More Aging Stuff
Posted by Neil Paine on July 16, 2009
In response to Tuesday's post about Win Shares and aging, reader Jason J thought about whether or not today's improved equipment (including much better shoes), training methods, and dietary regiments made it easier for older players to stick around in the NBA for longer periods of time. It's a great question, because the success of recent players like John Stockton, Karl Malone, Michael Jordan, Reggie Miller, Dikembe Mutombo, etc. well into their forties makes you wonder if modern basketball technology is helping older players extend their careers longer than ever.
As a follow-up to that earlier post, today I'm going to look at what percentage of league Win Shares were credited to players at each age, both in the history of the NBA and broken up by decade (the fact that we're essentially through with the Oughts now is handy as well). Now you may be asking, how am I going to use Win Shares as the measuring stick, since we're unable to calculate that metric for seasons prior to 1978? Well, similar to the way we estimate PER and Statistical +/- for pre-1978 seasons, I also have estimated Win Shares for those who played in the pre-turnovers/offensive rebounds/blocks/steals era. For instance, here's Bill Russell:
| Name | Year | Ag | Tm | Ht | Pos | G | Min | WS |
|---|---|---|---|---|---|---|---|---|
| BillRussell | 1957 | 22 | BOS | 81 | C | 48 | 1695 | 7.3 |
| BillRussell | 1958 | 23 | BOS | 81 | C | 69 | 2640 | 12.7 |
| BillRussell | 1959 | 24 | BOS | 81 | C | 70 | 2979 | 14.2 |
| BillRussell | 1960 | 25 | BOS | 81 | C | 74 | 3146 | 15.4 |
| BillRussell | 1961 | 26 | BOS | 81 | C | 78 | 3458 | 13.8 |
| BillRussell | 1962 | 27 | BOS | 81 | C | 76 | 3433 | 15.6 |
| BillRussell | 1963 | 28 | BOS | 81 | C | 78 | 3500 | 14.3 |
| BillRussell | 1964 | 29 | BOS | 81 | C | 78 | 3482 | 16.3 |
| BillRussell | 1965 | 30 | BOS | 81 | C | 78 | 3466 | 16.2 |
| BillRussell | 1966 | 31 | BOS | 81 | C | 78 | 3386 | 12.2 |
| BillRussell | 1967 | 32 | BOS | 81 | C | 81 | 3297 | 13.2 |
| BillRussell | 1968 | 33 | BOS | 81 | C | 78 | 2953 | 9.3 |
| BillRussell | 1969 | 34 | BOS | 81 | C | 77 | 3291 | 11.1 |
Like I say, they're estimates at best, but I think it gives us a decent way to pull off studies like this, where a sample confined to players from 1978-2009 is just not big enough to be acceptable.
So, armed with this data, we can take a look at what % of league WS were attributed to players at each age. I'm also going to break things down by height; the median height in NBA history is 78 inches (that's 6'6" for the lazy), so I'll track the percentages by age for the subgroups of those 6'6" or shorter and those 6'7" or taller. Here's the breakdown for all of NBA history:
| Age | 1950s | 1960s | 1970s | 1980s | 1990s | 2000s |
|---|---|---|---|---|---|---|
| <25 | 29.6% | 29.9% | 32.2% | 29.7% | 20.0% | 29.3% |
| 25-29 | 56.3% | 51.4% | 49.7% | 51.3% | 49.3% | 42.8% |
| 30-34 | 13.8% | 18.4% | 16.6% | 17.2% | 27.1% | 22.5% |
| 35-39 | 0.3% | 0.4% | 1.4% | 1.7% | 3.5% | 5.1% |
| >39 | 0.0% | 0.0% | 0.0% | 0.1% | 0.1% | 0.2% |
And here's a year-to-year graph of the phenomenon:
It looks like the uptick in productive older players was almost a one-time situation unique to the 1990s, which only seems to confirm that the stars who grew old in that era were a very special group of athletes. There were just so many superstars with staying power during that period -- Michael Jordan, Hakeem Olajuwon, David Robinson, Clyde Drexler, Charles Barkley, Karl Malone, Patrick Ewing, John Stockton... The list goes on and on. Truly, 'twas a great time to be a hoops fan.
Anyway, here's the data for just the "tall players":
| Age | 1950s | 1960s | 1970s | 1980s | 1990s | 2000s |
|---|---|---|---|---|---|---|
| <25 | 37.9% | 35.1% | 35.7% | 29.4% | 21.8% | 31.9% |
| 25-29 | 54.6% | 49.0% | 49.7% | 49.6% | 46.0% | 42.7% |
| 30-34 | 7.5% | 15.9% | 13.8% | 18.2% | 27.7% | 19.8% |
| 35-39 | 0.0% | 0.0% | 0.8% | 2.6% | 4.5% | 5.4% |
| >39 | 0.0% | 0.0% | 0.0% | 0.1% | 0.1% | 0.2% |
And the "short players":
| Age | 1950s | 1960s | 1970s | 1980s | 1990s | 2000s |
|---|---|---|---|---|---|---|
| <25 | 22.4% | 25.0% | 28.7% | 30.1% | 17.4% | 25.2% |
| 25-29 | 57.7% | 53.5% | 49.8% | 53.8% | 53.9% | 43.1% |
| 30-34 | 19.3% | 20.7% | 19.4% | 15.7% | 26.4% | 26.8% |
| 35-39 | 0.7% | 0.8% | 2.1% | 0.4% | 2.2% | 4.7% |
| >39 | 0.0% | 0.0% | 0.0% | 0.0% | 0.0% | 0.2% |
So, conclusions? Aside from essentially having more 30-somethings contributing from the start, there doesn't seem to be an appreciable difference between the shape of the graphs for bigger players and smaller ones. Both saw an increase in production from older players during the mid-to-late 1990s (the smaller players saw a distinct spike) as some of the aforementioned superstars neared the end of their illustrious careers. But during this decade, the 2000s, we have seen a return to the youth movement of the 1970s and early 80s, presumably fueled by the simultaneous retirement of greats like Jordan and the inclusion of young early-entry/prep-to-pro players like Kevin Garnett, Kobe Bryant, Tracy McGrady, & LeBron James. In fact, 2009 represented the greatest % of league Win Shares earned by the under-25 set in nearly 30 years, meaning that while today's older players don't stack up well vs. their 1990s counterparts, the future is nothing if not bright for the league as a whole because of an exceptional crop of youngsters on the rise.
July 16th, 2009 at 9:34 am
Wow!
This is so awesome I am at a loss for words (almost).
Turn it upside down, and the old guys are the 'bedrock' of the league; the youngsters are the sky; they just fill in any void of experienced players. It may make more sense that way.
What about the ABA? By snatching up kids from college (that the NBA wouldn't touch), significant younger talent would appear.
Huge tradeoff of older/younger guys right around the introduction of shot clock ('53 to 54).
The late-90s spike of older players' Win Shares might be exaggerated by whatever coincidence placed many older players on the best teams for a few years. You could do this by % of minutes played, but that might have the opposite effect. % of points and % of rebounds would be great.
Stupendous!
July 16th, 2009 at 9:50 am
Maybe it's even better with grid lines?
On the 'shorter players' chart:
Somewhere around 1998, 'old' guys (>30) approach 50% of all WS. Normally the 'prime age' (25-29) guys comprise about 50%, but at this time they're pinched down to maybe 25%.
In the late '80s, there may be more WS in the >40 group than in the 35-39 range, which is down to nearly nothing. This recurs around '93-94.
Very nice.
July 16th, 2009 at 12:14 pm
What effect did Bird's back injury and Magic's retirement due to HIV have on that 35-39 range in '93-94? Add those two back in and maybe get an extra year out of McHale and Worthy in the process, and the WS might bounce back up.
July 16th, 2009 at 12:36 pm
The late 90s uptick could have alot to do with that being the only era measured where underdeveloped 19 year olds were playing significant minutes.
July 17th, 2009 at 10:54 pm
I wonder what the correlation was to players hanging round longer and the spike in salaries that occurred during the nineties? Surely a bunch of quality guys stayed simply to cash in on the big paychecks that they didn't get in the 80s
July 18th, 2009 at 8:46 am
Neil,
Another great article! I have done some work with estimating win shares for players before 1973-74 and believe it can be done with a large amount of accuracy. What formula or methodology did you use for estimating the win shares of Bill Russell?
July 19th, 2009 at 10:50 pm
I suspect Team Expansion in the 1990s allowed Players like Jerome Kersey and Cliff Robinson to have above average length careers. (I wonder if this also contributed to the 84/85 draft blip we have observed.) Expansion together with the sorts of money players could now earn has meant they stay in the league (while they can) - as opposed to past generations where many a player retired to a better paying / more reliable income proposition.
Neil, You have charted the Win Share distribution broken up by age, but you have not indicated the relative proportions of people populating the different categories - I am not sure it is valid to seperate the 40 yr olds off and not the under 20s - especially since the latter group has probably always been larger...
July 20th, 2009 at 9:37 am
The graph is so spiky, it's hard to see a trend sometimes. How might you smooth it out some?
Given that the (5 years) age blocks are arbitrary, and compounded by the fact a player isn't a precise (integer) age in a given season -- what if you count all 23-25 year olds as '24', etc. You'd get an inflated total number of 'players' (or WS in this case), but they'd still add up to 100%.
For example, there's a crazy spike of over-30 player-WS in 1967 (I think), sandwiched between 2 sharp dips. (If you invert the image, that's what it looks like: hint) Chances are, some guys arbitrarily passed 'age 30', and some others happened to pass 'age 25'. Especially in a small (9-team) league, this is an issue.
July 21st, 2009 at 4:47 pm
Anyone home?