More on Changing Hitter Aging Curves

A few days ago, I looked at the possibility of major league hitters no longer showing any hitting improvement, on average, once they debut in the majors. I believe both the banning of PEDs and teams being able to evaluate MLB ready talent are the keys to this change.

To start off, I am going to look at a few of the question raised from the first article. The first question I would like to answer is “Was the aging curve change a one year fluke?”

To begin with, here is the wRC+ (link) curve going back six, ten year intervals.

Generally hitters have aged the same over the years with the exception of the past two 10-year intervals. From 1993 to 2002, players reached their peak a little slower and took a little longer to decline compared to the previous intervals. The other and more important change involves the most recent 10-year interval. Batters start at their peak and then decline at a faster rate. I expected the drop to be more than the 1993 to 2002 interval, but it was more than any other time frame. I wonder if pre-1993 amphetamine usage helped hitters perform better than previously thought.

The second question I would like to answer, “Is the aging rate change a one year occurrence or has it been happening for a few seasons?”

In the original graph from the first article, I looked at the 2006 to 2013 and 1998 to 2005 time frames. There was quite a difference in wRC+ between the two. I went back and got the values for each 7-year interval between and including the original curves.


The hitter aging peak became less and less defined the closer the time range gets to the present. The aging curve has slowly been changing over the past seven seasons.

The third and fourth questions I will answer are, “Are there less young players in the majors right now?” and “As the number of young hitters have decline are the ones playing at their offensive peak so they won’t improve?”

I went back and looked at the number of harmonically matched plate appearances for each of the 10 previous intervals I examine. Here is the percentage of matched pair plate appearances for six age ranges.

The number of matched plate appearances has dropped since the 1963 to 1972 interval until the last 10-year cycle. The graph is a little hard to read the exact values, so here is the percentages of matched player seasons from age 25 and younger (called young hitters from now on).

Time Frame 25-years-old or younger
1952 to 1962 17.7%
1963 to 1972 23.2%
1973 to 1982 20.1%
1983 to 1992 16.7%
1993 to 2002 13.1%
2003 to 2012 14.9%

The change makes sense. During the 70’s, players were granted free agency and teams were more likely to keep their young talent in the minors until they were major league ready. The percentage of young players continued to decline until the last interval when it increased a bit. So for 40 years, the number of young hitters declined by over 10% points (23.2% to 13.1%), but the aging curves stayed the same. From 1993 to 2002, the least amount of young hitters played, but the aging was the steepest compared to the other seasons.

Finally, I wanted to answer, “Do non-MLB hitters have their peak at 26-years-old?” I looked at the wRC+ aging curve for Triple-A batters from 2006 to present (extent of FanGraphs minor league database) and here is the aging curve.

Players in triple-A still have an up and down curve with a peak at age 25.

In conclusion, I think two items are causing the change in the curves. The first in the ban and harsh penalties on PEDs between the 2005 and 2006 seasons. Besides the ban on muscle building drugs, the crack down on amphetamines may have caused players to no longer chemically peak. The second effect is teams being able to better know when a player is able to contribute at their peak MLB level. I think this ability to evaluate talent may have existed before the PED ban, but increase production from PED masked the change. In the near future, I will look to see which stats are causing the change. In the meantime, let me know if you have any more questions.

Print This Post

Jeff writes for FanGraphs, The Hardball Times and Royals Review, as well as his own website, Baseball Heat Maps with his brother Darrell. In tandem with Bill Petti, he won the 2013 SABR Analytics Research Award for Contemporary Analysis. Follow him on Twitter @jeffwzimmerman.

27 Responses to “More on Changing Hitter Aging Curves”

You can follow any responses to this entry through the RSS 2.0 feed.
  1. Iron says:

    Out of curiosity, I sorted for qualifying age 30 hitters. 16 hitters came up. I plotted their WRC+ and WAR from the time they came into the majors until 2013. Almost all of them either kept improving or peaked in their late 20’s.

    Of the 16, 3 had their best WRC+ year at 30, 6 at 29, 2 at 28, 1 at 27, 2 at 26.

    So, my methodology is skewed towards productive 30 year old who batted well enough at 30 to be qualifying. But, at the same time, 14 of the 16 as individuals followed the traditional aging pattern one would expect. It leads me to question whether Jeff’s results are from individual players no longer aging the same way, or from some other selection biasing the data, such as teams washing young players earlier, or… something. It seems to me there is benefit to track individuals rather than age groups to see how actual players performance varies by age.

    Vote -1 Vote +1

  2. Josh says:

    “The change makes since”.

    Since when?!?!

    Vote -1 Vote +1

  3. cass says:

    Can we make Mantle, Mays, and Aaron take a walk of shame and shall we have endless debates about whether there should be an asterisk next to their names? How about MLB commissions a report on illegal drug use in the 1950s, 1960s, and 1970s? Maybe we need some Congressional hearings.

    The hypocrisy of amphetamines vs. steroids/HGH really bothers me. These were illegal drugs. The players knew they were wrong. They had to get them through shady sources because the team couldn’t provide them. It’s all documented in the widely-read book Ball Four. And yet no one cares.

    Vote -1 Vote +1

    • Mike Green says:

      Curt, the Mick seems a bit lifeless as he ambles down the Walk of Shame. The pick-me-ups don’t seem to be working any more.

      Vote -1 Vote +1

  4. olethros says:

    To test the influence of amphetamines, maybe graph league wRC+ by month for several sets of 5 year intervals or so, similar to the above. I bet the most recent two sets show a much bigger drop over time than previous years.

    Vote -1 Vote +1

  5. here goes nothing says:

    “As the number of young hitters have decline are the ones playing at their offensive peak so they won’t improve?”

    I suspect this sentence could be rewritten, because I don’t know what it means.

    Vote -1 Vote +1

    • Andy says:

      As the number of young hitters have declined are the ones who are playing now at their offensive peak, so they won’t improve?

      Vote -1 Vote +1

  6. Some Scientist says:

    It appears from your figures that last decade wRC+ declined as expected based the history. Basically, during the last decade the average 20-24 year old MLB player is producing more than at any time in history. It does appear that aberrant performance was observed from 1993-2003 in 35-37 year old hitters. What is interesting is that rather than declining any differently from history, players in the 2006-13 time-bin appear to reach a higher level earlier and then decline at the same rate as history would suggest. You could ask; was the average 22 year old MLB player of the most recent decade performing similarly to the average 24 year old in previous decades? The implication may be that since MLB players reach peak performance earlier they decline earlier but at the same rate. Shift the last decade’s curve 1-3 years to the right on your X axis and see what it looks like.

    In terms of decreased absolute performance by 28-32 year olds, what are the wRC+ for each time-bin? If you published your tables one could assess the significance of the peak ages, absolute performance in each time-bin, as well as the post-peak decline slopes instead of just eye-balling it. Based on your numbers, the percentage of players under the age of 25 is significantly lower in the last two decades than during the prior three decades. Your suggestion that teams are better at identifying the most productive years of younger players supports the argument that, as a group, players under 25 are performing at a higher level than at any time in history simply because lesser players (which would lower the mean wRC+ of the group) remain in lower levels of an organization for longer.

    Do Latin American players begin intensive training at an earlier age than U.S. players? Are there more or less athletes who spent time in college programs in the game? If so has this altered the age-related trends in your sample?

    Vote -1 Vote +1

  7. Nate G says:

    Help me understand the Triple-A aging curve. Wouldn’t that be very misleading because the best of the older hitters would be inherently be in MLB?

    Vote -1 Vote +1

    • bsball says:

      This seems like an important point. Players that continue to improve in AAA leave the group and move up to MLB. There is nothing similar going on in the MLB sample. Players only leave the data set when they are no longer able to play at that level.

      Vote -1 Vote +1

  8. tz says:

    Jeff, how has the wRC+ changed over the decades for the 25-and-younger players?

    It’s been in this past decade or so that teams have more regularly (1) held players in the minors for longer to prolong the team control period and defer arbitration eligibility and (2) drafted more “lower-ceiling” college hitters in lieu of “high-ceiling” high school hitters (like St Louis).

    Vote -1 Vote +1

  9. PackBob says:

    I wonder if there are any distinct groupings within the larger sets. For example, is there a subset that ages the same as in the past that is masked by a larger subset that now declines without first improving. Is it possible that a newer subset of rapidly declining players skews the overall curve?

    It might also be interesting to sort according to length of time and ages played, or to find the most common playing time frame. For example, if the most common time/age frame were 6 years from age 23-29, how does that that compare to the full set.

    Vote -1 Vote +1

  10. Devon says:

    Since wRC+ is relative to league average, a decline in older hitters would result in a rise in younger hitters, correct? If younger hitters are doing more of their learning in youth baseball academies (they seem to be more prevalent) rather than in the minors/majors, the flattening of the early curve makes some sense as well.

    Vote -1 Vote +1

  11. triple_r says:

    Why is the first graph in black-and-white?

    Vote -1 Vote +1

    • Micah Stupak says:

      Why is the second graph in the green version of black-and-…well, gray?

      Thought-provoking work, Jeff, but a real pain to read these two.

      Vote -1 Vote +1

  12. slash12 says:

    Could the results be skewed because players that are brought up incredibly early are exceptional players to begin with, and the pool of data gets polluted as the older less talented players are brought up later in their career? I wonder what this would look like if you only followed individuals that happened to come up at a young age, throughout their career.

    Vote -1 Vote +1

  13. badenjr says:

    Could it be that players are actually younger these days? I mean, MLB has cracked down on false birth dates the last few years. Could it be that when players were hitting their peak at age 27 before, they were really just 24 years old?

    Vote -1 Vote +1

    • olethros says:

      Don’t they usually fake their age the other way? I don’t recall anyone insinuating that Pujols was younger than his stated age, but the opposite accusation comes up in the comments on just about every article about him.

      Vote -1 Vote +1

  14. Nathaniel Dawson says:

    Is it possible that the overall talent pool of Major League players is getting better the last decade? We see evidence of this with pitchers — there’s no shortage of people that have noticed and remarked on the increase in pitch velocities the last few years. I have no idea how you’d measure change in absolute talent level for position players.

    Vote -1 Vote +1

  15. Bill James says:

    Yes, but these correlations are too selective to interpret the way you are presenting them. What is the r-value of these age/performance correlations you suggest? Therein lies the answer, my friend.

    On your next try, please consider using dashed-lines and/or more starkly contrasted colors in the graphs, especially when comparing more than 3 curves. And throw some plotted symbols in there, like a triangles, squares, etc., to better illustrate where the behavior of these curves exhibit separation from each other.

    Vote -1 Vote +1

  16. derekcarstairs says:

    Isn’t it more important to show how “good” ballplayers age rather than all ballplayers?

    Lichtman presented a chart showing that ballplayers since 1980 with 5,000 PAs tend to hold up very well until age 36. The finding is substantially different if all who have played in the majors are considered.

    Vote -1 Vote +1

  17. iron says:

    That’s a bit where I was going with my initial comment. Looking at individual players rather than a sample of age-groups showed players who came into the league over the last 8 years and are now 30 have aged exactly as previous studies would suggest. This implies the immediate aging curve decline in Jeff’s work is due to sampling problems and not representative of when current young players are actually peaking. Young players are promoted and demoted from mlb for many reasons that have nothing to do with age-related decline. I believe Jeff’s claim that something has changed recently. But I don’t necessarily buy his amphetamine conclusion. Individual players are still peaking in their late 20s as they had before. Teams are just selecting players differently in a way that skews the age sample.

    Vote -1 Vote +1

  18. Aristotle says helllo, and would like me to point out one of his classic logical fallacies–coincidence doesn’t equal causation. If the data is compelling and supports the lack of a peak, but assuming that it is PEDs seems pretty much like a wild leap because two things happened at the same time. Simiarly with talent evaluation. There is nothing to substantiate the link between those two things and the lack of a peak other than timing (or if there is more, its not in any of these articles). Additionally you ignored another, equally widespread and equally (if not more) important phenomenon–the increased performance in pitching. Pitchers have been dominating for a while now, in part because of new usage patterns in the bullpen, in part because an industry wide focus and understanding of pitch framing, and in part because of a better understanding of pitch counts and the times through the line up effect. All of these things could explain the lack of a peak in hitters as well. Finally, I think the fact that pitchers have gotten better while hitters got worse is at least a small argument against PEDs. If the lack of peak was caused by PEDs and hitters have gotten worse (pitchers never really had a peak, so that hasn’t changed), why hasn’t there been a similar effect on pitchers? Great article, really important articles, but the second part of the argument regarding causation is flimsy.

    Vote -1 Vote +1

  19. Baseball Bob says:

    I think this is very important research, if true. A real graph of aging, though, would probably incorporate the AAA and MLB data together: if you used MLE of some flavor, we could see players’ performance relative to age, without the twin masks: the AAA table is wrong because the best players go to MLB, and the MLB table doesn’t reflect the growth and peak that takes place in AAA. If THAT graph shows the growth/decline pattern we expect, then the conclusion that players are being kept in the minors until their peak would be reasonable.

    A related question not being asked is this: is this the right thing for MLB? If players are being brought up at peak, might there not be a pre-peak point at which they are MLB-ready, even though they have some growth left?

    Vote -1 Vote +1

  20. dominik says:

    I wonder if that has to do with the amount of time the kids now play. in earlier decades kids would play basketball or hockey in the winter and not really touch a baseball.

    now kids are getting paid lesson all year long and play paid “travel ball”. bryce harper played like 100 games a year in various tournaments and showcases as a 12 year old. some parents from cold regions even drive their kids to tournaments into warm regions just that they can play games in the winter too.

    this probably accelerates developement (when they basically play a pro schedule from age 14 on) but could also lead to earlier wear and burnout (in fact many 16 yo kids now have TJ surgery).

    Vote -1 Vote +1