Aging curves

[Crow]

Jeff, perhaps you were thinking of this old thread started by MikeG:

http://sonicscentral.com/apbrmetrics/vi ... p?f=2&t=45


DSMok1's thread was here:

http://sonicscentral.com/apbrmetrics/vi ... p?f=2&t=72
I went back in and added the Graphs.

And an old one by Ed K., if one wants to look at more detail:

http://sonicscentral.com/apbrmetrics/vi ... p?f=2&t=72

[Jeff Fogle]

Thanks very much for those links Crow...

[Crow]

Thanks for saying thanks and using the recovered threads.

[Mike G]

I quote myself:

Sorted by their age at entry to the NBA, there's a distinct cutoff between players who first appeared by age 20, and those 21 or older. Players who entered at 18, 19, or 20 have an average peak at age 25.4; those who enter at 21 or older have peaked at 27.1 years, on avg.

From the article linked above by Crow.
viewtopic.php?f=2&t=45

With all the players entering younger, and a sizable number whose careers are completed or have distinctly peaked, it seems the avg peak age, and avg final playing age, are now 1-2 years younger than they were just 10-20 years ago.

[Mogilny]

For the survivor bias issue, it might help to classify players into groups according to age of exit from the league. Look at all players who played until age 25, 26, etc. See what the trends are like within those groups and between groups.

Yeah, and since the survivorship bias results in a population of on average lucky rookie season-guys you could add a projected group of unlucky replacement players (or something) to the sophomore season-guys. This should lessen the effect of a false decline between the rookie and sophomore season. How to do this in a proper scientific way? I haven't got a clue, I just remember reading about it a couple of years ago.

[Jeff Fogle]

Given the information presented...is it better to call individual player development an "experience" curve rather than an aging curve...since the curve seems correlated to time in the league rather than calendar age? And, do we have evidence that a certain amount of "mileage" represents a wall elite players hit? Meaning, players peak at xxx number of minutes...and they're pretty much done as useful contributors at yyy number of minutes?

[Mike G]

There isn't a 'wall'. There are just averages.
We don't know, for example, how many years or minutes Kobe, Garnett, or LeBron will play.

After so-many minutes we can expect players to stop getting better, to decline each year thereafter, and so on.
There will always be players who seem to hit a wall, at any age or experience level.
There will also be those who seem to play forever without showing much effect of age.
But we've yet to see a player going strong into his late 30s who came into the league before age 21. Or 22, even.
http://bkref.com/tiny/CgSL5

[Jeff Fogle]

"We don't know, for example, how many years or minutes Kobe, Garnett, or LeBron will play. "

Right, past studies wouldn't have included their final stages. But, we're not talking about career years or minutes that stars will play later. The questions were about data that's been accumulated so far.

"After so-many minutes we can expect players to stop getting better, to decline each year thereafter, and so on."

I used "xxx" and you used "so-many." Is there a number that the stat field can give to the mainstream media for use in on-air conversations about trading strategies, playoff potential, etc...that says specifically what those minutes are? Understanding that we're dealing with averages of career curves rather than guaranteed certainty that findings will apply to any current player being discuseed.

"There will also be those who seem to play forever without showing much effect of age."
Who in your view has done that? Looked over a few guys in the link you provided. May not have seen the guys you were thinking of. Who specifically in your mind has seemed to play forever without showing much effect of age? Were you looking at Win Shares for that (the link ranked players on Win Shares after age 36) or a different metric? Is decline not universal after xxx number of minutes/years?

"But we've yet to see a player going strong into his late 30s who came into the league before age 21. Or 22, even."

Which would suggest there's a wall. Or, if that's a bad metaphor...reaching a period of decline. When marathoners say they "hit a wall" they don't actually stop cold. So, there's probably a better phrase to express the phenomenon.

So, I'll re-ask the question using "decline phase" instead.

Given the information presented...is it better to call individual player development an "experience" curve rather than an aging curve...since the curve seems correlated to time in the league rather than calendar age? And, do we have evidence that a certain amount of "mileage" represents the likely beginning of a decline phase that elite players hit? Meaning, players peak at xxx number of minutes...and they're pretty much done as useful contributors at yyy number of minutes?

[Mike G]

"There will also be those who seem to play forever without showing much effect of age."
Who in your view has done that? Looked over a few guys in the link you provided. May not have seen the guys you were thinking of. ...Is decline not universal after xxx number of minutes/years?

Heading that list were Malone and Stockton, Kareem, Parish, and Reggie. They all came into the league aged 22 or 23.
It's probably time to get a better look at the actual shape of the experience curve, as a function of minutes and entry age. My suspicion is that marathon minutes at a tender age -- and 'tender' might be just the right term for it -- is a lot harder on a player's body than equal minutes when he's an older rookie. Even when they don't break outright, they break down.

The earlier studies I did were aimed at seeing whether earlier entry was likely to produce a longer career, such that a young player might guess if he'd be better off financially to enter earlier or later. Some analysis had been declaring it was a given that earlier = more $$, by assuming everyone plays to the same age, regardless of how they spend their formative years.

The many arguments about the virtues of having a college career and college education might be circumvented by just allowing the evidence that skipping to the pros actually just means you retire earlier.

[Jeff Fogle]

Thanks Mike. Tried to respond late last night but kept getting timed out or something. The five guys you mentioned at the top of the list all showed clear declines in their mid 30's...so I wasn't sure what you were driving at. You used the term "forever," but then linked to a page ranking players on win shares after 36...where their individual pages then showed clear win share declines in the mid 30's that would follow the standard aging pattern.

Were you saying that some players have a lot longer peaks (plateauing at a peak for a long time I guess) before declining? And, that messes up the study? Want to make sure I'm on the same page with you.

Agree that it would be important to let college kids know that early entry may just mean early retirement. But, they'd probably want to take their shot at the money anyway, or act now for money just in case they suffer a bad injury. Very interesting that starting early saw the same kind of mileage issues. I would have guessed that starting early meant you were destined for bigger careers...with high peaks that took a long time to fall back to the point that you had to leave the league. Instead, it looks like from the data that the limits of what's likely to happen just get pushed forward a couple of years rather than lengthened...

[Mike G]

The five guys you mentioned at the top of the list all showed clear declines in their mid 30's...so I wasn't sure what you were driving at. You used the term "forever," but then linked to a page ranking players on win shares after 36...where their individual pages then showed clear win share declines in the mid 30's that would follow the standard aging pattern. ..

Yeah, I was speaking loosely. Of course they aren't still playing literally to forever. But Kareem was all-NBA 1st team at age 38, Malone was MVP at 35, 3rd-team at 37, etc. Parish, Stockton, Reggie. They not only started late into decline, they didn't follow anything like typical aging curves, which is to decline ever faster. That's why they top that list.
Mostly, they just played fewer minutes in their later careers.

Better players tend to have longer careers, so we have to look at tiers of players with similar ability. I separated players by their career PER through age 24, in groups of roughly 100. Figured by 24, a player is good, or great, or whatever he's going to be. If I did it again, I might choose age 25 or 26. But I didn't want to lose those who have flamed out by then.

Within each group of similarly-skilled players, I separated by age of entry. Sometimes there were only a couple of 19 year olds (as rookies) in a PER block, vs 40 or 50 age-22's.

At every ability level, little or nothing was gained, in career length or career minutes, by entering earlier.

I was going to reconstruct earlier tables in this thread, but maybe I'll start over instead. Soliciting ideas here.

[Jeff Fogle]

Thanks for the explanation MG. In terms of ideas, anything along the lines of adjusting “aging curves” (the title of the thread) to “mileage curves” (based on minutes played rather than calendar age—maybe using spaced out intervals of minute-thresholds) might be something that could have some impact on how things are perceived. You mentioned that old players often maintained a high level of quality within reduced minutes. Maybe there’s a point of no return where minutes “have” to be reduced after a certain mileage point. Or, maybe there’s a standard point where things go downward except for the handful of “freaks of nature” types. Could see that being of value to franchises or the media.

A 28-year old who just missed two years to injury may be in better shape to thrive the next two seasons than a 28-year old who played full campaigns. This (if true) would be more clear looking at mileage than calendar age.