Rethinking NBA Aging

Chaos or Order?

I’m dreading my next birthday. It’s not a milestone in any traditional sense, but 28 has a very important meaning in professional sports. Going back to the pioneering research done by Bill James in baseball, it’s been accepted by statistical analysts that players generally peak at age 27, meaning starting in April it’s all downhill for my basketball career, such as it is.
Lately, however, that statistical conventional wisdom has been under attack on both sides. In a controversial paper in the Journal of Sports Sciences that he summarized on Baseball Prospectus, baseball academic J.C. Bradbury found that baseball players peak much later than assumed, at age 29 or 30–more in line with the ad-hoc belief that existed before statistics were applied to the issue. Meanwhile, Wages of Wins author Dave Berri (with whom, coincidentally, Bradbury recently published an unrelated article) was studying peak age in the NBA and finding the opposite–a peak somewhat younger than 27, instead as early as 24 or 25.

For now, however, let’s side aside the issue of peak age. As interesting as that discussion is, the most thought-provoking piece I’ve read on aging in the NBA was a treatise from friends of BBP Bethlehem Shoals and Tom Ziller on Using Ziller’s trademark graphs, the duo explored the notion that some players are better able to respond to the inevitable effects of age by altering their game to remain productive long past what should be their peak. After closing with a comparison of legendary point guards Magic Johnson, Jason Kidd and Oscar Robertson, however, Shoals largely threw his hands in the air. His conclusion: “Prime is different for every player, like fingerprints, as much a function of how he ends up at and sustains star-level performance, not simply how long it lasts.”

It’s hard not to look at those charts and wonder whether we are silly to even discuss aging curves. As convenient as the notion is of someone entering the league, improving, peaking, then beginning to decline before dropping off and retiring, few if any players follow such a formulaic path. More likely, players improve and decline several different times throughout their career, whether because of injury, team fit or simply the randomness in both statistics and human behavior.

One of the most extreme examples cited by Shoals and Ziller continues to play out this season. Steve Nash is defying everything we know about aging with his performance since joining the Phoenix Suns before the 2004-05 campaign. Way back then, I wrote that “it seems unlikely that Nash is still a starter at the end of the six-year contract (last year believed to be at the team’s option) he got from the Suns.” Guess what? It is now that sixth season, and not only has Nash managed to hold on to his spot in the starting five, he is actually posting better statistics than he did during his MVP season. Nash’s best four seasons by WARP came at ages 29 and 31-34, and there’s a good chance we’ll be able to add this year’s age-36 season to that as well.
Even the statistics I used to employ the “delta,” or year over year method, to assess peak age in the NBA in this Unfiltered post last month back up the notion that players follow complex paths. Among players who played at least 250 minutes in consecutive seasons, 70.5 percent of 22 year olds improved the following season. That means nearly three out of 10 players got worse at an age when development seems like a foregone conclusion. Meanwhile, it’s not until age 36 that we see less than a third of veteran players decline.

It is this very kind of probability-based projection that is a key part of Prospectus’ PECOTA projection system, as we’ve attempted to replicate in our NBA equivalent, SCHOENE. “Variation,” Nate Silver wrote in introducing PECOTA in Baseball Prospectus 2003, “isn’t something that a projection system should seek to avoid, but something it should seek to embrace.”
Projecting what players will become with any sort of certainty is an impossible task, and thankfully so, since such a reality would surely be more dystopia than utopia. Why play the games at that point? It’s the outliers, the Nashs, that make following the NBA and all sports so interesting. What these examples ought to offer us is humility in our predictions.

At the same time, we have to make some assessments of player aging. Certainly, NBA teams do. As the Boston Celtics figure out the future of Ray Allen, for example, they will weigh his expected performance over the decline phase of his career against younger alternatives. I believe SCHOENE can help in this process not just by setting general expectations but also withof the kind of assessment made in the column on Gilbert Arenas’ future that kicked off this whole discussion. Scoring-minded point guards, like Arenas, tend to burn out faster than other players because when they lose the step that allowed them to create off the dribble prolifically they struggle to compensate.

Where SCHOENE cannot help as much is with the Nashes of the world. What has separated Nash’s performance in his 30 from those of players who were his equal in their 20s–some combination of his work ethic, his devotion to conditioning, the Suns’ training staff, luck and other factors–cannot be quantified and will always have to be assessed in a subjective fashion.
Ultimately, where I think I disagree with Shoals slightly is that I don’t believe that player aging is unique like snowflakes. Instead, the comparison to nature I would draw is to the night sky. To the untrained eye, the stars are a jumble of random lights, but somehow the ancients stared up there and saw constellations that resembled familiar shapes. Compare that to this graph showing year-over-year aging for the aforementioned group of players who played at least 250 minutes in each season.

At first glance, there’s a whole lot of random dots on that chart, but the regression line does show the trend we expect. (Because this is looking at performance year over year, as opposed to over time, it’s a line as opposed to the familiar parabola. If you add the differences, that’s what you’d get.) I guess it’s up to you determine which more accurately reflects the reality of aging: the chaos of individual year-to-year performance or the relative order of the average changes.

Myself, I’m left with this. Inspired by the example of Nash, I’m confident that my game no longer has to peak at 27. My basketball prime is well ahead of me. Look out, rec center.

Follow Kevin on Twitter at @kpelton.

Kevin Pelton is an author of Basketball Prospectus.

Scroll to Top