≡ Menu

Some teams, like the Rams have done a good job of fielding a very young roster; others, like the Raiders, have made a conscious effort to head in the other direction. Overall, the Rams are more representative of the current trend. NFL teams have made a shift towards younger players in the last three years, although you might be surprised by just how dramatic and sudden the change has been. The drop in Approximate Value (AV)-weighted ages of NFL rosters in the last three years is more than 50% larger than in any other three-year period in NFL history.

healy 1

Looking at the graph, there are two seismic shifts that changed the age distribution of the NFL in the Super Bowl era: the increase that started in the late ‘80s and the decrease in the last five years. These changes tell us about how changes in the collective bargaining agreement can change the NFL landscape in both subtle and dramatic ways.

First, the increase in NFL roster age in the 1980s coincides pretty closely with the introduction of Plan B free agency in 1989. It looks like the increase maybe starts a year too early. Remember, though, that the 1987 age may be skewed a bit by the three games with replacement players. Taking that point in mind, the increase from 1988 through 1993 coincides exactly with the introduction of limited free agency.

Given how little freedom Plan B actually gave to players, it may be surprising that Plan B had much impact on NFL roster age. However, Plan B did more than just make it possible for players such as Everson Walls to switch teams; it made their careers longer.

healy 1.5

The graph above―which shows the average career length for players who played their last season that year and played at least three years―shows an increase in career length about five years after free agency started.1 We’d expect free agency’s impact to take a little while to show up in this graph, just as it does.

Free agency thus appears to have perhaps extended average career lengths by about half a year. Before free agency, players like Walls would have retired without the ability to find the team that needed their skills. Plan B allowed teams to protect 37 players, giving freedom of movement to everyone else. The players who benefited from that system were exactly those whose original team no longer saw them as essential, who could more easily find a new team. Other explanations for the change are possible, such as better conditioning or nutrition, but the move to longer careers lines up pretty closely with the free agency explanation that I think that’s probably it, particularly in light of what the data seem to say about the current CBA.

In contrast to the career-lengthening impact of free agency, the current CBA seems is likely to lead to shorter careers. Players on cheap rookie contracts can fulfill roles that used to be carried out by veterans, causing those players to retire. This trend is suggested by the move towards younger rosters in the first figure. While it’s too early to see this trend in the ages of players who have retired, we can look instead at the chance that a player has left the league.

healy 2

This graph shows the chance that a player who is at least 33 years old and provided at least 5 points of AV does not play the following year. Notice the big increase in 2012. Since 1980, good players in their mid-30s were never more likely to retire than they were in 2012. We’ll see if the trend continues, but the overall pattern of teams valuing older players less fits with the other data. So I expect it’s likely to keep going this way, the Oakland Raiders defense notwithstanding.2

  1. The graph bounces around in the earlier years and even a fair bit more recently, so I took a moving average of the year and the one immediately before and after. Due to the moving average, one thing that doesn’t show up is that the longest average career length since 1960 happened in 2007. []
  2. I spent some time trying to predict how likely it is that the Raiders defense will have an AV-adjusted age of over 30 this year, and I think it’s possible if Khalil Mack starts off more slowly than I would anticipate. That would make them the first defense that could run for the Senate since the 2006 Dolphins. []
  • I’ve been doing research into this but it seems you are further along than I am so I hope you don’t mind me asking a question. Do you have the data for average age of roster for playoff and Super Bowl teams to see if there is a trend?

  • I don’t have it set up now to do playoff teams, but I do have it set up to do Super Bowl winners. Once you correct for the number of teams in the league, there’s no clear change in the age of SB champs over time. Correcting to a 32-team league, these are the numbers for the age rank for the champ (1=youngest, 32=oldest). I’ll put the average age in ().
    SB I-X: 19.3 (27.3 yrs)
    SB XI-XX: 19.3 (27.0)
    SB XXI-XXX: 17.9 (27.1)
    SB XXXI-XL: 26.6 (28.4)
    SB XLI-XLVIII: 19.2 (27.5)

    The average age is always a little on the old side, which makes sense (and makes teams like the ’92 Cowboys and ’13 Seahawks stand out). Teams may have gotten a little older after free agency with that bump in the fourth group, but maybe just a blip.

  • I clearly agree with point 1 but Im not sure about point two, at least the contract aspect of it. The rookie contract modification in the CBA only really affected a portion of the first round. The rookies drafted from round two through seven saw their contracts completely unchanged. Most of the first round was in a similar position as picks used to earn 5 year max length contracts which is exactly where they are now and at a similar price point.

    I think the shifts have come from a few spots in the CBA. There is a smoothing out of the franchise tag which makes that a bigger part of negotiations than in the past. The move from salary cap to cash minimums has changed some of the ways teams are doing business and having to plan for the long term. The financial equation for the salary cap was changed and thats the biggest one.

    The initial big drop came in 2011, which had I think two factors. With the financial equation changing the cap dropped big time. Teams in 2008/2009 would have been planning for cap raises at around $6 million per year which meant plans were for a cap in 2011 of $137-139 million. Instead it was $120. To comply with that veterans were going to get the ax. The other thing I would mention is that at the end of 2010 a number of teams essentially had a salary cap free roster purge that eliminated a ton of veterans who normally would have been players that the team was stuck with moving forward, which in turn would have led to more player friendly veteran contracts. Those were eliminated.

    The QB salary explosions also began around that time and may have led teams to consider planning for much money being invested in one player. Eli Manning jumped the market by over $3 million or so a year when he signed for $16.25 million in 2009. Rivers then hit $15. By 2011 you had Brady around $18 and Manning at $19 not to mention Vick at $16. Prior to the Eli deal teams were around $10 mil a season for the biggest QBs. Teams saw where this was headed (at the moment it looks like its peaked at $22M) and I would think prepared accordingly with changes to veteran spending to compensate.

    • Chase Stuart

      I’ll defer to you on cap stuff, but aren’t there two other factors in the new CBA that would shift the league towards younger players: 4-year contracts for all draft picks (used to be 3 for later round picks), and the vet minimum floor was raised.

      • 3 years was the minimum length, but in practice almost every team was already using 4 year deals, which was the max allowable length. The only exceptions were the Steelers, Ravens, Cardinals, Chiefs, and I think Lions who still went with the 3 year contracts. Those players rights were still controlled by the restricted free agent process however. By the vet minimum floor do you mean minimum salaries? Those were always moving upward though I cant remember the differece in the 06 vs 11 CBAs. The minimum salary benefit has always remained though (player signed to 1 year deal at min salary plus very modest bonus counts at same cap cost as third year player) to counteract the higher vet salaries.

        • Those are all great points. Assuming that the shift to younger players lasts, your explanation is much more complete than the one I wrote about. One other thing I thought of didn’t have to do with the CBA, but maybe a move towards analytics in front offices. For example, maybe teams are understanding aging curves for RBs better. But RBs haven’t gotten any younger (all the move to younger players is on defense), and the shift matches up too well with the new CBA. So everything you guys have been talking about sounds like close to the full explanation.

          • I think thats also a good point. I had written or talked about something on that topic once before but never really took a hard look at it to see if my theory had any validity. When you go back to the 90s-early 00s, you get a good deal of the dual role coach/GM. These guys almost always put more weight on veteran players to win now rather than a slower developmental process. Parcells was probably the worst with that. Those roles are more split now and many of the GMs are coming from a more business oriented background, where stats and analytics are a normal business practice.

        • Chase Stuart

          Thanks – yes, I was referring to the vet minimum salaries. I thought that, as part of the negotiations and the “give” on rookie salaries, the vet floor was raised quite a bit. I could be mistaken.

          • I got curious so I looked that up. If the old CBA has just been extended the 2014 salary scale would have been 385K for a rookie and 920K for a 10 year vet. Its 420K and 955K under the new CBA, so the raise was even across the board to keep the disparity equal between the two.

            • Chase Stuart

              Thanks, Jason. I guess my memory is off.

  • Nick Bradley

    This is all the more interesting with so many older QBs tearing up the league.

    I bet if you excluded quarterbacks and K/P this would look different

    • Yes, I thought about this, too. Just checked it out and the recent drop is slightly bigger if you leave out the kickers and QBs. I actually expected a little bit bigger difference.
      Including kickers and QBs: 1988 average age = 26.91; 2013 average age = 26.99
      Excluding kickers and QBs: 1988 average age = 26.82; 2013 average age = 26.84

  • Richie

    Why did you choose to use AV-adjusted age, instead of just average age? AV-adjusted would tell us the age of players who are getting the majority of the playing time. But average age would tell us the age of players who are taking up cap space and roster spots.

    Although, I would guess that maybe bench warmers are mostly young players anyway.

    • Yes, I wanted to get at the average age of the contributors, but I guess it could have been interesting to check out the unweighted average, too. Fewer of those end-of-the-roster spots may have gone to vets after the new CBA. I just checked it out and the picture looks pretty similar to the one for AV-weighted age in terms of the trend (with unweighted age about 1/2 year younger than AV-weighted age now).

      One differences: Drop in ’87 due to the strike substantially bigger for unweighted age.

      • Richie

        Cool, thanks.