Wednesday, April 9, 2008
Combine stats not created equal
By Bill Barnwell
Darren McFadden's 4.33 40-yard dash at the NFL combine confirmed that the Arkansas star has speed to burn. What we don't know is what that number foretells for his professional career. Is it a sign that he'll have the speed to succeed at the next level, where everyone's faster, or does a player's 40 time have nothing to do with his future as an NFL star? Are there aspects of the combine that are overlooked or, alternately, should the whole event be ignored altogether when projecting success for Sundays?
To find out, Football Outsiders compared the NFL combine performances for all running backs who performed in Indianapolis from 1999 through 2008 to three separate metrics of success: rushes, rushing yards and DPAR, our metric which measures how good a player is over the course of the season as opposed to a waiver wire-quality back. (DPAR is explained here.) Those figures were calculated by dividing a player's total carries, for example, by the number of years since he was drafted. This method of accounting punishes someone like Ricky Williams for missing years due to injury or let's say, "injury recovery."
We compared the NFL statistics to the combine numbers for each player using correlation coefficients, which measure the relationships between two variables. A coefficient of minus 1 is a perfectly inverse relationship; when A goes up 1, for example, B goes down 1. A correlation of 1 is a perfectly direct relationship, while a correlation of zero means that the two variables have nothing in common.
Also, since not every player runs, lifts or takes tests at the combine, we used only official combine results for the players involved, ignoring the rest of the running back pool, even if those backs ran at pro days. The combine is a sort of SAT, and comparing a pro day test to it is like swapping out your SAT score for an open-book test your homeroom teacher gave you with two months' notice.
Our research revealed several key findings
The 40-yard dash matters: When compared with all three metrics, the 40 time was the single most indicative measure of a running back's future success. That's not to say that a player can succeed strictly on the basis of a fast 40, but that a 40 tends to be the biggest tip-off of future playing time and success with that playing time.
We need to adjust 40 times for weight: Raw 40-yard dash times have a minus .36 correlation with DPAR, carries and yards; remember, a negative relationship here would indicate that 40 times (which are better as they decrease) are strongly correlated with those rushing metrics (which are better as they increase).
The thing is, not all 40 times are created equal. Brandon Jacobs' 4.56 40 is incredible when you consider his 267-pound frame. On the other hand, Ahmad Bradshaw's 4.55 40 was disappointing for a player who weighed in nearly 70 pounds lighter. Adjusting 40 times for weight helps translate a raw metric into something that's more indicative of a player's NFL potential.
Ready for a math headache? The formula for the adjusted 40 score is (Weight * 200)/(40 Time4). The multipliers are as such in the formula to ensure both accuracy as well as simplicity -- the scores that result revolve around a 100-point scale. The average adjusted 40 score of all running backs is 98.5; for all drafted running backs, it's 102.4; for all running backs selected in the first round, it's 112.1. Consider adjusted 40 score to be a sort of speed score -- a higher number is better.
This metric enjoys a .45 correlation with yards, carries and DPAR. It simply does a better job of predicting future success than raw 40 times, mainly by illuminating dramatic differences in size. The first-round running backs with below-average adjusted 40 scores are busts: Chris Perry, Trung Canidate and William Green. The guys with higher 40 scores are significantly more successful. Out of Kevin Jones, Ronnie Brown, Steven Jackson, Deuce McAllister, Edgerrin James and Adrian Peterson, only Jones looks to be a bust at this point of his career.
The metric is also better at identifying valuable players later in the draft. High adjusted 40 scores would have pushed players like LaMont Jordan, Rudi Johnson, Correll Buckhalter and the aforementioned Jacobs up on draft boards, while causing teams to avoid eventual busts like Mike Cloud, J.J. Johnson and Maurice Clarett. Even late-round or undrafted contributors like Ryan Grant and Rock Cartwright show up with high adjusted 40 scores.
It's worth noting one significant player who succeeded despite a low adjusted 40 score: Brian Westbrook of the Eagles. At the combine, Westbrook ran only a 4.57 40, which is very mediocre for a 200-pound running back. Reports say that on other occasions, he ran the 40 in the 4.37 range. The difference would push his adjusted 40 score up nearly 20 points, turning him from a player to avoid to one to pursue. That's a perfect example of how adjusted 40 score is a useful metric, but not the ultimate criterion for judging young running backs.
Vertical jump is a slightly less useful predictor: While backs aren't often asked to leap players, the vertical jump is apparently the best indicator of a player's athletic ability outside of 40 times. It has a .29 correlation with DPAR, and a .28 correlation with yards and carries. Jerry Azumah (a halfback in college) holds the combine record for running backs in the vertical jump with a 43½-inch leap, while players like LaDainian Tomlinson and Buckhalter also have shown off impressive hops.
The rest of the combine data is strictly for scouts: The raw numbers for the bench press, broad jump, shuttle drill and three-cone drill bear no relationship whatsoever to the future NFL success of a running back. Scouts might see things in a player's stride and cuts that fans can't, but if somebody's quoting a three-cone drill number for a halfback, it's relatively irrelevant.
That brings us back to McFadden. Some people have wondered if he has the lower-body strength to be an NFL superstar, but whichever team sees Mr. 501 line up for it in 2008 and beyond should be very pleased. McFadden's adjusted 40 score is a superb 120, 10th best in the past decade. Complaints about his weight or body mass index (BMI) being too low can be ignored, since neither has any sort of relationship with NFL success. He might not be Peterson unless he gets to start off with the Minnesota offensive line, too, but there's no reason to expect anything but success from the young man in the years to come.
Bill Barnwell is a contributing editor at FootballOutsiders.com and co-author of the upcoming book Pro Football Prospectus 2008.