Wide Receivers and Combine Numbers

In the second half of this pre-draft analysis, I’ll write about the “Combine Numbers” associated with wide receivers. As I noted yesterday, the bell cow of the Combine, the 40-yard dash, is random noise when it comes to predicting whether a receiver will be a success in the NFL.

Since some NFL athletes can run almost as fast as Olympic track stars, the score that relates most to track – the 40-yard dash – receives significant attention. However, the receiver with the fastest Combine 40 time in the period I studied, Jacoby Ford (4.22 seconds) in 2010, was selected in the fourth round by Oakland and played only four years in the NFL, accumulating 848 receiving yards. He showed some game-breaking return skills, but fumbled the ball too frequently. This is a decent, normal amount of production for a fourth-round wide receiver.

Darrius Heyward-Bey came in second at 4.25 seconds in 2009. He has stuck around for nine seasons. He accumulated 2,071 yards in his first four years, well under what the Raiders (again) would have hoped for from the seventh overall pick, and another 817 yards since. While he’s signed for just above the veteran minimum, I would not be surprised if he retires this off-season.

Up third is Jerome Mathis (4.26), drafted in the fourth round by Houston in 2005. Like Ford, he made an impact as a kick returner as a rookie, but had an issue with fumbling. He had 80 total receiving yards in a three-year career.

You get the picture. Some speedsters, like Mike Wallace, Julio Jones and Brandin Cooks, are very productive in the league. But just being able to run really, really fast is fairly meaningless on its own.

The other tests include a “bench press,” in which an athlete repeatedly lifts a 225-pound barbell until he feels he must stop. There’s a vertical leap from a standing position, a broad jump from a standing position, a 20-yard “shuttle run,” which measures the ability to accelerate and make turns quickly, a 60-yard version of the shuttle drill and a “three-cone” drill, which measures many of the same agility skills.

Players are also measured, with their height, hand size and arm length recorded within an eighth of an inch. They are also given an intelligence test, subjected to any of a number of medical examinations and (sometimes controversially these days) available for interviews from coaches whose goals are to surprise athletes with unusual questions.

Experts can make a lot of money by taking all this information and coming up with a formula for success. Unfortunately, no one has done that. Since statistical software these days is quite sophisticated, data miners can come up with a formula that fits a large number of past performers. But is that predictive? Even with my sample size of 444 drafted wide receivers, I suspect it’s limited.

A wide receiver needs many skills, and anyone who saw Jerry Rice play could tell you that his speed, while hardly close to the best on the field, was enough to get the job done. Rice reportedly ran a 4.59 in the 40-yard dash. He was drafted 16th overall in 1985. Larry Fitzgerald, a certain Hall-of-Famer, ran a 4.48 – same as A.J. Green, who is headed to Canton himself if he can maintain his production level a few more years.

What do they have in common that other receivers don’t? Fitzgerald was the third overall pick and Green was the fourth in their respective years. Scouts knew about them. Fitzgerald didn’t take part in most of the Combine events, so scouts likely didn’t have that type of comparative information down to the hundredths of seconds. Green did, and while his numbers were good, you can find better. What makes Green one of the best in his generation and Heyward-Bay, who had slightly better, but similar Combine numbers, disappointing?

I gave the 272 drafted players who ran most of the drills in the study a similarity score to Green’s. Strangely enough, one of the lowest scores came from Chad Johnson, who had an excellent career mostly with Cincinnati as well. But I look at his cone-drill time of 7.51 and his reputation as a bit of an eccentric (well, aren’t all good receivers a bit eccentric?) and I wonder if he simply didn’t feel like running the drill. That’s the worst time in that drill among wide receivers, and more indicative of a defensive lineman or tight end.

But some receivers had almost identical scores to Green’s, and some, like Michael Jenkins (2004) and Sammy Watkins (2014) went in the first round. And some, like Robert Herron (2014) and Pierre Garcon (2008) went in the sixth. Herron barely played and Garcon is one of the better 6th-round picks in recent years. Watkins was average for a number-four overall receiver pick and Jenkins was a bit of a disappointment for a late first-round pick.

There’s no magic in Green’s combination of results. In fact, the 100 receivers least similar to Green were a little bit better than the 100 receivers closest to Green. And the slightly smaller group of receivers between those groups averaged a tiny bit better than both groups. As for draft position, the top 100 averaged being picked 128th, the bottom 100 135th and the middle 71 126th.

Am I saying the Combine times don’t matter? Not even enough to be worth tracking? No. There are slight correlations between times and performance. Vertical leap seems to be the most important number, though it only correlates 15% with output. A receiver’s weight actually matters slightly more (18%).

The shuttle drill and 3-cone drill are difficult to use, because the top scouted players are far more likely to skip these drills. Why risk draft position on a poor performance (unless you’re Chad Johnson)?

I suspect that weight matters because it takes a certain amount of strength to deal with good bump-and-run coverage. And some of the low-round or undrafted receivers who have done wonders (going back to Bill Belichick’s penchant for finding Wes Welkers and Danny Amendolas) work out of the slot much of the time. If you want to find an exceptional athlete who was a little smaller but found enormous success out of the slot early in his career, Randall Cobb was picked by Green Bay at the end of the second round in 2011. More recently, the Chiefs have gotten a lot out of tiny Tyreek Hill, a fifth-rounder with elite speed, in many different roles.

That’s a lot to digest. What about a measure of simple competency? Take the 272 receivers with measurements in every category and look at those who scored in the top half in all of them. So that would mean someone who weighed more than 200 pounds, ran a 4.48 or better in the 40, a 4.20 or better in the shuttle, a 6.92 or better in the cone drill, leaped at least 36 inches and had a broad jump of at least 121 inches.

Who are these extraordinary people? The list is just 12 of the 272. Surely these are all players the scouts would agree are superstars in the making. Not so fast. There were two first-rounders (Heyward-Bey and Javon Walker, who ended up being slightly disappointing his first four years), four second-rounders – two who greatly exceeded expectations (Alshon Jeffery and Torrey Smith) and two who busted (Tyrone Calico and Chad Jackson). There was a fourth-round pick, Keenan Burton in 2008, who didn’t fare very well. In the fifth round, Roy Hall in 2007 had nine career receiving yards. The pair of sixth-rounders – Garcon and Jeff Webb (2006) were very solid picks. And the two seventh rounders, Marcus Maxwell in 2005 and Jeff Janis in 2014, were forgettable.

More recently, Kevin White (the seventh overall pick in 2015), fits this category and is looking like one of the bigger busts in recent history. DeVante Parker, also a first-rounder in 2015, has been OK. The only player to fit the bill last year was second-rounder Zay Jones, who hasn’t done much. And, notably, undrafted Tyrell Williams (2015) has been quite a find for the Chargers.

At any rate, the simple every-category competency measure is a poor fit for the data. Probably because smaller players are often more agile. Indeed, weight correlates 18% negatively with the shuttle drill and 23% negatively with the 40-yard dash. The broad jump is the only drill where weight seems to help a little.

So I made one last try – finding players who are in the top third by weight and top third in vertical leap. That’s 209 pounds and 37 inches. That gave me a set of 51 drafted receivers from 2001-14. What was their average performance? As a group, they outperformed their draft position by about 5%. Of the 15 first-rounders, eight significantly exceeded expectations and only three were very disappointing. This group also contains some breakout lower-round performers like Brandon Marshall, Josh Morgan, David Givens and Marques Colston. More recently, it identifies second-rounder Devin Funchess (2015), though there hasn’t been a first-rounder meeting the criteria the last three years.

I’m content to stop there. Being invited to the Combine means scouts think a player has a potential NFL career. That implies a minimum competency. This 209/37 test looks decent for identifying additional potential. And I think data mining the rest is only going to identify stupid coincidences in the remaining data set.

Who meets the criteria in the upcoming draft? D.J. Moore, Tre’Quan Smith, Dylan Cantrell, Allen Lazard, Jester Weah, Justin Watson and Reggie Bonnafon, among those with draftable grades. Of those players, Cantrell has unusually good scores in other categories, indicating solid athletic skills. His 4.59 40-yard dash, however, might concern teams and despite high vertical leap and broad jump scores, scouts say he is not an explosive athlete. He has a third-day draft grade. I’d be interested to see if he sneaks into the second round, though.

Just to wrap this up… if you play Front Office Football, you know the Combine plays a much different role in the game. Individual measurable skills correspond to specific potential ratings. This is obviously not realistic. But the alternative is to place more emphasis on the scouting abilities on your staff – with wider variety between teams. This would essentially render the Combine scores within FOF useless.

I think that makes for a less enjoyable game. You’d be drafting purely on scouted ratings and focusing on where your expectations differ from league-wide grades. At that point, it’s just guesswork with no skill at all. Or maybe I’d add something to indicate where your staff member has a particular insight. But if you played a hunch on your own with a Combine number, you’d be making a mistake.

I may change this whole system in the future, but for now, I think it’s a good idea to stick with a simulated Combine that means a lot more than the real-life version. Watching tape and making scouting opinions is full-time work for the few who really do this well. Just as messing with the intricacies of player contracts and agents is full-time work for a GM. That doesn’t make for an enjoyable game. My philosophy is that the game engine itself should be as realistic as I can make it. But the staff experience – your job as the team’s hybrid GM/coach/owner/etc – has to be enjoyable.

Wide Receivers and the NFL Draft

The NFL Draft begins Thursday, and I’ve been taking some time the last few days to study how the draft works. In particular, I’m looking at wide receivers and Combine scores.

This first post, however, will focus on performance alone.

The Amateur Draft is an important piece of team building. Perhaps it’s the most important piece, because teams receive on average eight new draft picks every year and only 53 players are on the team during the season. Trades are still very rare in the NFL, especially involving young players.

When you draft a player, he’s yours for four years. After that, he’s more-or-less a free agent and gets paid based on a combination of experience, expertise and the role he’s expected to play. There are very few important bargains in free agency. You fill holes in free agency; you don’t build a team. And when signing contract extensions after that fourth year, you have to pay the market rate.

To analyze the effectiveness of a draft pick under a hard salary cap, therefore, you should limit your analysis to those first four years. In other sports you can factor in the market and the style of a particular owner. Even the NBA has its “Bird exception” and its “Rose rule” and other odds and ends to turn the cap into something a little different. But in the NFL, the draft gives you that young base that shapes your team.

I chose wide receivers to examine because there’s a simple statistic that defines them quite effectively – at least when applied to hundreds of players over a long period… receiving yards. While you can dig deeper with receivers – adding points for first downs earned in passing situations and for touchdowns – simple accumulated receiving yards are quite meaningful.

The scope of this study is 2001-2014. That’s the last 14 years of receivers who have reached four years of experience. 2001 is a good place to begin a study of the most recent passing emphasis within the game. It was the last year before the eight-division, 32-team format. There were 444 wide receivers drafted during that period – more-or-less evenly dispersed in the seven rounds of the draft with a slight increase in the seventh round.

For each player, I tried to find Combine data (and where that wasn’t available, data from the respective Pro Day). I also tracked total receiving yardage through year four. Eighty-five of these players never caught an NFL pass. The highest of these 0-yard picks (and only second-rounder) was Dexter Jackson, 58th overall in 2008, a shorter (5094) small-college star who ran a 4.37 40-yard dash. Tampa Bay tried him in a few games as a return specialist as a rookie, then waived him late in training camp the following year.

Sixteen of the 444 wide receivers exceeded 4,000 receiving yards in their first four years – all the way up to A.J. Green of Cincinnati with 4,874. Eight of those 16 were first-rounders (five in the top eight picks, including Green), four seconds, two thirds, a fourth (Denver’s Brandon Marshall) and a seventh (Marques Colston of New Orleans). Both Marshall and Colston were drafted in 2006.

In fact, 27 of the 43 receivers who have exceeded 3,000 receiving yards were first-round picks. Part of this is self-fulfilling – you draft a wideout in the first round and you had better feature him in the offense right from the start. But scouting receivers isn’t as hard as scouting quarterbacks and the learning curve isn’t quite as steep, so coaches have a decent idea of who will and who won’t thrive in the NFL.

Average receiving yards by round drafted: 1st – 2,708, 2nd – 1,725, 3rd – 1,139, 4th – 813, 5th – 510, 6th – 340, 7th – 293. I broke down the first round, and while the average tails off noticeably in the last few picks of the round, the sample sizes get too small to provide significant delineation.

I also took a look at how individual teams draft. I did this through assigning a score to every pick based on the expected four-year yardage totals for each round. Every team drafted between 8 and 19 wide receivers during the 2001-14 study period (average of about 14).

The “best” team at drafting wide receivers was Indianapolis by a wide margin. Part of that is likely that the Colts could probably get away with lower receiver skill sets because Peyton Manning was so remarkable. Part was having amazing longevity in the form of Marvin Harrison (before the study) and Reggie Wayne. So Indianapolis was also one of two teams with eight receivers drafted (San Diego – average in quality of picks – was also at eight). Philadelphia, Chicago, Arizona, Jacksonville, Cincinnati, Pittsburgh and Dallas made up the rest of the top quartile. Minnesota, Tennessee, Tampa Bay, Seattle, St. Louis, Baltimore, Kansas City and (dead last) New England made up the bottom quartile.

Why New England? Perhaps Bill Belichick just isn’t great at evaluating college receivers. He’s a master at seeing low-round (or even undrafted) guys on other rosters who can help, but when he reaches for draftees, it’s not pretty. His highest receiver picks at New England were Chad Jackson (36th overall, second round) in 2006 and Bethel Johnson (45th overall, second round) in 2003. Combined, they didn’t reach 1,000 yards in their entire careers. New England is also the only team in the NFL that has not drafted a wide receiver in the first round in this millennium (Minnesota and Detroit are tied for first at four). You’d have to go back to 1994 and Michigan’s Derrick Alexander to find a Belichick-led team and a first-round wide receiver.

It’s not easy finding effective wide receivers, though. There is an element of lottery to it. Just look at 2008. In addition to Dexter Jackson, nine other wideouts were chosen in the second round. Four others were below 500 yards in their first four years. Four were above average for the round, including Donnie Avery, Eddie Royal, Jordy Nelson and DeSean Jackson. Of the group, only DeSean Jackson (4,085) was above 2,600 yards in four years. No receivers were taken in the first round. But second to Jackson in four-year yardage from that entire 2008 class was Davone Bess, who came out of Hawaii, ran a 4.60 40-yard dash, stood 5097 in height and went undrafted. Scouts seemed uncertain whether Hawaii’s incredible passing game in those days worked because of quarterback Colt Brennan (a sixth-round pick in 2008 who never played an NFL game) or Bess.

To wrap up this first post, I’ll talk about Combine numbers. I’ve tried to accumulate as much information as possible about the Combine tests for wide receivers. However, not all prospects run all of the drills and some information comes from Pro Days and some from the NFL Combine itself. I don’t want to throw out information, but a Pro Day might have more favorable (or less favorable) conditions. In addition, not running a test is hardly evenly distributed. Expected top picks are more likely to skip the tests, not wanting to jeopardize their standing with a poor result. And there’s less information about players unlikely to be drafted.

I have a 40-yard time on 441 of the 444 drafted players, ranging from 4.22 (Jacoby Ford, 2010, a fourth-rounder with average fourth-round production through four years) to 4.78 (Jonathan Smith, a seventh-rounder who accumulated 77 career receiving yards). There’s vertical leap data for 372 players, broad jump for 334, a shuttle drill for 297 and a three-cone drill for 288. In addition, I have height for every player and hand size and arm length for most.

I will start out by saying that 40-yard dash time hardly correlates at all with four-year receiving yard totals, despite a positive 16% correlation with draft position. That’s what we’re up against with any study of Combine numbers. More soon.