February 2024, National Signing Day. Georgia signs its third consecutive recruiting class ranked No. 1 in the country, fifteen prospects rated four-star or above, an average rating of 92.78 by the 247Sports composite. The same day, a Mountain West program signs a fifteen-player class with an average rating of 84.1, eight of those prospects considered three-star recruits. Three years later, in November 2027, four of the Mountain West signees will have been All-Conference selections at their level; three of the Georgia signees will have started a single game; two of the Georgia signees will have entered the transfer portal by sophomore year. The recruiting class composite scores, as published every February, gave each team a single number that summarized the talent infusion. The numbers were accurate at the moment of signing. They were also, three years later, much less predictive of on-field production than the rankings themselves suggested.
Recruiting class composite ratings are, in 2026, simultaneously the most influential and the most misunderstood numbers in college football coverage. The 247Sports composite, the Rivals rating, the On3 industry ranking, ESPN’s class rankings — each produces a single annual number that ranks every FBS recruiting class against every other. The rankings are heavily covered in mainstream media, debated extensively in fan communities, and used as a proxy for program-building success on radio shows, podcasts, and Twitter. The analytical literature on the relationship between these rankings and subsequent on-field success is, at this point, mature and clear. The numbers do predict something. They predict less than the public conversation usually assumes. And the way they fail to predict is structured enough to be informative if you read the data carefully.
I have been writing about college football analytics since 2018, with a particular focus on the gap between recruiting hype and on-field reality, and the variable I find myself most often calibrating expectations around is the one this article is going to unpack. Recruiting class composite scores — what they actually measure, how they correlate with future performance, where they break, and how to read them without falling for the National Signing Day theater, is the subject of this article.
The origin: where recruiting composite scores came from
The history of recruiting rankings in college football starts with newspaper sports writers in the 1960s and 1970s, who produced informal annual rankings of incoming classes based on word-of-mouth from coaches and scouts. The modern era began with Tom Lemming, whose newsletter-and-magazine work in the 1980s and 1990s produced the first systematic public recruiting rankings. Lemming’s ratings — five-star to one-star, with composite class scores — established the framework that the digital-era successors would refine.
The internet generation arrived with Rivals.com in the late 1990s and Scout.com shortly thereafter, both producing competing recruiting databases and ratings. The two services frequently disagreed on individual prospect grades, which created consumer confusion. 247Sports, launching in 2010, addressed this by producing a “composite” score that averaged the individual ratings from multiple services. The composite became, within a few years, the dominant industry standard.
The composite rating works by combining the prospect ratings from 247Sports’ own staff, Rivals, ESPN, and On3 (when available). Each service’s rating is converted to a 247Sports-equivalent scale, weighted, and averaged. The resulting composite produces a single number per prospect (typically 0.70-1.00 on the scale) and a single weighted class score per team. The class scores have, since 2010, been the most-cited public recruiting rankings in football.
The analytical question — how well do the composite class scores actually predict next-season or career-long on-field production? — was studied seriously starting in the mid-2010s. Bill Connelly, then at Football Outsiders, published the first widely-circulated empirical work on the correlation. The results were, at the time, surprising to many fans: the correlation between recruiting class composite rank and subsequent SP+ rating was meaningful but smaller than the conventional wisdom suggested.
How recruiting composite scores work: in plain language
The basic mechanic combines two layers: individual prospect rating and class-level aggregation.
Each prospect, during their senior year of high school, is evaluated by multiple recruiting services. Each service produces a star rating (one to five stars) and a numerical grade (typically 0.70-1.00 on the 247Sports scale). The 247Sports composite weights the individual service ratings and produces a single number for each prospect. The weighting is proprietary but generally treats each service’s evaluations with comparable weight, with adjustments based on the service’s historical predictive accuracy.
The class-level aggregation, for any given team, weights the individual prospects’ composite ratings by their relative importance. Top prospects are weighted more heavily than depth-chart signees. The weighting formula has changed over the years; the current 247Sports formula gives more weight to the top players in each class to prevent teams from “padding” their class scores with low-rated players just to raise their overall ranking.
The resulting class score, for an elite program, falls in the 290-320 range on the 247Sports composite. Mid-tier Power Four classes are typically 240-280. Group of Five classes are 180-230. The very top of the recruiting rankings — Georgia, Alabama, Ohio State, Texas, in recent cycles — produces class scores in the 295-318 range with notable consistency.
The critical component: what the rankings actually predict
The single most important fact about recruiting composite scores, from the analytical literature, is that they predict program-level performance more reliably than individual-player outcomes. A team that consistently ranks in the top-10 of recruiting class composites over a 5-year window is, in expectation, a top-25 SP+ team over the same window. The correlation at the program level is about r = 0.55-0.60 over 5-year windows.
The correlation at the individual-player level is much weaker. A five-star recruit has, historically, about a 45-50% chance of becoming an NFL Draft pick. A four-star recruit has about a 20-25% chance. A three-star recruit has about an 8-10% chance. These probabilities are meaningful but, by any standard, leave substantial room for prospect-by-prospect variance.
The single best-known statistical anomaly is the three-star quarterback who becomes an elite college player. The list of players who match this description — Joe Burrow (three-star recruit, eventual #1 overall NFL Draft pick and Heisman winner), Lamar Jackson (three-star recruit, Heisman winner and NFL MVP), Brock Purdy (three-star recruit, NFL playoff starter) — is long enough to be a recurring topic in recruiting analysis. The recruiting services, on average, underestimate quarterback potential because the position depends heavily on traits that high-school competition does not stress.
Composite scores vs the alternatives: a comparison
The major recruiting rating frameworks, side by side:
| Rating system | Methodology | Strengths | Weaknesses |
|---|---|---|---|
| 247Sports composite | Weighted average of multiple services | Most stable, smooths individual-service noise | Reflects industry consensus, can miss outliers |
| 247Sports proprietary | 247Sports’ own staff ratings | Bold takes, captures individual evaluator’s read | Single evaluator’s biases, less stable |
| Rivals.com rating | Rivals’ own staff evaluations | Long history, established methodology | Independent system, can diverge from consensus |
| On3 industry rating | On3’s evaluation team plus NIL signal | Newer, incorporates market data | Less historical track record, NIL signal noisy |
| ESPN class rankings | ESPN’s recruiting team evaluations | Mainstream visibility | Less granular than dedicated services |
The composite is, in my experience, the most useful single number for analytical writing because it smooths the noise of any individual service. Cross-checking with PFF or ESPN’s coaching-staff projections can identify prospects whom the composite undervalues. The full evaluation uses multiple data points; the composite is the entry-level anchor.
What the data needs: inputs
Recruiting composite calculation requires the prospect-level ratings from each contributing service. The data is mostly held by the services themselves, with public-facing versions published on 247Sports.com, Rivals.com, and ESPN.com. The historical archives — going back to roughly 2000 for systematic data — are searchable on these sites.
For analytical work that requires bulk data extraction, the cfbfastR ecosystem includes some recruiting data, though the depth is less than the dedicated recruiting databases. Pro Football Focus publishes its own prospect grades for top recruits, which can serve as an independent check on the composite. On3 publishes individual prospect grades alongside their NIL valuations.
The harder input is the verification of “early enrollment” status and other roster timing details that affect when a recruit’s production translates to college performance. A recruit who enrolls in January (early enrollee) has a substantial head start on a player who arrives in summer. The class-level production analysis benefits from accounting for these timing variables, which the headline composite scores do not.
Building the analysis: a working framework
The practical workflow for using recruiting composites in analytical writing:
- Anchor projections to a 5-year recruiting window, not a single class. Single-class variance is large; 5-year windows are much more stable predictors.
- Compare class composite rank to current SP+ rating. A team consistently recruiting top-10 with a top-30 SP+ is underperforming; the math says they should be better. A team recruiting in the 40s with a top-20 SP+ is overperforming, almost always due to coaching.
- Look at position-specific class strength. A team that landed three five-star linemen but no four-star quarterbacks has a different developmental profile than a team that did the opposite.
- Watch the early enrollee group. Players who enroll in January are typically the most impactful immediately; they get the spring practice headstart that summer arrivals miss.
- Cross-check with the transfer portal flow. A team with strong recruiting and high outbound portal traffic is losing developed talent; this is a developmental signal that the recruiting rankings alone don’t capture.
Where this gets weird: common mistakes
The pitfalls of recruiting-composite writing.
Single-class overconfidence. A team that has one elite class is not necessarily “back.” A program that has one mediocre class is not necessarily “broken.” Recruiting variance year-over-year is real even at elite programs. Five-year windows are the analytical floor.
The “blue chip ratio” mistake. Some analytical writing reduces recruiting strength to the percentage of a team’s roster composed of blue-chip recruits (four-star or above). This metric correlates with team strength but is a derivative of the composite, not an independent measurement. Treating it as separate evidence is essentially double-counting.
Position-blindness. An elite linebacker class does not translate one-for-one to an elite quarterback class. The positional value of recruits varies dramatically — a top-50 quarterback is much more valuable than a top-50 specialist. Most public coverage flattens this, treating all recruit ratings as fungible.
Coaching-stability confounds. A team that signs an elite class and then changes head coaches in December has effectively wasted much of that class’s value. The transfer-portal era has accelerated the rate at which coaching changes destabilize previously-strong recruiting outcomes. Composite-based projections should adjust for coaching stability.
The NIL distortion. Since 2021, recruiting decisions are increasingly driven by NIL availability rather than pure school selection. A team’s recruiting composite can be inflated by NIL money in ways that don’t translate to on-field performance the same way as pre-NIL recruiting did. The relationship between class composite and SP+ outcomes is, in 2026, probably noisier than the historical track record suggests.
When composite scores shine: use cases
The applications where the framework has earned its keep:
Long-run program-building evaluation. A coaching staff that consistently lands top-15 recruiting classes is doing something replicable. Programs at Alabama, Georgia, Ohio State, Clemson, USC, and Texas in recent windows have all demonstrated repeatable top-tier recruiting. The math suggests these programs will, in expectation, dominate the next decade of college football.
Predicting program trajectory. A team that has improved its recruiting composite from a 40s ranking to a top-20 ranking over three classes is, in expectation, going to climb the SP+ rankings within two to three years. The lag is real (the new recruits need to develop) but the directional signal is reliable.
Hall of Fame and award projections. The composite rating of a player as a high school senior correlates with their eventual NFL Draft slot, college All-American selection, and Heisman voting. The relationship is loose but real. A five-star recruit who becomes an NFL first-round pick is the median outcome, not the exception.
Coaching impact identification. A coach who turns three-star recruits into All-Conference players consistently — Iowa under Kirk Ferentz, Wisconsin in the pre-realignment era, James Madison under Curt Cignetti — is doing something that the recruiting composites do not capture. Identifying these coaches before they leave for major programs has been a quiet specialty of long-form analytical writing.
A working example: the 2020 Cincinnati Bearcats
The 2020 Cincinnati Bearcats are a useful composite-scoring case study. The program, then in the American Athletic Conference, had recruited in the 50s-60s nationally for most of the preceding decade. Their 2020 recruiting class ranked 65th overall. Their 2021 class ranked 70th. By any composite-rating projection, the Bearcats were a solid Group of Five team with no realistic path to College Football Playoff contention.
The 2021 Cincinnati team went undefeated in the regular season and became the first Group of Five team selected to the four-team College Football Playoff. The roster was almost entirely composed of three-star recruits, plus the kind of overlooked four-star recruits whom the composite had pegged as marginal. The development under head coach Luke Fickell — combined with strong portal additions and a strong assistant coaching staff (which included future Texas head coach Kalen DeBoer) — produced a team that wildly outperformed its recruiting-rating projections.
The Cincinnati case study is, in one sense, a “look how the rankings can be wrong” story. It is also, more importantly, a case study in where the rankings systematically miss. Three-star recruits with elite competitive intangibles, in a strong developmental coaching environment, can produce All-American performance. The composite rates the raw athletic and physical traits more reliably than the developmental and competitive variables. A team built around player development can beat their recruiting rankings consistently; a team built around recruiting alone, without strong development, can fall short of theirs.
The limits: what composite scores cannot tell you
The honest version names the limits.
Recruiting composites cannot predict individual-player development. A five-star recruit can flame out due to injury, off-field issues, scheme misfit, or simply not improving as the level of competition rises. A three-star can become an All-American through development, coaching, and competitive drive. The rankings capture the talent baseline; they do not predict the path.
Recruiting composites cannot fully capture quarterback potential. The position depends on traits — pre-snap reading, decision-making under pressure, anticipation, accuracy under duress — that high-school competition stresses unevenly. The composite ratings are, position by position, least reliable for quarterbacks. Combining with PFF prospect grades and college-coaches’ rankings produces a more nuanced read.
Recruiting composites cannot model coaching changes. A class signed for one coaching staff and developed by another is a different class than one signed and developed under the same staff. The post-recruitment coaching context is, in 2026, increasingly volatile; the rankings cannot adjust for this in real time.
Recruiting composites cannot capture the NIL era’s market dynamics. A recruit who chooses a particular school for $1.5 million in NIL deals is making a different decision than one who chooses for “the program” or “the coach.” The composite ratings are calibrated on a pre-NIL or early-NIL dataset; the post-NIL dataset is still being collected.
One additional limit: the composite ratings themselves are subject to revision after the fact. Players sometimes get re-ranked late in the cycle based on senior-year film or all-star game performance. The “final” composite published in February can shift in the weeks before signing day. Comparing rankings across different points in the recruiting cycle requires care.
Frequently asked questions
How predictive are recruiting rankings of NFL Draft success?
Five-star recruits are drafted by the NFL at about a 75-80% rate; four-star recruits at about 45-55%; three-star recruits at about 20-25%; two-star recruits at about 8-12%. The rates have shifted slightly upward over time as the recruiting services have improved their evaluation methodologies. The composite captures meaningful talent signal; it does not perfectly forecast the draft.
Why do three-star quarterbacks become elite at higher rates than other positions?
The position depends on traits — anticipation, processing, pre-snap reading, throw-from-pocket accuracy — that high-school competition does not stress consistently. A quarterback in a high-volume passing high-school offense can look elite in stats and on film without developing the harder mental skills. Conversely, a quarterback in a run-heavy high-school offense may have minimal opportunity to display the traits that matter at the college level. The composite ratings, working off the film and stats they can see, systematically underestimate the position.
How does the transfer portal change recruiting math?
The traditional metric — class composite as a predictor of program quality — assumed that recruited players would, mostly, develop and play at their original program for three or four years. The transfer portal has shortened that assumption. A team can sign an elite class and lose three of the top prospects to the portal by their sophomore year. Modern composite-based projections need to adjust for outbound portal risk, which the historical track record was not calibrated to anticipate.
Where can I see the rankings?
The most-cited public source is 247Sports.com, which publishes the composite alongside individual prospect grades. Rivals.com, ESPN.com, and On3 all publish their own competing rankings. The composite is published throughout the recruiting cycle, with the “final” version locked in around National Signing Day in February each year.
Sources and further reading
- 247Sports recruiting rankings — the most-cited public source for recruit ratings and class composites.
- Bill Connelly’s Study Hall — the analytical writing that has consistently connected recruiting rankings to subsequent on-field performance.
- Rivals.com — independent recruiting evaluation that contributes to the composite and produces its own class rankings.
- On3 industry rankings — newer ranking service incorporating NIL signal alongside traditional evaluation.
- Pro Football Focus prospect grades — independent player-level evaluation for top recruits, useful as cross-check on the composite.
The National Signing Day moment that opened this article — top-ranked class signed, average composite rating in the 92s, the radio shows already projecting national championships — is the most theatrical moment in the college football calendar. The math, run quietly the next morning by the analytical writers, tells a more measured story. The rankings predict something meaningful. They do not predict everything. The teams that win, year after year, do so by combining recruiting with development, coaching stability, and the kind of program-level execution that no single February number can capture. For the broader frame on how recruiting fits into preseason projections, our guide to returning production is the natural companion piece.
