This is our old blog. It hasn't been active since 2011. Please see the link above for our current blog or click the logo above to see all of the great data and content on this site.

Win Shares and Aging

Posted by Neil Paine on July 14, 2009

...AKA part one of what I'm sure will be a very long series before we've said all we want to on the subject.

Now I'm preparing you guys ahead of time, this is mostly a data/graph dump, and there's a lot of selection bias going on here (then again, I challenge you to find an aging study where there isn't selection bias). But using a sample of all player-seasons since 1978 with >2000 MP, here are the numbers on how a player's age affects his rate of Win Shares per 3000 minutes. The first focus will be every player in the sample, and the average change in their WS3K by age:

Age #Obs Avg∆
20 10 2.02
21 27 1.68
22 62 1.22
23 129 1.17
24 216 0.47
25 247 0.34
26 266 0.19
27 258 0.19
28 244 -0.37
29 198 -0.21
30 176 -0.45
31 138 -0.32
32 109 -0.10
33 75 -0.41
34 55 -1.03
35 33 -0.87
36 16 -0.89
37 9 -0.31
38 7 -1.89
39 6 -0.43
40 2 -1.60

If you prefer that in graph form:

So that basically jibes with what we expect to see in an aging curve -- you peak at around age 27, then slowly go downhill until your mid-30s, when you fall off a cliff (unless your name is "John Stockton", apparently: the Pasty Gangsta still had an absurd 11.65 WS3K at age 40!). Still, this is the general curve for every type of player... Can we find differences by position?

Here are centers:

Just like the aggregate player pool, centers apparently peak at 27 (also, keep in mind that the sample size is ludicrously low for Cs). Here are the forwards:

Again, age-27 peak, slow decline until early-to-mid 30s, and then you start to lose it for good. Finally, the guards:

I think the guards are the most interesting, because some peak as early as age 25, some peak at the conventional 27, and some even hang around to peak as late as age 32! Among guards who played more than 1 season of at least 2000 MP, here's the distribution of peak seasons by age:

Age Pct
23 5.7%
24 11.5%
25 16.1%
26 19.3%
27 12.0%
28 9.9%
29 7.3%
30 7.3%
31 3.6%
32 5.7%
33 0.5%
34 1.0%

Here's another interesting tidbit: those who peaked early (before age 26) burned themselves out too soon -- they performed a half a win worse per season from age 30 on than their normal-aging counterparts, averaging 6.84 WS3K after the big 3-0 (as opposed to 7.33 for everyone else).

And finally, there's the question of whether guards tend to lose it quicker, and at an earlier age, than other positions. Here's a table showing what percentage of players in the study were represented at each age checkpoint:

252 Guards 323 Others
Age #Obs Pct #Obs Pct
20 2 0.8% 8 2.5%
21 4 1.6% 23 7.1%
22 21 8.3% 41 12.7%
23 53 21.0% 76 23.5%
24 103 40.9% 113 35.0%
25 112 44.4% 135 41.8%
26 116 46.0% 150 46.4%
27 108 42.9% 150 46.4%
28 100 39.7% 144 44.6%
29 91 36.1% 107 33.1%
30 83 32.9% 93 28.8%
31 63 25.0% 75 23.2%
32 52 20.6% 57 17.6%
33 29 11.5% 46 14.2%
34 15 6.0% 40 12.4%
35 12 4.8% 21 6.5%
36 4 1.6% 12 3.7%
37 2 0.8% 7 2.2%
38 2 0.8% 5 1.5%
39 3 1.2% 3 0.9%
40 1 0.4% 1 0.3%

And let's graph those percentages side by side:

As you can see, on average guards do disappear sooner than players at other positions, starting at age 32. The conventional wisdom is that smaller players who rely on their speed/quickness tend to lose that ability (and therefore their viability as NBA players) in their early 30s; bigger players, of course, lose it at that point too, but it doesn't matter as much for them because they were never very fast to begin with. Based on these results, I'd say the conventional wisdom has been confirmed -- guards do appear to "lose it" sooner than big men, starting in their early thirties. That makes the recent re-signings of aging PGs Jason Kidd and Mike Bibby particularly risky, and the buyer should certainly beware when it comes to Allen Iverson, who could end up being the poster child for this phenomenon.

6 Responses to “Win Shares and Aging”

  1. Jason J Says:

    It would be interesting to see how much that has changed over time. With the better training methods and diet, the improved shoes and equipment, and the easier travel, it seems like players have been able to hold it together better in their 30s since the 1990s. I'm thinking of Jordan, Stockton, Karl, Reggie, Porter, Barkley, H. Grant, Antonio and Dale Davis, Detlef... A lot of guys played pretty impressive ball into their mid-30s. It's possible that you might get a different result if you set your perameters to include only players drafted after 1980 (or something like that). Of course losing Kareem would set back the centers quite a bit!

  2. MCT Says:

    Some numbers that tie into Jason's point, but from a different angle:

    The 1983-84 season was Kareem Abdul-Jabbar's 15th season in the NBA. At the time, he was one of only seven men to have played 15 seasons in the NBA/ABA. Today, there are 98 players who have done this.

    1984-85 was Kareem's 16th NBA season. At the time, he was one of only four or five men to have played 16 seasons in the NBA/ABA, depending on how you define "NBA" (the others were Dolph Schayes, John Havlicek, Paul Silas and Elvin Hayes; Schayes only played 14 seasons in the NBA per se, but the NBA gave him credit for the season he played in the NBL before the NBL-BAA merger). Today, there are 57 players who have done this (58 counting Schayes).

    In 1985-86, Kareem played his 17th season in the NBA. He was the first man ever to play 17 seasons in the NBA/ABA. Today, 34 players have done this.

    I won't keep going year by year -- I'm sure you get the idea -- but Kareem was also the first man to play 18, 19 and 20 years. Today he is one of 17, 8 and 4 men, respectively, to have reached those stages of longevity (three players have one-upped Kareem by playing 21 years: Moses Malone, Robert Parish, and Kevin Willis). Some of this may be attributable to players entering the NBA at younger ages, but not all of it; Moses Malone aside, the trend began with players who entered the pros at a time when almost everyone played at least three years of college basketball, if not four.

    Note: in all cases above, years played totals reflect the number of seasons in which a player appeared in an NBA game. Kevin Willis was on an NBA roster in 22 different seasons, but he missed the entire 1988-89 season due to injury, so he only played in 21. I also know of at least two more pre-1984 players with a 15-year spread between their first and last NBA seasons -- Carl Braun and Dick Barnett -- but neither actually played 15 seasons in the NBA. Braun only played in the NBA/BAA for 13 seasons due to spending two years in the military service. Barnett is only credited for playing in the NBA for 14 seasons because he jumped for a year to the ABL, a short-lived early '60s attempt at challenging the NBA which is not generally recognized today as a major league.

  3. Johnny Twisto Says:

    Fascinating MCT, I had no idea how unusual 15+ seasons was at the time nor how common it has become.

  4. Chicago76 Says:

    Another possible explanation of the "guards losing it" is the potential pool of competition due to normal height distribution in the general population. There is a relatively large pool of =6'5" talent trying to fill the majority of an NBA team's roster at F-C.

    Once you make it as a regular in the league at F-C, the higher availability of roster spots and lower competition via height distribution means it is easier to hang on to those spots.

    For a guard, there were always be more competition via college PGs, SGs, and undersized forwards transitioning to shooting guard. It follows that turnover and decline would happen earlier at the guard position.

  5. Adjustable Dumbbells Says:

    I really enjoy this website. Could let me know how I can subscribing with it? By the way I stumbled upon this blog through Bing.

  6. Roxana Behn Says:

    Review by A.J. for Apex JD2.2 Flat Bench
    This bench was easy to assemble, and, once put together, the bench was solid, i.e., it didn't wobble. The padding is good and the overall quality is good. The only downside I have seen from the bench so far is that it could use a wider base for doing exercises like dumbell flies... This could be considered a good thing, though: you have to use your core and bring the weights up evenly. This could also be due to that fact that the bench sits on thick carpeting. For $40 and free shipping, I have no complaints.