Player Regression

Re: Player Regression

Postby Gary Gorski » Wed Nov 17, 2021 4:25 pm

One thing to follow up with as well - remember this game is not 2K. We do not use a rating system that overrates everyone trying to appease players. A 2 star player in the game is literally an average bench player. So a 2 star player is not worthless and does not suck - he's just your normal bench guy.
User avatar
Gary Gorski
WS Development
 
Posts: 8914
Joined: Thu Apr 27, 2006 3:56 pm

Re: Player Regression

Postby KW77 » Wed Nov 17, 2021 5:06 pm

i'm not expecting you to let every player be at the top of his game at 35, but the vast majority of players in that screenshot would be out of rotations in the CSL based on those ratings. There's a middle ground to be had between superstar and deep bench player. Both extremes are unrealistic. So default can be bang in the middle of those two.

To answer your question- I would expect an all-star at age 30 to be a capable starter 3-5 years later unless they were injured, with a chance to still be very good (like the Lowry, etc examples), or a chance to totally fizzle out into a bench player or useless bum (like the screenshot).

I think the default can be a middle ground, and then a modifier where the user can make a call whether your regression model is realistic or not.
KW77
Junior Member
 
Posts: 103
Joined: Wed Apr 29, 2020 2:11 pm
Location: CO

Re: Player Regression

Postby Cleasby » Wed Nov 17, 2021 5:07 pm

I do think the slider model ddspf has for player development and player regression is an awesome feature. So I can see why that could be a great add for debates like this.
Cleasby
Senior Member
 
Posts: 872
Joined: Sun Feb 15, 2009 8:28 am
Location: U.K

Re: Player Regression

Postby Gary Gorski » Wed Nov 17, 2021 6:26 pm

KW77 wrote:i'm not expecting you to let every player be at the top of his game at 35


Literally no player other than LeBron can be considered to have been at the top of their game at 35. Chris Paul has been very good at his age but he's still had a drop off from his late 20s with the Clippers.

but the vast majority of players in that screenshot would be out of rotations in the CSL based on those ratings
.

Even with no tweaks 7 of the 20 players in that screenshot are 2 star or higher which means they are at least average bench players and again I already said that I had made changes that should bump that up. You really think that in 7 years players who are currently either bench players (Hart, Brown, Nunn, Morris, Colson) or are starters who are likely never to be all stars (Kuzma, Brooks, Graham, Nurkic, Capela, Grant, VanVleet, LeVert) are either going to be in the league for that first group or anything but a role player at best for that second? The majority of those 20 players won't be serious parts of rotations in 7 years so that part is already accurate and I've already tweaked things in my build that should handle the other 7.

To answer your question- I would expect an all-star at age 30 to be a capable starter 3-5 years later unless they were injured, with a chance to still be very good (like the Lowry, etc examples), or a chance to totally fizzle out into a bench player or useless bum (like the screenshot).


It's just not that clean. Of anyone who was ever an all star at all from the players currently in the league 35+ years old you have these tiers at the 33-35 stage of their career

Very high level - James, Paul
High level/starter - Lowry
Starter - Anthony, Horford
Starter/role player - Aldridge, Howard, Millsap
Role player - Iguodala
Role player/useless - Dragic, Rondo

From the 33-34 crowd you have

Very high level - Curry, Durant
Starter - Conley, B.Lopez, Westbrook
Role player - Love, Rose
Useless - D.Jordan

So out of the 11 in the first group you have 4 who were "capable starters" through 35. Aldridge was at 33 but not now and Melo has been a good scorer still to this day but I wouldn't classify him as anything great. Lowry was an all star at 33 but he's not as good today as he was two years ago by a noticeable margin. Even out of those 11 though all of them had regression to some degree from their highest point by 33 other than James.

Out of the 8 in the second group I would be surprised if Lopez was starting in two years and DeAndre probably won't be in the league in two years because he's already washed.

I think this is more the model to follow as opposed to either still being very good at 33-35 or a bum. But that's just the list of those 42 guys who had ever made any all star appearance period. That's not even half of the 33+ year old guys who have played this season. So that means that the majority of players do turn into a pumpkin by 33 let alone 35.

Again, in summary, yes - the guys who are still all-star caliber players late in their career need to have the regression toned down so they are still at least capable players in that 33-35 range and they will BUT the majority of players in that
range should not be very useful at all and even the really good players start showing regression in the early 30s - it's just not as drastic as the game has it now. Guys who are marginal starters in their prime should not also remain marginal starters into their mid 30s. Regression has to start for nearly everyone in the early 30s - it just has to be fine tuned a bit so that more outcomes are likely and players can fall into any of those above categories.
User avatar
Gary Gorski
WS Development
 
Posts: 8914
Joined: Thu Apr 27, 2006 3:56 pm

Re: Player Regression

Postby Pialdret » Wed Nov 17, 2021 8:33 pm

So just had a look at NBA seasons from 02-03 until today and from what I quickly gathered was that their really aren't that many great seasons being played by players if any at all post 33 years old. Only Lebron, Curry, KD, Chris Paul, and Kobe have had truly great season at age 33 or older. There has also only been around 20~25 high level starter seasons during that time period for players 33+, with those players being:

Gary Payton, Sam Cassel, Karl Malone, Shaq, Kidd, Nash, Iverson, Allen, Duncan, Carter, Billups, KG, Manu, Jason Terry, Antwan Jamison , Pierce, Dirk, D Wade, Zach Randolph, Marc and Pau, Aldridge, Lowry, Conley.

So unless you were an all star during your or hall of fame talent during your prime, chances are the average nba player is going to see some significant decline once they hit 33.
Pialdret
Junior Member
 
Posts: 16
Joined: Sat Sep 11, 2021 4:58 am

Re: Player Regression

Postby Flaming Homer » Thu Nov 18, 2021 4:40 pm

Cleasby wrote:I do think the slider model ddspf has for player development and player regression is an awesome feature. So I can see why that could be a great add for debates like this.


This
Flaming Homer
Junior Member
 
Posts: 20
Joined: Fri Jul 21, 2017 10:44 am

Re: Player Regression

Postby Gary Gorski » Thu Nov 18, 2021 10:03 pm

I am not a fan of the slider idea here because it's not the right answer to this issue. The answer is to get a proper distribution of players who continue to play well for a few years, players who slightly regress and players who basically just fall off a cliff. A slider is just going to emphasize the same degree of regression to each player and that is not what I am after.

I have made significant changes in the b11.0.2 build and I look forward to feedback on how the results are now.
User avatar
Gary Gorski
WS Development
 
Posts: 8914
Joined: Thu Apr 27, 2006 3:56 pm

Re: Player Regression

Postby Johnny Slick » Thu Nov 18, 2021 10:41 pm

One thing I think you ought to bear in mind - and this is part of why I'd like some kind of config file or number value to adjust aging - is that teams make decisions on older players that have very little to do with their current value. Very often a team will play a younger player over a comparable if not better older guy because:

- The younger player is cheaper - even a guy on the vet minmum makes a lot more money than a rookie on a minimum contract.

- In years past (honestly not sure how it works now) you couldn't put players over a given age on your reserve roster or assign them to the G-League or the CBA.

- While the older player might be better *now*, the younger player at least has a chance of being better in a couple of years.

- In the current NBA, a bad team has very little incentive to play a veteran, especially if said veteran is better than younger players it has access to. If OKC can win 25 with Al Horford and 21 without him, why on Earth would they ever use him?

- Even if the bad team doesn't have a comparable younger player, they *still* have to ask themselves if said player is still going to be a worthy part of the rotation when they're a playoff contender again, in, say, 3 years or so. A 33 year old might still be good enough. At 36? That's a harder question.

- I'm not sure if younger players are actually more versatile than older players but the fact that they have room to improve means they could potentially be directed to improve in a certain direction. A guy like Russell Westbrook at the age he's at is going to be a usage-high, not terribly efficient scorer who will get lots of assists, rebounds, and turnovers. If you had Westbrook at 23 on a team with 2 established stars, there's at least that theory that you could retrain him to be more of a 3rd option type. Again, I'm not actually sure that this is *true*, but it's certainly a thing that coaches think about.

- There are certain varieties of coaches that work better with younger players than with older ones. These coaches tend to drive away vets, particularly vets who don't buy in. Of course there are the opposite but coaches who work better with vets tend to be the "just do your job and I'll try to make things as easy for you as possible to do so" type whereas the ones who work with the younger guys better tend to be the "I am a strategic GENIUS and I need you to do everything I tell you to do, no questions asked" type.

- A vet who has a bad season is much, much less likely to be given another chance to prove that it was just a bad season than a younger player.

All that said, it leads me to say that a lot of the "vets losing their jobs from age" is less "they're actually getting worse" and more "they could, theoretically, still produce in a vacuum but there are a hundred reasons why to play a younger guy instead". There was a study done in the NFL that found that for quarterbacks roughly 50% of "aging" occurred in their final season. Some of that is, I think, a tendency for some NFL QBs to completely fall off a cliff in terms of production (thinking of Peyton Manning in particular, although Manning had essentially 2 seasons as a replacement level QB, not one), but a lot of it is related to the stuff I brought up above, especially the bit where older players don't get second chances. I'm sure the average NBA wing ages more quickly than the average NFL quarterback simply because there's more athleticism required, but the eye test tells me that lots of guys, especially in the current era, are still just as good or at least at like 95% of the level they were at when they were 25 as they are at 33 or 34. In real terms I'm sure that, like, Lebron and Steph Curry have lost a step or two in terms of pure speed, but they've made up for it with basketball IQ stuff and frankly most of the ratings DDS PB uses are based on outcomes, not pure footspeed, etc.
Johnny Slick
Junior Member
 
Posts: 110
Joined: Mon Jul 10, 2006 2:06 am

Re: Player Regression

Postby Gary Gorski » Fri Nov 19, 2021 1:55 pm

Johnny Slick wrote:One thing I think you ought to bear in mind - and this is part of why I'd like some kind of config file or number value to adjust aging - is that teams make decisions on older players that have very little to do with their current value.

Very often a team will play a younger player over a comparable if not better older guy because:

- The younger player is cheaper - even a guy on the vet minmum makes a lot more money than a rookie on a minimum contract.

- In years past (honestly not sure how it works now) you couldn't put players over a given age on your reserve roster or assign them to the G-League or the CBA.

- While the older player might be better *now*, the younger player at least has a chance of being better in a couple of years.

- In the current NBA, a bad team has very little incentive to play a veteran, especially if said veteran is better than younger players it has access to. If OKC can win 25 with Al Horford and 21 without him, why on Earth would they ever use him?

- Even if the bad team doesn't have a comparable younger player, they *still* have to ask themselves if said player is still going to be a worthy part of the rotation when they're a playoff contender again, in, say, 3 years or so. A 33 year old might still be good enough. At 36? That's a harder question.

- I'm not sure if younger players are actually more versatile than older players but the fact that they have room to improve means they could potentially be directed to improve in a certain direction. A guy like Russell Westbrook at the age he's at is going to be a usage-high, not terribly efficient scorer who will get lots of assists, rebounds, and turnovers. If you had Westbrook at 23 on a team with 2 established stars, there's at least that theory that you could retrain him to be more of a 3rd option type. Again, I'm not actually sure that this is *true*, but it's certainly a thing that coaches think about.

- There are certain varieties of coaches that work better with younger players than with older ones. These coaches tend to drive away vets, particularly vets who don't buy in. Of course there are the opposite but coaches who work better with vets tend to be the "just do your job and I'll try to make things as easy for you as possible to do so" type whereas the ones who work with the younger guys better tend to be the "I am a strategic GENIUS and I need you to do everything I tell you to do, no questions asked" type.

- A vet who has a bad season is much, much less likely to be given another chance to prove that it was just a bad season than a younger player.


I'm going to respectfully disagree with most of this. It's really much simpler in almost every case. If the team can win they're going to play their veterans. If the team isn't any good then they're either going to play the veterans for a short period in the hopes of showcasing them for a trade or they're going to not play them at all as was the case with OKC and Horford and now Houston and John Wall. Horford starts in Boston now after being told to basically go home for much of last season with OKC. Take Brooklyn for example. A lot of people think Cam Thomas could be a very good player - he averages 5 mpg right now while 33 year old Patty Mills averages 26.1 per game. Mills is no world beater so why are they playing him instead of the rookie who absolutely has upside? Golden State - Kuminga is a lottery pick who's already had a couple of highlight reel plays. He plays 8.3 mpg while 33 year old Bjelica gets double the minutes and 38 year old Iguodala is out there for almost 21. Why? Nobody thinks that in 3 years Thomas or Kuminga won't be better than those guys. In fact in 3 years I doubt Mills, Bjelica and Iggy will even still be in the league.

I'm going to single out this quote in particular

A guy like Russell Westbrook at the age he's at is going to be a usage-high, not terribly efficient scorer who will get lots of assists, rebounds, and turnovers. If you had Westbrook at 23 on a team with 2 established stars, there's at least that theory that you could retrain him to be more of a 3rd option type. Again, I'm not actually sure that this is *true*, but it's certainly a thing that coaches think about.


Westbrook has been a high usage inefficient scorer his whole career. His career average FG% is 43.7 and 3P% is 30.5%. At 23 Westbrook was on a team with Kevin Durant who was absolutely an established star, James Harden and Serge Ibaka - a team that went to the NBA Finals. I think this only verifies my approach to the game in that players are what they are and its virtually impossible to get them to be something else. There was no way you could "retrain" Westbrook. That was one of the great frustrations of the Pistons second championship run - Rasheed Wallace was an excellent low post player but he had no desire to do it. Even one of the most respected coaches in the game (Larry Brown) could not get him to play differently. That Pistons team was also a great example of the veteran debate. They took the #2 overall pick, an incredibly highly touted prospect and he got to play 4.7 mpg while 35 year old Elden Campbell played 13.7 and 33 year old Mehmet Okur played 22.3 and shared the starting C duties until Rasheed was traded for. The Pistons won the championship. In fact Milicic hardly played for Detroit in his entire time there because they were chasing more titles.

but the eye test tells me that lots of guys, especially in the current era, are still just as good or at least at like 95% of the level they were at when they were 25 as they are at 33 or 34. In real terms I'm sure that, like, Lebron and Steph Curry have lost a step or two in terms of pure speed, but they've made up for it with basketball IQ stuff and frankly most of the ratings DDS PB uses are based on outcomes, not pure footspeed, etc.


This is just not correct. If that was the case then how do you explain the majority of 33 and 34 year olds playing far fewer minutes and putting up way worse numbers in terms of the ones who are playing on teams that are trying to win?

At 27 Danilo Gallinari averaged 19.5 ppg and played almost 35 mpg (two years removed from missing a season because of an ACL injury) - now at 33 he averages only 7 ppg in 18.7 mpg. The Hawks have championship hopes - why were they giving Hunter 27.8 mpg to average 11 points and Reddish 22.8 mpg to average 11.6 points when if Gallinari was 95% as good as his prime he would be way better than either of them?

At 25 DeAndre Jordan led the league in rebounding with 13.6 rpg to go with his 2.5 bpg and 10.4 ppg. Now at 33 why has he gone from starter to either DNP or barely playing in just 14 games averaging 5.7 rpg and 4.8 ppg? The Lakers had to move Davis to center (which he has said multiple times he hates and does not want to do) because Jordan is so awful at this point. If he was 95% as good he'd be exactly what they could use to protect the rim, rebound and catch lobs off the pick and roll from Westbrook and LeBron.

At 25 Nic Batum averaged 13/7.5 rpg/5.1 apg - now he averages 9.6/5.7/1.6 still in a starting role on a playoff hopeful team. The Clippers are desperate for anyone to do anything other than PG so why has Batum taken a backseat to Reggie Jackson, Terance Mann, Eric Bledsoe and Luke Kennard on offense?

I've already pointed this out multiple times in this thread - Steph, KD, LeBron are complete outliers. They are 3 of the most talented people EVER to play the game. There are not "lots of guys" who are as good or 95% as good at 33/34 than they were at their primes at 25-28 years old. That doesn't mean there are none and it doesn't mean that most of the players are washed and have no business playing at 33 but virtually all of them have had a decline even the ones who are still good players. In his prime Horford was a threat for 20/10 every night. He's still good enough to start and give you 14/8 maybe but that's not almost as good as he was in his prime. That's a 30% drop in points and 20% drop in rebounds. That's what has to be replicated. There has to be a handful of guys who have either barely dropped off or dropped off but still are good enough to play and there has to be a majority of players who have either dropped from star to decent rotation player or who have dropped to barely good enough to be in the rotation.
User avatar
Gary Gorski
WS Development
 
Posts: 8914
Joined: Thu Apr 27, 2006 3:56 pm

Re: Player Regression

Postby Johnny Slick » Fri Nov 19, 2021 3:32 pm

I mean, also, just from guys I've watched a lot of over the past couple years (I live in Chicago and have season tickets for the Bulls), I'm also seeing Demar DeRozen and Thad Young playing as well in their early to mid 30s as they ever have. And you bring up Al Horford as a big example the other way: Horford was a 20/9 guy per 36 when he was 27 but he's not a "14 and 8" guy now; he's a 16 and 10 guy now who gets fewer minutes. And at that, his age-27 season was a scoring outlier: the other 3 "prime" seasons sandwiched around that, he averaged 17, 18, and 17 (I went with 26-29 because if you go back to 25 his usage was even lower). That's simply not as big of a decline as you imply it is. It is one, sure, and if you wanted to downgrade the endurance and some of the athleticism-based stats, sure, but on a per-minute basis he's nearly as productive as he was in his prime.

I'm kind of confused as to why you think a vet playing substantial minutes on a playoff contender over a rookie is an argument in favor of what you're saying. Yes, absolutely, on the flip side you have teams who would rather go after "proven vets" in their rotation. I didn't list every single factor that real GMs use, just a bunch of them that real GMs use to discriminate against older players (I'm not calling this ageism, don't get me wrong; most of the time those are perfectly valid reasons for wanting to play the younger guy).
Johnny Slick
Junior Member
 
Posts: 110
Joined: Mon Jul 10, 2006 2:06 am

PreviousNext

Return to DDS: Pro Basketball 2022 FirstAccess

Who is online

Users browsing this forum: No registered users and 2 guests