How Development Differs Across Positions, and a Picture of The League Today
Humans develop. We go from infancy to childhood. From there, we take the path of adolescence towards adulthood. Societies develop, from the stage of traditional society all the way up through high mass consumption. NBA players also develop. We are very interested in the development of NBA players because analysis of it can help us not only understand how an athlete performs now, but also give us an idea of how they will perform in future years. While we want a model that can be applied in general, I believe it’s important to at least begin by breaking down at least by position. Players at certain positions develop at different speeds or times than others. That’s an important thing to consider when trying to look into the crystal ball of an NBA player’s future.
I chose a timeline including the last five years of NBA action (the 2012–13 NBA season through the 2017–18 season), and gathered every single player from this timespan. This is a good sample for the modern NBA.
There were 3,304 player data points from the past 5 years of NBA action.
I then used three catch-all stats — VORP (Value over Replacement Player), Win Shares, and Box Plus-Minus — and took percentiles for each player, for each year, for each of the catch-all stats. I then took the average of those three percentiles, producing a rough estimate of a player’s standing in the league.
I took averages of each position at each age for the past 5 years. Here are the results:
Here’s how they look layered on top of each other:
Firstly: Most players don’t actually get better as they approach their forties. Usually, only the world’s best ballplayers are able to sustain a career that long. Those players’ performances are actually much worse than earlier in their careers. They’re also often the only players left at that age. While they’re worse than earlier in their careers, they’re still playing at a level better than or roughly equal to their younger peers. Take 2016, for example. These were the four oldest players in the league: Vince Carter, Tim Duncan, Kevin Garnett, Andre Miller (basketball-reference). The average value over replacement player of the group was 0.7. The average VORP of the league that year: 0.63. This doesn’t mean that as players age, they get better. The average VORP of those four in any given year is 3.45. Now, 2016 was somewhat of an outlier — at age 39, we had two of the 3 or 4 best power forwards of all time, a top 15 shooting guard of all time, and… Andre Miller, I guess. He was good.
So, take the numbers for the older players with a heaping scoop of salt.
What else do we notice about this data?
We see that centers take less time to develop than other positions — there isn’t a huge difference between the talent level of centers in their early twenties, mid twenties, and late twenties.
We also observe that guards come along slower than wings and bigs. This isn’t a big surprise — often, their games are more passing, handling, and shooting-oriented than bigs, and from a physical standpoint, there’s definitely a larger learning curve.
Point guards tend to fall off sooner than other positions. After reaching 32, the drop off is steep — from the 51st percentile at age 31 to the 32nd percentile by age 33. That’s Jerryd Bayless/JR Smith range to Jabari Bird/Nik Stauskas range.
It’s around age 26 when point guards and centers reach the same general level of play. And from age 26 up through age 30, players reach their primes. For all positions, age 31 or 32 appears to be when a steep drop-off occurs. When viewing the percent of the league that exists at each age, it’s clear that 33 is a drop-off point as well, signaling that in the prior one to two years, players recognize they can no longer keep up with the rest of the league.
If I were to do this again, I might include only players who played until their upper 30s, so we could get a better picture of how the best players in the league develop. I’d need to include more years in order to maintain a solid sample size, which may compromise a bit of the “recency factor.”
When evaluating athletes through various stages in their development, it’s important to remember that certain subgroups — whether that be position, playstyle, etc — exist within sports, and that sometimes generalizations about those subgroups can be more helpful than generalizations about the entire league. We now know, based on the data, that guards take longer to develop, while bigs tend to reach their peak from a much younger age, often in the first few years of play in the league.
