Age and WAR (position players)

Note: I’ve added a 4th graph at the end of the post, covering only the years 1982-2011.

A couple of graphs relating bWAR/age, and OPS+/age. I’ll leave the observations to you folks.

(1) As a general followup to the graphs in my Ryan Zimmerman post, here’s a graph showing the number of seasons of four different WAR levels, for all position players, for the years 1901-2011:

(2) I thought it might be interesting to look at the career patterns of those with the longest careers. The next two graphs cover the top 200 players in Plate Appearances, for the years 1901-2011; they all had at least 8,225 PAs. First, bWAR:

(3) … and then, OPS+:

(4) This last graph covers only the years 1982-2011 and the top 200 players in PAs during that time. This angle grew out of comments by kzuke, who noticed a local peak at age 31 in the original graphs (and here it is again!), and Mike L., who described a pattern of difference between WAR-Avg. and WAR/700 PAs.

I originally thought to cover just the last 20 years — the you-know-what era — but realized I’d have to take in another decade to allow for the full aging process. Even so, I limited the ages to 22-37, since the years before and after would contain less than half the original 200 players. The fewest players covered by any age in this graph is 116 for age 27 and 126 for age 22.

This graph mixes two WAR lines (red and blue, pegged to the Y-axis on the left) and one OPS+ line (green, pegged to the Y-axis on the right). I included some data labels on the OPS+ line to emphasize the distinction. Lastly, note that all WAR averages in these graphs are based only on those who actually played at the given age.

65 thoughts on “Age and WAR (position players)

  1. 1
    Dr. Doom says:

    GREAT stuff, JA. I’ve always wanted to see good graphs of this very thing. Here they are!

  2. 2
    Mike L says:

    Love the ratios between avg WAR and WAR/700 at the beginning and end of the age curve as compared to the middle ten years. Terrific visual-really great work

  3. 3
    kzuke says:

    That local peak at 31 again! We’ve seen that before haven’t we? You think this may be due to two sets of players: those who peak at 27 and those who peak at 31? It seems unlikely that the “Average” curve is that of a typical player.

    • 6
      John Autin says:

      kzuke, there was a tiny local peak at 31 on the 4-WAR line of the 3B graph I ran with the Ryan Zimmerman post, but I doubt that’s what you’re referring to. I suspect that was pure noise from a very small sample.

      I don’t know if this one is any more meaningful; the uptick at 31 is only about 3% higher than age 30. The median values (which I didn’t graph) for age 30-31 are 4.2 and 3.9, respectively.

      If I get a chance, I’ll look at the individual players from 30-31.

    • 8
      topper009 says:

      It would be interesting to see a histogram with age vs peak season from the same group

    • 12
      Voomo Zanzibar says:

      I propose that that peak at 31 is a product of the phenomenon of the Saturn Return. Here’s a tidy explanation (it’s an astrology thing):

      So, it is not really a peak, but a return to full capability after the trials that occur at 29-30.

      • 15
        John Autin says:

        It disturbs me a little that I was thinking the same thing, from a non-astrological angle.

        Do the stars portend another peak at 51? I’d like to be able to predict my own performance in 2015.

        • 17
          Voomo Zanzibar says:

          No, it’s every 28-30 years.
          So, 56-59.

          And look, I’m not really a new agey kinda guy, but Astrology, if you dig beyond the Sun Sign blurb in the daily paper, and you get a full chart analysis, well, you might see that it is like the Sabermetrics of Life.

          • 18
            Voomo Zanzibar says:

            By the way, the spell check on this website red-lines “sabermetrics.” Something wrong with that.

    • 13
      John Autin says:

      Re: local peak at 31 in the 2nd graph —
      There were 199 players who were active at both age 30 and 31. (George Sisler was out for the year at 30.) Of those 199:
      — 99 had more WAR at age 31, 94 had less, and 6 stayed the same.
      — 70 increased by at least 1.0 WAR, and 70 decreased by at least 1.0 WAR.
      — 44 increased by 2+ WAR, while 36 decreased by that amount.
      — 9 increased by 5+ WAR (led by Babe Ruth’s +9.1 after recovering fromhis famout “bellyache”, and Nap Lajoie surged +6.9 after an injury-shortened year), while 7 decreased by 5+ WAR (led by Cal Ripken’s -8.1 and Don Baylor’s -5.9).

      Still nothing really jumping out at me to explain the spike at 31.

      • 26
        bstar says:

        John, great stuff by the way. Do the guys that had the good years at 31 look in general like they are power guys? There’s a lot of differing opinions out there about when power actually peaks. Some have recently suggested it’s EARLIER than 27, but I always suspected the opposite, that it may be 28 or 29.

        • 32
          John Autin says:

          Good question. On first glance, I think there is a preponderance of big-power years among the top age-31 WAR
          seasons in my study; the top 10 in WAR averaged 33 HRs. But of course, power and WAR are correlated, so I’d have to think of a better way to test that.

          • 34
            BryanM says:

            John ; do the folks who sign 32 year olds to 250 million dollar contracts know about this stuff? Just saying..

          • 36
            John Autin says:

            BryanM @34 — From what I read, I think that all MLB front offices include folks who not only believe in the statistical approach, but are starting with more sophisticated data and doing more with it than I could dream of.

            But there does still seem to be some variation in how much weight those studies carry when the chips are down.

          • 38
            John Autin says:

            Further to BryanM — Don’t forget, these are tendencies. There are outliers.

            If you’d signed a 10-year, max-$ deal with a 32-year-old Honus Wagner, Willie Mays, Babe Ruth, Hank Aaron, Stan Musial, Tris Speaker, Ted Williams, Eddie Collins, Nap Lajoie, Barry Bonds, etc. — you’d have no regrets.

            (Well, maybe about Bonds….)

          • 44
            bstar says:

            I am assuming you are referrring to the A Pujols contract, Bryan. Doesn’t he deserve to be put among those group of outliers? We ARE talking about the best position player to play baseball since Willie Mays, aren’t we?

          • 45
            John Autin says:

            One more note on Albert’s deal: Since 1901, first basemen and DHs age 38 and up have posted just 4 seasons of 4+ bWAR:
            – 5.5, Edgar Martinez, 2001 (38)
            – 4.9, Darrel Evans, 1987 (40)
            – 4.4, Jake Daubert, 1922 (38)
            – 4.2, Joe Kuhel, 1945 (39)

            But it’s still plausible that Albert earns his pay.

          • 47
            Bells says:

            well, don’t forget that GMs and people running a baseball club own a business where the bottom line is revenue, of which money might be a part. It would be a bit facile of them to put forward a formula for WAR as the only factor in them spending on a player. Not only do you ‘overpay’ Pujols or anyone in the long run to ensure you have them in the short run, but you also ‘overpay’ them for their ‘wins contributed’ because a) it’s more efficient to have a high WAR guy than several low-WAR guys, b) they will sell millions of dollars of merchandise and put people in seats to watch games, which makes the money of the contract back.

            To paraphrase a classic sports adage, ‘winning (above replacement) isn’t everything’

          • 54
            John Autin says:

            Bells @47 — You make some good points. One quibble, though: as Ed has noted, it’s too simple to say that “it’s more efficient to have a high WAR guy than several low-WAR guys.” That’s broadly true, but each case must be analyzed individually, depending on the real-life alternatives. There are plenty of situations when a team would do better to buy two 4-WAR guys than one 8-WAR guy.

        • 48
          Bells says:

          sorry, that was supposed to say ‘revenue, of which WINNING might be a part’. Obviously money is a part.

          • 51
            bstar says:

            Exactly. There’s no way to measure truly his off-the-field worth, but I’d estimate it might be at least a third or fourth of his contract right there. And what if they, I don’t know, win 2 World Series in the next six years like the Cardinals just did? Is anyone going to be questioning his contract then?

          • 61
            Hartvig says:

            I know in one of his abstracts Bill James looked at just that issue for Nolan Ryan at the height of his popularity. He concluded that it had very little impact on the days that he pitched or, I think, on the attendance for the season on the whole.

            That being said:
            a) It’s also true that during Mark Fidrych’s great 1976 season and, to a much lesser extent, during Fernando Valenzuela’s first couple full seasons there were noticeable increases in attendance in both home and away games in Fidrych’s case. In Valenzuela’s case the difference was great for away games, mostly because the Dogers were already leading the league in attendance and there just weren’t that many more seats available. And

            b) winning also increases attendance and it seems reasonable to assume that for the first half of his contract anyways Pujols will help the Angels win more games.

      • 33
        Hank G. says:

        At the risk of coming up with an explanation for a random event, how about this?

        After a player peaks at age 27, he starts his decline phase. For a couple of years, he is in denial, but by the time he approaches 30, he recognizes that he isn’t the player he was before. If he is good enough to continue playing, he starts making adjustments for his declining skills (becoming more selective at the plate, increasing his physical training, etc.). This causes a temporary reversal in his performance, but it doesn’t last, as his physical skills continue to erode.

        • 39
          bstar says:

          Sounds like it might have some truth to it, Hank. Good theory.

        • 49
          K&J says:

          This seems very plausible, Hank.

          Is there a chance the peak coincides with impending Free Agency, or is 31 after the age that most players become FAs for the first time?

        • 55
          topper009 says:

          I can’t answer this question yet, but is it possible there is a legit psychological effect when you turn 30 and people finally admit to themselves they aren’t “young kids” anymore and work a little harder and have more discipline?

        • 58
          bstar says:

          What about walk rate contributing a little bit to this uptick at 31? Isn’t walk rate gemerally considered to be the single skill that takes the longest to mature? Maybe a lot of these guys start tinkering with a more disciplined approach at 29-30, and we finally see the results of an uptick in OBP at 31. Just a thought.

          • 63
            Hank G. says:

            Being more selective at the plate would presumably contribute to a higher walk rate.

            BTW, years ago Bill James expressed a similar explanation about how some fastball pitchers were able to adjust to the declining speed of their fastball and continue to be successful pitchers while others weren’t. I’m sure that’s the inspiration for my idea.

  4. 4
    John Autin says:

    BTW, I’ve added another graph for the top 200 in PAs. The new one shows weighted OPS+ by age.

    • 5
      Tmckelv says:

      So am I reading your graphs correctly? It says, very basically, that the peak in WAR is age 26 and the peak in OPS+ is 29?

      If that is correct, it kind of makes sense (a little) that defense drops with age earlier than hitting.

      Not to create extra work for anyone, but do you think an oWAR vs. dWAR by age graph would be helpful?

      • 9
        John Autin says:

        I’m definitely thinking of graphing the WAR components. And I welcome any other suggestions.

        There are limits to how much drudge-work I’ll do. For instance, I’d rather have done a much larger pool than 200 players, but that’s the limit of a saved search in the Play Index. I could split the target group into several saved searches, but there are already 21 separate followup searches per graph, one for each age (with attendant copy/paste into Excel); so to cover 1,000 players would take 105 followup searches, and that’s too much.

        So I’d especially welcome any ideas for improving the efficiency of this process, as well as suggestions for meaningful samples of 200 players or less.

      • 11
        John Autin says:

        Tmckelv re: factors in WAR peak at 26 / OPS+ peak at 29:

        I do think that defense starts declining before hitting does, and the same for baserunning and DP avoidance. Those things are a drag on WAR, but aren’t measured in OPS+.

        Another potential factor is that OPS+ gives equal weight to OBP and SLG (I think!), whereas OBP is actually somewhat more important. If, as I suspect, power hitting peaks later and lasts longer (but somewhat at the expense of OBP), then that would skew the OPS+ graph in favor of older hitters, as opposed to a hypothetical perfect measure of batting value.

        • 22
          BryanM says:

          And If , as I suspect, but can’t prove, (yet ) the event-based base running elements of WAR understate the decline in offensive value from age-related slowness, there is another reason to suspect the old guys are credited with a little too much ( Many players who can run a little when they first come up get to zero SB by 31 or so, but their run scoring keeps declining.)

          • 28
            bstar says:

            Good luck in your study, Bryan. You’ve already provided some cool info on the subject to date.

  5. 7
    MikeD says:

    The Weighted Average OPS+ shows what I’ve always suspected: 39 is the new 20.

    • 14
      Voomo Zanzibar says:

      If 39 is the new 20 then for my birthday next month I’ll be expecting a keg in my shower stall, a pool table in my living room, a girlfriend who can roll joints with one hand and needs sex three times a day, no responsibilities before one in the afternoon, ankles and shoulders that fully rotate, and a shortstop named Spike with a 66 ops+.

  6. 19
    MikeD says:

    I suggest that all those things are good things but the last. I will take the last on my favorite team for much of the prior.

  7. 21
    BryanM says:

    John , BTW your first chart is 4 histograms; which are sometimes shown as Bars instead of curves, but they are counting instances of x by variations in y — an amazing bit of research just terrific. i can tell you from personal experience that 67 is the new 56, but i forget what the old 56 was.

  8. 23
    Andy says:

    I love the post and the graphs–don’t let my following flip comment give any impression otherwise:

    Am I the only one who thought of Everybody Loves Raymond “Hank’N’Pat” upon seeing the title?

  9. 25
    BryanM says:

    Speaking of peaking at age 27; I did a little study on on BRef looking at scoring ability by age. Number of seasons where a player has scored at least 40% of times on base (including roe) , ( min 300 PA) sort of a combined power-speed peaking, because both sluggers and speedsters tend to do well on this metric, Anyway, it’s never been done by a 40 year old; Andre Dawson was the only one to do it at 39, 17 times by 38 year olds ( most runs ; George Brett), 41 times by 37 year olds, led by Sam Rice, Kiki Cuyler, Raul Ibanez and Jim Thome.
    The peak, you guessed it, is at 27 , 420 times, run leaders Todd Helton and Chase Uttley. The 25 year old group is a fun collection; in order of most runs scored , lu blue, Rogers Hornsby, Rafael Furcal, Albert Pujols, Zoilo Versailles, Snuffy Stirnweiss, Jimmy Foxx, Stan Musial, Elgardo Alfonzo, Jim Thome and Vince Coleman

    Forgive me if a little OT, but it seemed like fun

  10. 31
    Mike L says:

    John A., I’m looking at it a second time, and it really tells a great story of usage. The prodigy comes up and gets some playing time, looks like an impact player, and after a couple of years, starts to play full time, and the gap begins to narrow between WAR and WAR/700. As he gets more senior (mid twenties) his manager moves him up in the batting order so he gets more plate appearances per season, and the gap gets narrowest-and it stays there, even though his production begins to decline just a bit every year from 26-31. He crosses over an imperceptible line where his reputation is better than his performance-his plate appearances stay stable, but his production has taken a step down (32/33). By 34, the manager is being much smarter as to how he uses him. The plate appearances are down, the production is down, but he’s being platooned at times, rested against the tough lefthander, etc. so the decline in his WAR/700 isn’t as severe as the decline in overall war. If he’s managed to keep a job into his late thirties, he’s still producing, but only in very defined situations with a lot fewer plate appearances.

    • 35
      John Autin says:

      Nicely put, Mike L, and I do think there’s something to that. The playing time of the very young and the very old is more discretionary than that of the mid-career player — in short, the team isn’t depending on them, so if they don’t produce, their PAs are more likely to be cut back.

      I wonder if the picture would be much different if we just looked at the last 20 years — not only because of the (presumed) effect of PEDs prolonging productivity, but because of long-term big-$ (often foolhardy) contracts that make it harder (or at least more painful) to cut the playing time of an unproductive 35-year-old.

      • 41
        Mike L says:

        I’d like to see those numbers, although the 35 year old isn’t unproductive-his WAR is still 2.57, but he’s barely at 60% of where he was at his peak.

  11. 43
    BryanM says:

    @38 — Just so, John – The very best have a business value off the field as well..

  12. 52
    John Autin says:

    FYI, especially to kzuke & Mike L — I’ve added a 4th graph at the bottom of the post, covering the last 30 years only. The local peak at 31 shows up again. And it does seem (just by eye) that the downward slope in the mid-30s is a little more gentle.

    • 56
      Mike L says:

      That’s interesting. John A, the downward slope does look more gentle, but to what do you ascribe the narrower band of performance generally? Both peak WAR and peak OPS+ are distinctly lower in the last 30 years.

  13. 62
    BryanM says:

    I yield to nobody in admiration for Pujols, and obviously people who own baseball clubs are often astute businesspeople who can see commercial value off the field . The deal could certainly be a good one for the Angels, but it is odd that the best players often go from being underpaid relative to others ( Albert has never been in the top 15 the whole time he has been the best in the game) to being not very good value for money. i hope that doesn’t happen to Albert; he’s a great asset to the game

  14. 65

    You can certainly see your skills in the article
    you write. The world hopes for even more passionate writers like you who aren’t
    afraid to say how they believe. All the time go after your heart.

Leave a Reply

Your email address will not be published. Required fields are marked *