What Does the NFL Draft Really Produce? (Part I)
by Scott Kacsmar
This is the week when analysts rush to grade the NFL draft that just happened before any of the players have even played a down. We are going to do something very different here today by looking only at drafts where every player has retired from the NFL, ensuring complete career data. I had been sitting on this idea for four years, but this offseason was finally the time to complete it.
With the recent retirements of Peyton Manning, Charles Woodson, and Matt Hasselbeck, every player drafted before 2000 is now retired. In 1994, the NFL made important changes by introducing a salary cap and shrinking the draft to seven rounds like it is today. Even a draft as recent as 1992 had 12 rounds, if you can believe it.
By studying the 1994-1999 drafts, which saw a total of 1,459 players drafted, we have the first six examples of what a seven-round NFL draft really produces. Today, with the advent of "Draft Twitter," there are more supporters and detractors of draft prospects than ever before. Late-round picks that would have garnered little interest in the past have a new glimmer of superstar hope because "that guy I follow had him as a second-round pick for two months." Sifting through the takes can be more troublesome than warranted now. While so many current draft grades are favorable and fan expectations are sky-high for most prospects, we know that within a few years many of these players will turn out to be average at best. There just cannot be that many studs in one draft. We know several first-round picks will bust, and a few late-round gems will surface. Getting two starters and two contributors is a great haul in theory, but we will have to see what the data says below.
Judging player value is never easy, especially for non-skill positions, and in this timeframe we are lacking snap data, which FO only has back to 2002 right now. A mid-round pick that does not start games, but plays on special teams and is a multi-year backup, can be valuable to a team. Only 22 players on the 53-man roster can start in a given week, but the benefits of a good backup are hard to quantify. How do we know the player is good if he never has to see the field? A team keeping a player around seems like a good argument for his worth, but even Will Blackwell lasted five seasons in Pittsburgh. Victor Allotey, a seventh-round guard drafted in 1998, was the study's most unique example, lasting six seasons (three each with Buffalo and Kansas City) despite never playing a regular-season game. Some fans on Reddit even believed him to be a homeless ex-player. Needless to say, Allotey never left much of an impact despite his lengthy career.
So while we are heavily using games played and games started here, there are obvious limitations to the usefulness of that data. Years of experience were also collected, though if a player was cut or traded during the season, then that still counted as a full season with that team. This also includes Allotey-esque seasons where the player may not have played a single game. For the most part, years of experience were coordinated with the data at NFL.com, except for obvious cases of difference such as Walter Jones not playing in 2009 or a data error with Bernard Williams (banned after failing 15 drug tests) making a comeback in Philadelphia that never happened.
Weighted career Approximate Value (AV) from Pro-Football-Reference was also collected for this study. Like any stat, it is imperfect, but it makes a grand attempt to quantify every player-season at each position, and the results are more than respectable. The top players in AV in this study are Peyton Manning (177), Ray Lewis (160), Derrick Brooks (144), Marshall Faulk (133), Marvin Harrison (124), Randy Moss (123) and Warren Sapp (120).
We will start with a broad view of the draft results before chopping things down into smaller bits. Supplemental draft picks were not included. Player-for-player trades were not considered either, though I am not sure there were any Manning/Rivers-level deals in any of these years anyway.
Draft Results: By Career
It is often said that the average NFL career lasts three to four years. The NFL Players Association has quoted an average of 3.5 years, though some question their methods. It is reasonable to think careers are getting shorter today than they were in the period of this study, but my results are much closer to the NFL's proposed figure of 6.0 years from 2011. Naturally, adding in the numerous undrafted players that try to make a team each summer would bring down the average, but can you really call that a career if you never make the final cut? An average career length of five-plus years looks very accurate for drafted players.
|NFL Draft: Average Career Length|
|Draft||Players||Years of XP||Games||Starts||AP1||Pct.||PB||Pct.||CarAV|
In terms of average games played, the number is a tick under 65, or the equivalent of just over four seasons. That is more in line with the number I was used to hearing for average career length, but that is from a games played perspective only.
The columns for first-team All-Pro (AP1) selections and Pro Bowlers include how many players earned those honors and are not the total number of honors each draft accumulated. The highest individual honor one can receive, a bust in the Pro Football Hall of Fame, is something that could take decades to figure out a percentage for from these classes. Currently, just nine of the 1,459 players are inducted (0.62 percent), but look for Peyton Manning, Charles Woodson, Ray Lewis, Randy Moss, Terrell Owens, Alan Faneca, Brian Dawkins, Champ Bailey, Tony Gonzalez and perhaps a few more to at least double that amount. Ultimately, each draft class from 1994-1999 should have two to five inductees.
Regardless of current Canton busts, the 1996 draft is the cream of the crop from this era, with the highest average of starts and AV, and most individual honors (116 Pro Bowls and 43 All-Pros). It had the rare occurrence of a wide receiver going No. 1 overall (Keyshawn Johnson to the Jets), leading arguably the best wide receiver class ever with Terrell Owens, Marvin Harrison, Terry Glenn, Muhsin Muhammad, Eric Moulds, Joe Horn, Eddie Kennison, Bobby Engram, and Amani Toomer all surpassing 500 receptions and 7,500 receiving yards. (In a few years, 1996 vs. 2014 wide receivers should be a great debate.)
The 1995 and 1999 drafts were statistically not as impressive, though 1995 did see a few key injuries derail incredible careers, including Terrell Davis. Tony Boselli was another player on the fast track to Canton, and he really started the amazing run of offensive tackles that happened in this era with Jonathan Ogden, Orlando Pace and Walter Jones following in his footsteps. That run of talent likely led to the idea that tackles are "safe" picks at the top of the draft, and nearly-certain 10-year players. Then the likes of Robert Gallery and Jason Smith came around, along with more recent disappointments such as Matt Kalil, Eric Fisher and Luke Joeckel. But let's keep living in the '90s here.
The concept of a 10-year player has always been questionable when football is such a tough sport with major health risks, but roughly one in five players drafted do last that long.
|NFL Draft: Career Length Splits|
|Draft||Players||10+ Yrs||Pct.||0 GP||Pct.||0 XP||Pct.||0 GS||Pct.|
Then you have the unfortunate 20 or so players drafted each year who never even get credit for one season (0 XP) because they were cut before the regular season. I am not sure this happens as much these days, with smaller draft classes and practice squads willing to gobble up many of those players. Anyone drafted today also has a Wikipedia page with details of when they were cut and signed to practice squads, whereas we are a bit in the dark with past flame-outs from the mid-'90s.
Just fewer than 14 percent of the drafted players never played in a regular-season game. We'll soon break that down by round; as you can imagine this happens much more often in the last rounds. Roughly a third of players never start a game, which would be fine if they were kickers or punters, but only 17 of the 1,459 players were drafted for their legs. One of those 17 was the only player in this study to finish with negative career AV: kicker Steve McLaughlin, on whom the Rams foolishly wasted a third-round pick in 1995, their first year in St. Louis. McLaughlin only played in eight games and was just 8-of-16 on field goals. Well, at least they didn't trade up to get him in the second round.
Draft Results: By Round
You have probably read dozens of draft studies that reach the same obvious conclusion: the first round produces the best careers, the second and third rounds are not bad, but after the 100th pick or so, it is a real crapshoot to find any significant value. Things were no different from 1994 to 1999.
|NFL Draft: By Round Averages|
|Round||Players||Years of XP||Games||Starts||AP1||Pct.||PB||Pct.||CarAV|
The average first-round pick lasts nine years and starts over 100 games, which makes the "10-year starter" concept much more feasible when you are talking about a high pick. More than one-fifth of first-rounders end up as All-Pro selections, and more than a third make at least one Pro Bowl, and this was back when they didn't invite fifth alternates like today. By the time you get to the seventh round, you are lucky if that player starts a dozen games. Honestly, just making a roster is impressive. Here is the breakdown of career length from the previous section, but by round.
|NFL Draft: By Round Splits|
|Round||Players||10+ Yrs||Pct.||0 GP||Pct.||0 XP||Pct.||0 GS||Pct.|
More than a quarter of sixth-round picks and nearly a third of seventh-round picks never played a regular-season game. Every first-round pick played at least one game, but five did fail to log a start in their career: Andre Johnson (1996 Redskins), Craig Powell (1995 Browns), Reggie McGrew (1999 49ers), Dimitrius Underwood (1999 Vikings) and John Avery (1998 Dolphins). All five were selected between picks 24 and 30, but Johnson may have been the worst of them all, never even playing a game with the Redskins and only logging three appearances in his career. Underwood also never played for his drafted team and was the most memorable name of the group to me, as he had a rather interesting story.
[ad placeholder 3]
Can you recall any of the four second-round picks to never play a game? Leon Bender (1998 Raiders) is the most tragic story of this study. Just five weeks after he was drafted 31st overall, the Washington State defensive tackle passed away at the age of 22 from a seizure disorder. James Manley (1996 Vikings), Jimmy Oliver (1995 Chargers), and Shane Hannah (1995 Cowboys) were the other three without a game played.
None of the six drafts produced more than four Pro Bowlers in any of what we call today the "Day 3" rounds (Rounds 4 to 7). Some of the steals found in Rounds 4 to 7 include Terrell Davis, Zach Thomas, Jamal Anderson, Rodney Harrison, Matt Birk, Matt Hasselbeck, Derrick Mason, La'Roi Glover, Tom Nalen, Aaron Smith, Joe Horn, Al Harris and Donald Driver. Not all of them shined for the team that drafted them, but we will get to that later.
The round with the highest average AV was 1996's first round (53.0). The "deepest" later round in this study was the 1996 draft's fifth round with an average AV of 18.7. It produced six 100-plus-game starters in Zach Thomas, La'Roi Glover, Joe Horn, Fred Miller, Chris Villarrial, and Marcus Coleman. The most barren round also belonged to 1996, as its seventh round averaged a study-low 8.5 starts and 4.1 AV. That makes sense given how much talent went in the first six rounds of 1996.
The year-to-year results were fairly consistent by round, with one notable exception coming in the seventh round. Since free agency was new, the concept of compensatory picks would not come into play until the 1995 draft. So there were only 28 picks in 1994's seventh round, but then 39 to 52 picks in the 1995-1999 drafts. That may be the reason why 1994 has the best seventh-round metrics with 4.8 years of experience, 49.1 games, 18.8 starts, the only two All-Pro selections, three of the seven Pro Bowlers, and 11.1 AV.
Draft Results: By Position
Are offensive tackles the safest picks? Unfortunately, there was too much gray area in figuring out the positions, so we took a more general approach by using offensive line, defensive line, and defensive back instead of breaking things down into tackle, center, safety, defensive end, etc. There still is some gray area between defensive line and linebackers, but we generally went with what the official NFL site had, unless it was a clear error like Tom Barndt being listed as a guard when he was a defensive tackle at Pittsburgh and with the Chiefs.
|NFL Draft: By Position Averages|
|Position||Players||Years of XP||Games||Starts||AP1||Pct.||PB||Pct.||CarAV|
No position is drafted more than defensive back, with the two trenches not far behind. Punters and kickers have the longest careers, which was expected, but the offensive line is not far behind at 6.2 seasons. Wide receivers have the shortest careers at 5.0 years, which also makes sense given how dependent they are on their quarterback and situation.
[ad placeholder 4]
However, the high-priced quarterbacks tend to have the shortest leash, averaging a study-low 43.7 games played. Remember the names of busts such as Jim Druckenmiller, Akili Smith, Cade McNown, Ryan Leaf, and Heath Shuler? None were able to start 25 games in their careers as teams were rightfully quick to dump them. Backup quarterback careers can last long with the player rarely making an appearance, but ideally you want to draft a quarterback to be your starter. Twenty-three of the 63 quarterbacks never started a game, but those were all players drafted 100th or later.
Offensive linemen average 50.5 starts, while no other position is above 40. That sounds impressive, but linemen are used to starting the game with the expectation of going the distance. Most other positions can be rotated in, but that is where the line has an advantage in these numbers. It always feels weird to talk stats at all with offensive linemen, but in terms of AV, they lead there as well as the only position to crack 20 on average. Of course, playing with a great quarterback can really help the offensive production, which is key in accumulating AV.
By these numbers, it looks really hard to draft a good tight end, though it seems like this is just a reflection of the time period more than anything. Tony Gonzalez, arguably the greatest tight end ever, was drafted in this time in 1997, but the rest of the bunch is a very unimpressive group led by Ken Dilger, Freddie Jones, Desmond Clark and Kyle Brady. Gonzalez, and later Antonio Gates, really helped push the athletic, pass-catching tight end into becoming a major part of offenses again. From 1994 to 1999, only five tight ends were drafted in the first round. That doubled to 10 in 2000 to 2005, which does not even include the prolific Gates (undrafted) and Jason Witten (third round in 2003). We have now seen another lull with two first-round tight ends since 2011, and if you played fantasy football last year, you know it felt like Rob Gronkowski or the field in most weeks.
(Ed. Note: There's also an issue where P-F-R's AV formula seems to underrate tight ends by scoring them the same way wide receivers are scored. -- Aaron Schatz)
Obviously the round matters in finding positional value, and none of the main positions do better than they do in the first round. Setting the kickers aside, this graph shows the progression of average AV for each of the eight positions by round.
Each position follows a pretty similar decay, with the offensive linemen notably spiking above their peers in the fourth and seventh rounds. Nothing is lower than a fifth-round quarterback -- something I showed in a previous study of drafted quarterbacks -- but the reason they are at zero AV here is because only Kevin Daft and Jay Barker were drafted in this period.
There are really two main types of draft analysis: what the player does, and what the drafting team gets out of the player. So far we have just covered the former, but Part II of this article, next week, will focus on the latter.