about most individual recruits. Both of those are true, and a serious recruiting take has to hold them at once.
Gridpex's desks are model-driven, AI-assisted columns. Every figure is generated from our own data and ratings — not invented. We don't fabricate reporters, quotes, or sources. Published Sun, Jun 28, 2026 · Gridpex Scout desk (recruit model).
Discussion
Weigh in on the analysis — the best takes rise to the top.
Discussion
Weigh in on the analysis — the best takes rise to the top.
0 Replies
Sign in to join the discussion.