Investigate / Football / Prediction
What Recruiting Rank Predicts in the NFL
A top high-school football recruit is much more likely to reach the NFL. Once a player gets there, recruiting rank has a smaller relationship with how much he plays.
Result
Seventy-one percent of recruits ranked 1–32 recorded an NFL regular-season snap inside five years of their class. The rate was 22% for ranks 201–300. Among NFL entrants, better rank still tracked early playing opportunity, but the relationship was modest.
- 70.9%
- of recruits ranked 1–32 entered the NFL inside the fixed window.
- 21.9%
- of recruits ranked 201–300 did the same.
- 707
- entrants in the four-season playing-opportunity analysis.
The first threshold
The top 32 were three times as likely to reach the NFL.
“Entered the NFL” means at least one regular-season snap from one through five years after the recruiting class. A draft pick without a snap does not count. An undrafted player with a snap does.
The entry rate falls at every step down the list. It moves from 70.9% in the top 32 to 40.2% for ranks 33–100, 29.3% for ranks 101–200, and 21.9% for ranks 201–300. The top-32 advantage appeared in all seven classes.
Entry intervals: top 32, 64.6%–76.4%; ranks 201–300, 19.0%–25.1%. A sustained role means at least 500 offensive or defensive snaps in two of a player’s first four NFL seasons.
Drafted is not the same as played
The outcome stays tied to a regular-season snap.
- Drafted in window
- 586
- Entered in window
- 707
- Undrafted entrants
- 167
- Drafted, no snap in window
- 46
After entry
Rank had a modest relationship with early playing time.
The next question only uses the 707 recruits who entered on time. Their opportunity window covers four consecutive seasons beginning with NFL debut. A missed season remains a zero.
Spearman correlation measures whether two orderings move together. Lower recruiting numbers are better, so a negative estimate means better-ranked entrants tended to play more. The correlation with primary offensive or defensive snaps was -0.159, with a 95% interval from -0.232 to -0.078. That is a real association, but not a strong one.
Points are Spearman rho estimates. Lines are 95% intervals from resampling recruiting classes and recruits within each class. Offense and defense rows include zeros for players on the other side of the ball, which dilutes those separate estimates.
Snaps describe opportunity and availability. They do not describe how well a player performed, and a special-teams snap is not interchangeable with an offensive or defensive snap. This is why the analysis stops short of an NFL “value” measure.
Exact rank and broad groups
The information continues between the usual labels.
Four broad groups tell the main story, but they discard some useful ordering. In held-out recruiting classes, a smooth model of exact log rank had an area under the curve (AUC) of 0.670. The four-group model scored 0.640.
AUC asks how often the model ranks an entrant above a non-entrant. A value of 0.5 is chance ordering and 1.0 is perfect. Log loss grades the predicted probabilities and penalizes confident mistakes, so lower is better. Exact rank also won that comparison, 0.587 to 0.596.
Fitted probabilities from the confirmed log-rank entry model. These are population estimates, not claims that adjacent rank slots identify different players precisely.
Held-out class performance
Exact ordering beat the four fixed groups.
- Smooth log-rank AUC
- 0.670
- Smooth log-rank log loss
- 0.587
- Four broad groups AUC
- 0.640
- Four broad groups log loss
- 0.596
Consensus check
The Composite and 247Sports’ own list were effectively tied.
On the same 1,525 recruits, the Composite recorded 0.649 AUC and 0.626 log loss. The 247Sports in-house list recorded 0.647 and 0.627. Every paired uncertainty interval included zero, so this sample does not show an NFL-entry advantage for the consensus.
Position groups
The rank gradient appears across the field.
Top-100 recruits entered at a higher rate than ranks 101–300 in every broad position group. The size of the gap varies, and so does the base rate.
Those differences are descriptive. Position groups occupy different amounts of the recruiting list, NFL teams need different numbers at each position, and players often change positions in college. The chart cannot tell us that one position is easier to evaluate.
Right column shows ranks 101–300 first, then the top 100. “Skill” combines running back, receiver, and tight end. “Athlete / unknown” preserves recruits without one fixed position label.
Sample and scope
Limits
This study covers the 247Sports Composite top 300, not every high-school recruit or college player. It is a predictive analysis. College development, injury, scheme, transfers, and opportunity all sit between recruiting rank and an NFL snap.
Measurement details
The primary exposure is the 247Sports Composite national rank preserved on each historical class page. The study keeps ranks 1–300 from 2011 through 2017 and retains ties. Two duplicate source identities were resolved explicitly. The identity crosswalk linked 945 recruits to NFL records, confirmed 1,153 as non-NFL identities, and left no unresolved cases after 14 manual decisions.
NFL entry requires at least one regular-season snap from class+1 through class+5. Early opportunity covers four consecutive regular seasons beginning with debut for on-time entrants. Draft selection, entry, total snaps, primary offensive or defensive snaps, and special-teams snaps remain separate fields.
Predictive comparisons leave one recruiting class out at a time. Uncertainty uses 1,000 resamples of classes and recruits within class. The current 247Sports historical pages do not display their final-snapshot dates. Internet Archive captures and contemporaneous final-ranking coverage establish that the class lists existed before NFL outcomes, but this is not a row-by-row reconstruction of every signing-day page.
Sources
Recruiting ranks: 247SportsComposite Recruit Rankings historical class pages. Provenance checks included anarchived 2015 class page and contemporaneous final-ranking coverage fromMaxPreps. NFL player, draft, and snap-count records come fromnflverse-data, licensed CC BY 4.0. Raw recruiting pages and player-level identity links are not republished.