Penalized Regression of WOWY data

[AEnigma]

Yeah in terms of standard deviations that is probably the most potent summary I have seen of Russell’s era-relative advantage.

[Moonbeam]

Here is a spreadsheet with up to 100 positive coefficients for each 5-year window for Ridge, Lasso, and ENet. I'll see if a spreadsheet with the full data is navigable and post separately if so.

So basically Russell and Magic look awesome and everyon else looks not so awesome

Also, uh:



Yikes!

Kind of! There is a bit of potentially useful information lost by looking at percentiles only. The caveat to that is because the ridge models are choosing different penalties for each window, the degree of "outlierness" is not always directly comparable. But yes, Russell and Magic are beasts in these pure versions!

[OhayoKD]

When your spankage per 100 poss is so high you still spank your own teammates

Here is a spreadsheet with up to 100 positive coefficients for each 5-year window for Ridge, Lasso, and ENet. I'll see if a spreadsheet with the full data is navigable and post separately if so.

So basically Russell and Magic look awesome and everyon else looks not so awesome

Also, uh:



Yikes!

Kind of! There is a bit of potentially useful information lost by looking at percentiles only. The caveat to that is because the ridge models are choosing different penalties for each window, the degree of "outlierness" is not always directly comparable. But yes, Russell and Magic are beasts in these pure versions!

Nah, too late.

I'm officially blacklisting anyone who has played with or against russell from my ballot for this and all future top 100's,

[lessthanjake]

I think we should remember that it’s easier to finish at the very top of these in a smaller league. The top players in a lot of time periods for these spreadsheets are actually completely random people who barely played. So them being at the top is obviously just noise. The larger the league, the more low-playing-time players like that you’ll get in a given time period, and the more players like that there are the higher the chances of there being ones that randomly have a really high score. So I wouldn’t equate Bill Russell constantly being at the top as necessarily being better than a guy more recently who isn’t at the very top but is consistently ahead of everyone but random guys. For instance, if we look at the Ridge version, Steph isn’t at the very top in any time period, but as far as I can tell there’s not any particularly meaningful player above him in the last 5 time periods. Same with Steve Nash for like 5 consecutive time periods in the 2000s. Granted, Bill Russell is still the very top meaningful player in *more* timeframes than anyone else—and that strikes me as a great data point for him—but I do think other players have shorter but still significant timeframes where they look similarly good as Russell when we take into account the added amount of random-noise guys in later years.

[homecourtloss]

When your spankage per 100 poss is so high you still spank your own teammates

So basically Russell and Magic look awesome and everyon else looks not so awesome

Also, uh:



Yikes!

Kind of! There is a bit of potentially useful information lost by looking at percentiles only. The caveat to that is because the ridge models are choosing different penalties for each window, the degree of "outlierness" is not always directly comparable. But yes, Russell and Magic are beasts in these pure versions!

Nah, too late.

I'm officially blacklisting anyone who has played with or against russell from my ballot for this and all future top 100's,

Really interesting looking at these in this form. A few things:

—This thread is carrying the internet right now (for me at least )
—Russell might be underrated at anything under #1 with how he looks here
—Wilt even with the caveats mentioned looks great
—Magic’s stock has to be going up now I would think.
—Akeem/Hakeem seems underrated in his early Rockets days. There doesn’t seem to be any other Rocket around in those mid late 80s.
—McHale, Dennis Johnson, and Parish look really, really good.
—Bill Walton…what could have been. Is there any era in NBA history in which he wouldn’t be a monster impact force? Everything he does translates immediately. The 1986 to 1990 segment that has only his 1986 season seems to corroborate the impact he had on that team, especially given that McHale missed 14 games in 1986
—Horace Grant looks even more impactful in this. His impact lasts all the way through his mid to late 30s. A high motor, high IQ, unselfish, defense first player seems to be underrated even though many on this particular board have been singing his praises for a while. On that note, A.C. Green also looks good.
—KG and Shaq have already been discussed

[homecourtloss]

Here is a spreadsheet with up to 100 positive coefficients for each 5-year window for Ridge, Lasso, and ENet. I'll see if a spreadsheet with the full data is navigable and post separately if so.

A few requests for Moon if possible:

can you create a graph for Ralph Sampson, Hakeem, Rodney Mcray, and Robert Reid, Ridge and Lasso?

And another for Jordan, Pippen, Horace Grant, and BJ Armstrong, Ridge and Lasso?

[Doctor MJ]

.

Yay! Thanks Moonbeam!

At present, I only have data back to 1952 because of the minutes requirement I'm using. I've thought about modelling minutes based on available box score stats for earlier periods so we can maybe get something for Groza, Feerick, etc.

- Rochester Royals if we can get good numbers at least back to their joining of the BAA. (NBL back to '45-46 would be amazing, but the data is super sparse)
Key players: Bob Davies, Arnie Risen, Bobby Wanzer, Jack Coleman, Arnie Johnson

Yeah, this one's tough. Thanks for trying but I'm not sure what to do with it at present.

- Minneapolis Lakers ideally back to their joining of the BAA.
Key players: George Mikan, Jim Pollard, Herm Schaeffer, Slater Martin, Vern Mikkelsen, Clyde Lovellette

Note: No Schaefer due to no MP.

This is telling. I note that Mikkelsen's big drop off comes with the addition of 1957. So Mikan & Pollard retiring and Martin leaves, Mikkelsen stays and the team falls off, and it's like he's the guy who comes back from the restroom to find all his friends left him with the bill.

Granted, I think Mikan & Pollard are the clear top two guys, and I think it's incredibly impressive what Martin did in St. Louis, so I'd rank them in the same order as this graph suggests, but it doesn't look quite as disappointing with this context.

- Syracuse Nationals
Key players: Dolph Schayes, Paul Seymour, Red Rocha, Earl Lloyd, George King, Red Kerr

Quite closely bunched together at the time of their chip.

- Philadelphia Warriors
Key players: Paul Arizin, Neil Johnston, Jack George, Tom Gola, Wilt Chamberlain

Okay, so what we're seeing here is the Warriors getting Arizin back one year, but taking a big leap the next year with the arrival of Gola, which seems to interpret Gola as the keystone rather than Arizin.

If we didn't have the prior stint with Arizin before his military service, this would seem a more reasonable interpretation to me, in part because Arizin's statistical improvement the next year could be argued to only be possible with Gola's addition.

As is though, the fact Arizin showed he was basically the only one who could score like this in this era, and the champion Warriors rode him all the more strongly in the payoffs, I'd still say that the most likely conclusion here is that Goal is getting credited with Arizin getting his sea-legs back.

That doesn't mean I don't think Gola's impressive here though. I frankly think there's pretty good reason to think that Gola may have been a better player than Johnston. I think that sounds pretty heterodox given how we've perceived the two players on this board over the years, but of course Gola was a big-time college star and Johnston was the furthest thing.

Gola was being drafted on to a Warrior team that already had two scorers in Arizin & Johnston, and fit in by leaning on his all-around game doing less scoring than in college. As we look back from the future with access to all these NBA stats, it's easy to be skeptical that Gola could have scored more in the pros, but the reality that from Arizin & Johnston to Chamberlain it just always made sense for him to focus on things other than scoring.

With this in mind, I think we really need to make sure we don't dismiss him lightly.

Wilt continues to look amazing, and I remain impressed but curious to see shorter-stints regressed.

- Boston Celtics
Key players: Bob Cousy, Ed Macauley, Bill Sharman, Bill Russell, Tom Heinsohn, Frank Ramsey



- Boston Celtics
Key players: Bill Russell, Sam Jones, John Havlicek, KC Jones, Tom Sanders, Bailey Howell



- Boston Celtics
Key players: John Havlicek, Dave Cowens, Jo Jo White, Paul Silas, Don Chaney, Don Nelson

Okay, taking these Celtics altogether:

Russell is clearly a common thread like no other.

Heinsohn begins basically tied with Russell...in a period that basically just includes their rookie year and nothing else. I am not going to put much stock in his numbers with this starting point.

I will say though I was looking to see if Ramsey surpassed Heinsohn, and there really doesn't seem to be anything suggesting he did. He may still have been the better playoff performer, but I'll be keeping keeping Heinsohn above Ramsey on my lists despite continuing to wonder what might have been if Red had turned Ramsey loose more.

Similar with Cousy & Sharman. I respect Sharman so much I kinda wanna make the argument for him, and before Cousy, but I think realistically I won't be championing Sharman in the 100. I suppose that's true of Cousy as well in practice, but Cousy's a clear cut Top 100 guy to me.

Havlicek looking quality. I've long had him as my #2 for this extended Celtic era, and so that feels re-assured.

Cowens really does not look as impressive as I was expecting though.

I'm a bit disappointed that Sam Jones doesn't look stronger. I'm always doing a little revisionist analysis when I put Sam ahead of Cousy. I think I need to ponder whether I really have enough ground to stand on there.

I'd really love to totally buy into KC Jones' late numbers, but that's largely in a period where he's retired and part of that whole "peaking at the end" concern.

- St. Louis Hawks
Key players: Bob Pettit, Cliff Hagan, Lenny Wilkens, Clyde Lovellette, Zelmo Beaty, Lou Hudson

This is interesting.

So first we have Hagan having the edge across the first stellar era over Pettit, and then we get Wilkens with the edge in what's universally considered Pettit's extended run as top player.

Then we have Zelmo who surpasses Wilkens just before he leaves, after which he goes up as he thrives in the ABA, and the Hawks fall off a cliff. There's more to the Hawks falling off a cliff, but this is making me more and more confident that Zelmo is a guy we all need to try to get our head around better.

If only so we get to type "Zelmo" more often, which is just fun.

- Philadelphia 76ers
Key players: Wilt Chamberlain, Hal Greer, Chet Walker, Billy Cunningham, Luke Jackson, Wali Jones

Wilt on top again of course.

Cunningham really looking solid for more seasons that I was expecting. I've been strongly considering both Hal & Chet over him, but I'd probably go back Cunningham as the 2nd of that bunch now.

I'm very cautious about those Jackson numbers which peak in an injury-soaked period that includes years of him playing < 24 MPG.

- Los Angeles Lakers
Key players: Elgin Baylor, Jerry West, Dick Barnett, Rudy LaRusso, Wilt Chamberlain, Gail Goodrich

Wilt again.

West looking incredible.

Baylor looking quite spotty but certainly a general positive.

Barnett looks pretty solid, the rest really not so much.

- New York Knicks
Key players: Walt Frazier, Willis Reed, Dave DeBusschere, Dick Barnett, Earl Monroe, Bill Bradley


[/img]

Okay, so this is interesting.

We have Frazier start out with a lead over Reed that Reed then matches and briefly surpasses. With that early lead I think we should note that the turnaround of the Knicks came mid-season in '67-68 when with Frazier getting put in the game mid-season. By the following post season he was a clear cut star.

How did he get off the bench? New coach named Red Holzman came in and quickly decided that's how things should be, and the team got better.

All this raises a question of whether Frazier's getting the last-piece-of-the-puzzle benefit where whoever joins the core right when the biggest breakthrough happen looks great with regression. If some other guy, like Reed would have had the same last-piece-of-the-puzzle effect we should not give this credence but we'll never truly know.

I think at the least Frazier's better health, despite not having great longevity himself, continues to give him the nod in my book.

Have to say DeBusschere looks a bit more solid than Barnett.

Is Bradley missing?

- Milwaukee Bucks
Key players: Kareem Abdul-Jabbar, Oscar Robertson, Bob Dandridge, Jon McGlocklin, Greg Smith

Wow these guys are bunched together during their window. Makes me think we should be very cautious about using it to come to any big conclusions, but Top 100 contenders Kareem, Oscar & Bobby certainly don't look bad around those years.

[Doctor MJ]

^Duncan/KG weren't in the doc yet when Doc made his post.

I appreciate that eminence.

It's frustrating when people jump to the idea that people are deliberately excluding guys who should obviously be included when other things are both way more likely and way more socially positive.

It would be really great if we can try not to do that here.

[Doctor MJ]

My guesses would be Duncan/KG/Dirk/LeBron/CP3/Steph based of the more granular stuff, but who knows.

I would enjoy having some of this stuff in a spreadsheet/table to browse for sure.

Regardless, I do think the onus is finding arguments for the non-100th-percentile guys over the 100th-percentile guys.

Eh...not sure I agree with this.

First off arguments have been made that involve much larger samples and which do not rely on data tied largely to when players happen to miss games:

Spoiler:

(Will circle back to this later)

More importantly, this(just like real RAPM) is not designed to distinguish between 71 or 72 Kareem or 96-98 Jordan. So sorting players into whether they hit the 99th or 100th percentile is kind of missing the point. If you want to compare the highest highs, rapm and rapm approximations are not designed for that as they are curving those highs down.

What matters here is frequency

Kareem is at or higher than the 90th percentile 12 times scoring at the top level for nearly a decade. You might also note that he goes down when 72, 77, and 1980 are introduced. "peaks," which replace down-years in terms of on-court results , but where Kareem doesn't miss any time. There is also srs suppression from 74-onward(you might notice that jordan by comparison is suddenly skyrocketing when srs for all the top teams goes up after being well behind the pace for what is conventially considered his prime)

And yet with all the above, Kareem still is constantly hovering around the top and then adds a bunch of value later.

Yet, applying a very arbitrary filter for one-offs, you've found a way to get him tiered below a shitton of players he looks as good or better than when we do year-by year analysis or focus on concentrated samples of off, and he is very clearly, "by impact" a much more clear cut era #1.

I think the onus is on you to explain why --this-- matters more than all the other arguments/evidence people have made/offered, especially when we're sneaking in MJ, Bird, and Shaq alongside actual(emperical) impact kings like Magic and Russell and more consistent contenders(at least by this metric) like Wilt(who I do not think "seems right" according to your priors).

Also FWIW, I'm not sure putting all your stock in this does all that for Bird because even by the seasonal inputs of a guy who had him higher than Magic, he still fell down to 14th.

I'm not why regularization should be seen as changing the meaning of percentiles. While the mechanism of regularization could change the order of the players, percentiles are still percentiles.

I focus on the guys who are consistently on the top because of it just gives more confidence that they could achieve it as a matter of course.

Now, of course these are coarse metrics, so I wouldn't want to argue for a player purely because he has a 5-ish percentile edge even if it is sustained for significant duration...but that doesn't mean I wouldn't look to ask and try to explain what I do see while making use of the other information at my disposal.

Re: onus. I should have avoided that words. It would have been better to have said that there's an opportunity than to have implied an obligation.

Re: even...a guy...higher...still fell down. Honestly not sure what you're referring to, so feel free to point me.

What I will say though up front is that there's really never a time where I'm looking to change my opinion about player rankings based on the player rankings of a statmaker. I may look at both his stats and his arguments and find them persuasive, but my conclusions are my conclusions.

To repeat a go-to of mine:

I can have a debate with someone and conclude that they are probably right and I'm probably wrong, but not change the direction of my assessment, rather only decrease my confidence in that assessment.

I don't believe in adopting others conclusions as your own, and I think people in the room should feel comfortable bringing up their viewpoints even if they don't think they know better than others, simply because diversity of perspectives increases the chances that we become aware of a new optimal. All involved, but especially those new to the room, should respectful and courteous when they do this of course, but if you're in a room like this, you should be seeing it as an opportunity to refine your understanding even if you've been around the block many times before.

[LukaTheGOAT]

I mean, "apples to apples" have played a big role in the shift towards "two-way rules". That's where the idea of Russell, Lebron, and Kareem as a tier onto themselves, and Duncan, Hakeem, and KG's elevation largely come from.

As always, things are never so cut and dry, different WOWY metrics can lead to drastically different results. The same DARKO you praised in the other thread, sees MJ as the outlier of his team.

First iteration of DARKO WOWY:

Read on Twitter

This WOWY provides predictive ratings which update at an individual game-level, as opposed to requiring multiple years of data, so it is quite good for showing year by year impact. This doesn't have issues with overly smoothing multiple years of data, thus ignoring aging/injury.

Like usual you read selectively:

Well, no, not necessarily. After, all accuracy writ large does not preclude bias distorting a specific comparison. That is why it's worthwhile to distinguish between box and non-box, even with more "accurate" metrics. When we do so with LEBRON and RAPTOR, we see that the box-prior boosts Steph relative to his impact. Shaq's point is fair.

Fwiw, the metric at the top of the that graph(and one that consistently tests at or near the top more than the others) is darko which to my knowledge is
A. either only using RS data or is using full-season data with no playoff filtering
B. Has a predictive component (the line)and a descriptive one(the dots).
C. It curves descriptive data up and down based on typical aging curves(keep that in mind). Crucially it only curves future data-points, it does not curve retroactively(30+ years cannot bolster earlier ones, under 30 years can bolster(or cripple) 30+ ones)

...

Woah! the most accurate metric has...steph=cp3? Duncan>Lebron? Lebron>Curry>Shaq?

This is an example of where understanding how a stat(and who it might be biased against) informs how we interpret it.
-> Lebron's trajectory is far higher than anyone's up until...2011 which, at age 26, is typically around a player's "peak". Lebron looks dissapointing in that peak so everything after is capped. As everything before is typically "pre-prime" it's curved down as it happens.

-> Duncan has a phenomenal start and doesn't falter until he's 30. His arc mantains allowing it to leapfrog Lebron and by extension everyone else.

-> Steph has a slower start than CP3 but is great at the years that should be his peak so is able to quickly ascend to match.

-> The data does not include Shaq''s suprising early years so his trajectory is not as high as it should be(though tbf, im not sure that actually puts him at a disadvantage relative to steph)

Clearly this data has a bias towards players with "clean" trajectory: Be really good early -> don't dissapoint in a season placed in your mid-20's -> dominante darko

Does this make the stat useless? No. But it does mean you should factor in what it's doing when you extrapolate. This approach is probably accurate on a large scale, but it may distort things in specific comparisons. Account for that and you can still get some use here. Lebron was outlier-valuable for players his age through 24 and then not so valuable in what would typically be a peak year by age(Lebron was aged 26 for 2011). Duncan was really valuable early on and stayed that valuable right up till 30. Steph was underwhelming at the start but was really impactful at 25 and 26. Chris Paul had high impact from the getgo.

I made a point about not taking metrics at face-value and you turn this into "ohayo praised DARKO"

Regardless Jordan still comes out significantly behind Lebron who per DPM is second to two-way-big Duncan(no idea what his WOWY looks like though).

Curiously, like with moonbeam's, Jordan's impact seems to peak during the second three-peat rather than the 80's though I guess with how the stat works it's basically saying Jordan is a big outlier for his age in his 30's which he was.

Looking at the breakdowns of 1985, 1988, and 1992:

Read on Twitter

Seems like Magic has it throughout the 80's and then Jordan comes out tops in the 90's

As per usual you can't read? DARKO WOWY, is not DARKO. And as usual, you speak on stats you don't know anything about. Just only few on this board bother to check you on this.

Furthermore, you clearly have stated MJ's case over Magic is questionable in terms of prime value, no? So would this not serve as a data point in this case to refute this.

[OhayoKD]

As per usual you can't read? DARKO WOWY, is not DARKO.

So it isn't spiking or deflating player performance by checking how players compare relative to expected trajectories?

If I've counted correctly, Jordan trails Magic's trajectory through the end of 1990 and matches a second Magic peak(somehwere between 87-90) as of 1993.

That big spike comes during the second-three-peat where Jordan is aged 32-34(magic was forced to retire at 29 and came back one season at 36).

Now you can go with 96-98 Mj just being >>>>> "prime" MJ(in which case we actually have data-ball for MJ's "peak" which now doesn't look that great), the reason Jordan is spiking in his 30's is because he was unusually good at 30+ and not so unusually good at what would be his typical peak. Actually Magic and Bird also benefit from this(notice how they go up as they age?) though they for different reasons they are not participating during the height of expansion where 30+ players were peaking

You can also see this with Lebron whose trajectory is well ahead of the field during his mid-late 20's is mysteriously upsurped by 30+ Jordan, and then reaches newer heights than everyone else in his 30's. Is this because 30+ Lebron was more valuable than mid-20's Lebron? No. It's because Lebron was a bigger outlier relative to other players.

Additionally because you are relying strictly on WOWY, players are at the mercy of timing. Duncan barely misses games up until 2004 so he doesn't do too well higher. But if we swtich to Darko DPM, Duncan is kicking everyone's ass because Darko now has access to that early stretch where Duncan was already one of the best players ever by the age of 22. The only player who tops Duncan's early trajectory is Lebron but then Lebron has 2011 at an age where Darko expects you to peak and Lebron's trajectory collapses to 2nd best behind the big fundamental.

Darko is a great stat which is very good at what it does. But you have to understand what it does to use it properly. Darko suggests that Jordan was crazy good for a 30+ guy during the second-three peat. It also suggests he was good but not crazy good for a guy in his 20's/mid-20's during the 80's and early 90's.

I know how the stat works, which is why I'm not just looking at which lines are higher on the chart. Lebron at game 1600 was almost certainly not a better player than peak Micheal. And while there's certainly evidence to suggest the second three-peat might have been Jordan's apex in terms of "era-relative situational impact", he was not a waaaay better there than he was in the years some consider the best ever.

[Moonbeam]

I think we should remember that it’s easier to finish at the very top of these in a smaller league. The top players in a lot of time periods for these spreadsheets are actually completely random people who barely played. So them being at the top is obviously just noise. The larger the league, the more low-playing-time players like that you’ll get in a given time period, and the more players like that there are the higher the chances of there being ones that randomly have a really high score. So I wouldn’t equate Bill Russell constantly being at the top as necessarily being better than a guy more recently who isn’t at the very top but is consistently ahead of everyone but random guys. For instance, if we look at the Ridge version, Steph isn’t at the very top in any time period, but as far as I can tell there’s not any particularly meaningful player above him in the last 5 time periods. Same with Steve Nash for like 5 consecutive time periods in the 2000s. Granted, Bill Russell is still the very top meaningful player in *more* timeframes than anyone else—and that strikes me as a great data point for him—but I do think other players have shorter but still significant timeframes where they look similarly good as Russell when we take into account the added amount of random-noise guys in later years.

Yeah, I think the way rotations have changed throughout history have had an impact here. As you say, there tend to be more "Ed Nealys" that seem like interlopers in later time periods due to what is likely greater instability in minutes for players and more player movement leading to more opportunities for "team-seasons" within the 5-year windows.

[Moonbeam]

Here is a spreadsheet with up to 100 positive coefficients for each 5-year window for Ridge, Lasso, and ENet. I'll see if a spreadsheet with the full data is navigable and post separately if so.

A few requests for Moon if possible:

can you create a graph for Ralph Sampson, Hakeem, Rodney Mcray, and Robert Reid, Ridge and Lasso?

And another for Jordan, Pippen, Horace Grant, and BJ Armstrong, Ridge and Lasso?

Here are the 80s Rockets:





Hakeem dominates as expected.

And the 90s Bulls:





Grant looks pretty great here. He did have the benefit of playing for good teams throughout his career, but the fact he was able to be a positive contributor for all of them (also confirmed via RAPM I believe) is certainly a signal that he is one of the better unheralded guys of the era. B.J. Armstrong looks amazing to start, but I think a lot of that is joining a team that took off and was great. Once the Warriors seasons creep into the sample, he plummets as expected.

[Moonbeam]

- New York Knicks
Key players: Walt Frazier, Willis Reed, Dave DeBusschere, Dick Barnett, Earl Monroe, Bill Bradley


[/img]

Okay, so this is interesting.

We have Frazier start out with a lead over Reed that Reed then matches and briefly surpasses. With that early lead I think we should note that the turnaround of the Knicks came mid-season in '67-68 when with Frazier getting put in the game mid-season. By the following post season he was a clear cut star.

How did he get off the bench? New coach named Red Holzman came in and quickly decided that's how things should be, and the team got better.

All this raises a question of whether Frazier's getting the last-piece-of-the-puzzle benefit where whoever joins the core right when the biggest breakthrough happen looks great with regression. If some other guy, like Reed would have had the same last-piece-of-the-puzzle effect we should not give this credence but we'll never truly know.

I think at the least Frazier's better health, despite not having great longevity himself, continues to give him the nod in my book.

Have to say DeBusschere looks a bit more solid than Barnett.

Is Bradley missing?

Yeah, Bradley got cut out. It's because there's another Bill Bradley who played one season only, and averaged fewer than 18 MPG for that season, so he wasn't on the plot.

Here's the updated one with the Knicks' Bradley:

[DraymondGold]

Kind of! There is a bit of potentially useful information lost by looking at percentiles only. The caveat to that is because the ridge models are choosing different penalties for each window, the degree of "outlierness" is not always directly comparable. But yes, Russell and Magic are beasts in these pure versions!

Hmm, so this is interesting. It suggests the raw numbers might not be good to compare across era (at least in ridge, and presumably in the other methods as well?).

If percentiles / rank (adjusted for sample size) are a better way to compare players, perhaps we might look something like:
-Average Percentile at a shorter timescale (e.g. best 5 samples in a row), medium (bets 10 samples in a row), and full-career timescale (every sample) for the standard top set of players (e.g. the list Doc mentioned earlier in this thread for a shorter list, or maybe something like the Top 20 players from last year's + Curry/Durant/Paul/Walton for a larger list)

Or some variant of this methodology, if we wanted to do a more systematic but fair comparison of players across era in this new metric.

...

On another note, I wonder if there are any trends on what the standard deviations are in the ridge values over time.

I checked in the spreadsheet you sent, and it looks like...
-there's a big peak in standard deviation in the 60s
-then it dips until a minor peak around ~90,
-a dip until another major peak around ~06,
-a dip around ~12,
-then the biggest peak is in recent times.
Of course this is only based on the Top 100 players included in the spreadsheet, so the trends across the full player list might be different (e.g. this only includes positive RWOWY players). As for what to interpret from this, I'm not sure. Presumably it relates to the varying penalties for each window? But as for why those trends occur, I'd have to think more. But I'm open to ideas if anyone has any!

[Moonbeam]

Kind of! There is a bit of potentially useful information lost by looking at percentiles only. The caveat to that is because the ridge models are choosing different penalties for each window, the degree of "outlierness" is not always directly comparable. But yes, Russell and Magic are beasts in these pure versions!

Hmm, so this is interesting. It suggests the raw numbers might not be good to compare across era (at least in ridge, and presumably in the other methods as well?).

If percentiles / rank (adjusted for sample size) are a better way to compare players, perhaps we might look something like:
-Average Percentile at a shorter timescale (e.g. best 5 samples in a row), medium (bets 10 samples in a row), and full-career timescale (every sample) for the standard top set of players (e.g. the list Doc mentioned earlier in this thread for a shorter list, or maybe something like the Top 20 players from last year's + Curry/Durant/Paul/Walton for a larger list)

Or some variant of this methodology, if we wanted to do a more systematic but fair comparison of players across era in this new metric.

...

On another note, I wonder if there are any trends on what the standard deviations are in the ridge values over time.

I checked in the spreadsheet you sent, and it looks like...
-there's a big peak in standard deviation in the 60s
-then it dips until a minor peak around ~90,
-a dip until another major peak around ~06,
-a dip around ~12,
-then the biggest peak is in recent times.
Of course this is only based on the Top 100 players included in the spreadsheet, so the trends across the full player list might be different (e.g. this only includes positive RWOWY players). As for what to interpret from this, I'm not sure. Presumably it relates to the varying penalties for each window? But as for why those trends occur, I'd have to think more. But I'm open to ideas if anyone has any!

Great questions! I think the percentiles are the best for cross-era comparison because the penalties do indeed change with each window based on what is deemed best through cross validation. In general, the penalty that is best is a function of how much variation there is in the scoring margins and how much information there is about the impact of players through a WOWY lens (via injuries, changing teams, rotation size, etc.). Eras with a lot of transactions or a lot more injuries would likely have stronger information signals about players and perhaps require less penalization as a result. For the 60s, I'd guess the overlap with the ABA would be one factor, and for recent times, load management means there are more 'Without' games for a lot of players (though I'd imagine there are more players average 18 MPG per team as a result).

[homecourtloss]

Here is a spreadsheet with up to 100 positive coefficients for each 5-year window for Ridge, Lasso, and ENet. I'll see if a spreadsheet with the full data is navigable and post separately if so.

A few requests for Moon if possible:

can you create a graph for Ralph Sampson, Hakeem, Rodney Mcray, and Robert Reid, Ridge and Lasso?

And another for Jordan, Pippen, Horace Grant, and BJ Armstrong, Ridge and Lasso?

Here are the 80s Rockets:





Hakeem dominates as expected.

And the 90s Bulls:





Grant looks pretty great here. He did have the benefit of playing for good teams throughout his career, but the fact he was able to be a positive contributor for all of them (also confirmed via RAPM I believe) is certainly a signal that he is one of the better unheralded guys of the era. B.J. Armstrong looks amazing to start, but I think a lot of that is joining a team that took off and was great. Once the Warriors seasons creep into the sample, he plummets as expected.

Thank you, Moon. I’m trying to think of any other trio that played that many seasons and minutes together do that well together—you have essentially the entire Bulls run with Jordan, Grant, Pippen in the 90th+ percentile, with BJ Armstrong looking very strong early.

[eminence]

Thank you, Moon. I’m trying to think of any other trio that played that many seasons and minutes together do that well together—you have essentially the entire Bulls run with Jordan, Grant, Pippen in the 90th+ percentile, with BJ Armstrong looking very strong early.

The Mikan Lakers looked kind of similar, with Mikan/Pollard/Martin all on top of the league.

Similarly, could I get a recent Warriors graph? (Steph/Dray/Klay/Andre/KD, can't really think of a 6th I'd be that interested in)

[Colbinii]

Any chance we can do the bulls with Grant/Rodman/Kukoc/Purdue?

[Colbinii]

Thank you, Moon. I’m trying to think of any other trio that played that many seasons and minutes together do that well together—you have essentially the entire Bulls run with Jordan, Grant, Pippen in the 90th+ percentile, with BJ Armstrong looking very strong early.

The Mikan Lakers looked kind of similar, with Mikan/Pollard/Martin all on top of the league.

Similarly, could I get a recent Warriors graph? (Steph/Dray/Klay/Andre/KD, can't really think of a 6th I'd be that interested in)

Livingston or Bogut