Do NFL Combine Numbers Predict Success?

The question that never settles is whether the numbers say anything about how a prospect will play in the league, or only about where he will be picked. These are separate questions, and correlation analysis is the tool that keeps them apart.

Two outcomes, two correlations

A combine metric can be tested against two different targets: draft position and later NFL production. A drill that lines up with the draft slot tells you how scouts weigh it. A drill that lines up with production tells you whether that weighting was justified. It is interesting that the two often disagree, which is why a single “does the combine matter” verdict is usually the wrong framing.

The 40-yard dash is the best example. A 2021 study of more than 1,000 combine invitees across three draft classes found a clear link between 40 times and where players were picked at several positions: wide receivers (rs = .436), tight ends (.356), offensive linemen (.373), linebackers (.573), and safeties (.510), where rs is the so-called Spearman rank correlation coefficient. Therefore, as shown, faster players tended to be drafted earlier.

One detail is worth internalizing before reading any table like this. Draft position is coded by pick number, so the smaller is better. That flips the expected sign: the same work reports a negative correlation between vertical jump and draft slot for receivers (about −.286), which means a higher jump goes with a lower — that is, better — pick.

The draft slot is not production

Predicting where a player is drafted is not the same as predicting how he plays, and this is where the combine's reputation gets complicated. A metric can shape draft position while barely moving with on-field value. The 40 is the standard example: it weighs heavily on where skill-position players come off the board, yet the link between raw testing and career production is consistently weaker than the link to draft slot. Short-area burst and explosiveness — the 10-yard split, the vertical — tend to carry more of whatever signal exists.

A useful rule of thumb from applied statistics: a coefficient below about 0.3 is weak, while one closer to 0.6 is where a relationship starts to look dependable. Most single drills, measured against production, sit at the low end. The practical reading is that speed is priced more heavily into draft capital than into production supports.

Moreover, the Relative Athletic Score, created by Kent Lee Platte in 2013, compresses a player's testing results into a single 0-to-10 figure, graded against the historical record for his position. Against a general measure of career value, composite athleticism shows a moderate positive correlation — stronger and more consistent than any lone drill, and clearest among defensive positions. The relationship is not uniform along the scale, though: the middle band separates players poorly, and prospects with good tape routinely beat their scores.

Run it yourself

An interesting feature is that none of these analyses require proprietary data, since combine results, draft slots and career-value figures are all public. We can, for instance, correlate two columns to get Pearson’s r for any pairing you care about — a position group’s 40 times against career value, for instance — and square the result to see the share of variance one variable explains in the other. For draft order specifically, prefer the Spearman rank version, since pick numbers are ranks rather than measured quantities.

The same public inputs feed the mock exercises. If you would rather watch a fast riser reshape a board than read a coefficient, set up a draft room, run it with other people, and see where the testing actually pushes each prospect in practice.

A few cautions keep the exercise honest. Remember that correlation is not causation. Range restriction quietly shrinks the numbers because the combined data mostly covers invited and drafted players rather than the full talent pool. Small-position samples cause coefficients to jump from year to year. And a single r assumes a straight-line relationship, so it can hide threshold effects — a level of speed a cornerback has to clear, past which more adds little.

The takeaway

Combine numbers explain draft behaviour better than they explain football. That is not a reason to ignore them, but a reason to read each coefficient in context: check the sign, the sample, and above all, which outcome is being predicted. A prospect’s 40 time will tell you a fair amount about his draft weekend and considerably less about his third season.

The group that will test next spring is on college fields right now, and this season’s front-runners — with Indiana setting an early pace — are where those measurements, and the arguments about what they mean, will come from.