NBA is weak?

[Mike G]

Reading NY Times online, I was drawn to a John Hollinger column with a questionable headline.
https://theathletic.com/5205999/2024/01 ... henderson/

I can't find the Times header now, but it was along the lines of: The NBA is weaker than ever.
Well now I can't read the article. Basically it says the existence of 5 very uncompetitive teams -- Det, SAS, Por, Was, Cha -- is an indication of weakness across the league.

And I was thinking just the opposite: Teams with plenty of talent and decent (or better) coaching are almost unable to win a game due to the extremely high level of competiton.

Is this difficult or impossible to prove one way or another?

For openers, I will mention that my Improved Players sheet shows that returning players are averaging about 4% less eWins per minute than last year. Other rates (PER, WS, BPM) may concur.

[Mike G]

Here's a quick look at players whose health and availability have increased this year over last year; and the bottom of the list, ranked by estimated eWins added (or lost) due to minutes change.
Stats are from a few days ago, when the season was half over; so I just doubled this year's minutes for the Min+ projection.

Code: Select all

min+   eW+     Player           min+    eW+    Player
1669   6.5   Karl-A Towns      -1312   -4.8   Ja Morant
1054   5.0   Anthony Davis     -1738   -4.3   Mason Plumlee
1172   4.6   Zion Williamson   -1305   -3.9   Evan Mobley
2082   4.4   Cade Cunningham   -1476   -3.5   Trey Murphy 
716    4.0   Giannis           -1495   -2.9   Draymond Green

836    3.5   Kevin Durant       -791   -2.7   CJ McCollum
1048   3.2   Brandon Ingram    -1087   -2.5   Darius Garland
904    3.2   Tyrese Maxey       -807   -2.3   Tari Eason
1141   3.2   Goga Bitadze      -1024   -2.3   Zach LaVine
491    3.1   Nikola Jokić       -769   -2.3   Tyler Herro

min+   eW+     Player           min+    eW+    Player
726    3.0   LeBron James       -917   -2.3   Anfernee Simons
738    2.9   Kawhi Leonard      -513   -2.0   Kyrie Irving
903    2.9   Collin Sexton      -918   -1.8   Wendell Carter
1611   2.8   Duncan Robinson    -240   -1.7   Joel Embiid
1318   2.7   Coby White        -1252   -1.6   Markelle Fultz

793    2.6   Jusuf Nurkić       -475   -1.4   Nic Claxton
705    2.5   Paul George        -586   -1.4   Robert Williams
973    2.4   Khris Middleton    -356   -1.3   Bam Adebayo
713    2.4   Desmond Bane       -298   -1.3   Donovan Mitchell
862    2.3   Dario Šarić        -597   -1.3   Bradley Beal

min+   eW+     Player           min+    eW+    Player
898    2.2   Michael Porter     -343   -1.2   Jamal Murray
1034   2.1   Devin Vassell      -607   -1.2   Thomas Bryant
632    2.1   Day'Ron Sharpe     -747   -1.2   Marcus Smart
340    2.1   Shai               -723   -1.1   Ben Simmons
669    2.0   Jaren Jackson      -448   -1.1   Trendon Watford

519    2.0   Andre Drummond     -822   -1.1   Jordan Nwora
565    1.8   Damian Lillard     -749   -1.1   Caleb Martin
908    1.8   Obi Toppin         -339   -1.0   Franz Wagner
1109   1.8   Payton Pritchard   -467   -1.0   Larry Nance Jr.
884    1.8   Cam Thomas         -312   -1.0   Jimmy Butler
515    1.8   Stephen Curry      -455   -1.0   Deandre Ayton

It's not that most of the gainers are really improved on a per minute basis; their gain in projected contribution is simply from better health so far. Losers likewise.
Given that there is likely some kind of selection bias at work, it certainly looks like more competitive players have been on the floor this season.

[Crow]

"Five teams sport scoring margins of minus-9.0 per game or worse this season. In the last five seasons, only four teams total have been that bad, and three of them barely crept past the threshold."

He also talks of the 5th worst being far worse than usual. And how a number of the 5's few wins are against each other, how many of the losses were blowouts and the botton dropping earlier than usual.


Hollinger goes on to note 17 teams with a winning record (18 today). That, not noted by Hollinger, is highest end in last 5 years but was matched several times. 8 projected 50+ win teams would also be a recent top but matched.


I'd say it is mostly about the bottom and some about the top. No acknowledged super prize in lottery but bad teams still want the most pingpong balls (perhaps because draft is still vague)... and don't know how to win, with bad Coaches and / or GMs (and owners) playing substantial roles.

[v-zero]

Wins are zero sum. You can't assess the talent in the league by looking at its worst teams. Imagine a league in which all but five teams are made up of clones of 28 year old LeBron, and the remaining teams are made up of normal NBA talent. Those five teams would look even worse than these five, yet it would be impossible to argue that the league wasn't the most talented it has ever been by an unfathomably huge margin.

The league is currently offensively geared. Coaches and front offices are finally all aligned on the fact that three at 35%+ is indeed better than two at 45%+. Teams are mostly agreed that moving the ball through the air is better than moving your feet to move the ball. Entire varieties of player have become extinct due to these glaringly obvious innovations.

Offensively, at the very least, the league has probably never been this good before, and offensive basketball in the NBA has probably never been played so close to its limits of theoretical efficiency. All the low hanging fruit has finally been picked.

[Crow]

League average 3pt fg% at .366 = .549 efg%. League average 2pt fg% is .544. Close but arguably 3pt fga rate should go a little higher.

Last season, league average 3pt fg% was at .361 = .5405 efg%. League average 2pt fg% ws .548. Close but arguably 3pt fga rate looked like it should go a little lower.

3pt fga rate season to season went from .387 to .393 and has been rewarded.
3pt fg% has some volatility. Up from last season but only 4th best ever. I assume by moving average it is going up over time, so it was probably correct to think it would increase this season. All-time high is .367. Is that the limit or will it be exceeded?


Would be interesting to look by top 10 teams vs rest of league, now and over time.



If the 3pta over 2pta advantage has essentially been played out, it is logical for players and teams to seek to up the fta rate. Hopefully they don't succeed long-term. Fta rate went down from .266 last season to .258 but it had gone up 2 previous seasons.

If the 3pta over 2pta advantage has essentially been played out, it would also be logical to turn more attention to improvement in lineup efficiency.

LIneup efficiency trends over time analysis would take several other more comprehensive measures but it appears that the frequency of lineups over 700 minutes for season has generally been going down over last 15 years. Frequency of lineups over 200 minutes is also down. Frequency of lineups over 50 minutes was also drifting down mildly from 2007-08 (first records) thru 2018-19, then experienced about a 25% fall off in 2019-20. Moving back up some recently but not near levels of 2018-19 and before.

[Mike G]

The last 8 seasons have the 8 lowest TO% of all time, this year being lowest ever (12.1). The 55- year median is 14.1
Last 5 years are the 5 best FT% years ever, this year being the best yet (.785). The 78-year median is .751
Last 9 years are 9 best eFG% seasons, in almost yearly improvement. Now .546 (best ever) vs median .482

ORtg last year broke the previous mark by 2.8; this year only a pt/100 better than that at 115.7 -- 50 year median 106.8
Again last 8 years are top 8, increasing yearly.

No year has seen more 3FGA AND a higher 3fg%
2FGA are more likely to create FTA, and .785 is still better than .545

[DSMok1]

John's point, if you read that article, is the bottom of the league has never had this many really, really bad teams. Not that the overall level of the NBA is bad.

That speaks more to an imbalance in the talent across the NBA than anything else. A high standard deviation in team quality. Hopefully the teams on the bad end are accumulating very young talent that will turn them into good teams eventually.

Overall talent level in the league itself cannot be derived by looking at the league alone. There is no reference point. All we can say from the league itself and year-to-year deltas is whether the deltas are higher or lower than average year-to-year deltas. In other words, we can show the shape of the year-to-year change curve, but we cannot derive the linear component (i.e. steady improvement).

[Mike G]

"To review, the league has arguably never been this bad, this broadly."

That's a funny way of saying the bottom is really bad.
And that's the line that seems to have become the headline that appeared in the NY Times.

There are 426 players who played last year and have returned this season. Projecting their minutes to double this year, and simply averaging their summary stats, last year and this year, the majority have declined -- in PER, WS/48, BPM, and my own eW/484.

Here they are summarized by age, in 2-year blocks that overlap. Most players are included twice -- e.g., the 19 to 21 group includes those who are now 21, as does the 20 to 22 group. These are the avg differences from last year to this year, with minutes this year being doubled (since we are at mid-season).

Code: Select all

Age to    #   min   e484    PER   WS/48    BPM
19  21   40    96   -.04   -0.2   -.003    0.1
20  22   62    59   -.06   -0.6   -.016   -0.4
21  23   69    35   -.01   -0.3   -.003   -0.1
22  24   84   -82   -.12   -1.1   -.006   -0.5
23  25   97   -93   -.17   -1.8   -.029   -1.5

24  26   86   -25   -.11   -1.2   -.024   -1.1
25  27   69   -16   -.04   -0.2    .001   -0.0
26  28   56    -4   -.10   -0.6   -.001   -0.2
27  29   50    96   -.13   -0.9   -.007   -0.4
28  30   42   152   -.07   -0.6   -.003   -0.2

Age to    #   min   e484    PER   WS/48    BPM
29  31   41    35   -.12   -1.0   -.011   -0.7
30  32   35  -131   -.14   -1.1   -.015   -0.8
31  33   25  -258   -.19   -1.5   -.013   -0.8
32  34   20  -159   -.21   -1.7   -.011   -1.0
33  35   22   -33   -.17   -1.3   -.005   -0.9

34  36   19    40   -.05   -0.4    .007    0.1
35  37    9    38    .12    1.2    .020    1.2
36  38+   9  -126   -.10   -0.7    .004   -0.3

3rd column is # of players in the age group.
While minutes are up and down, proficiency is almost universally down for all age groups.
Not included are players who didn't even make a team this season.

What would account for such overwhelming avg dropoff, other than increased competition?

[Mike G]

EDIT: Sorry, I had some nonsense numbers up here earlier. It looked as if the whole NBA is 'below average'.

Here are the individual age groups' minutes-weighted rates, last year and this year, for those still in the league:

Code: Select all

Age to    #  23min   e484    PER   WS/48   BPM    24min   e484    PER   WS/48   BPM
19  20   13   1058    .54   11.2   .040   -3.3     1195    .62   11.6   .053   -2.6
20  21   27    928    .76   13.0   .046   -2.3     1005    .89   13.7   .065   -1.5
21  22   35   1128    .92   14.4   .080   -1.0     1174   1.06   15.4   .086   -0.3
22  23   34   1265    .85   14.0   .098   -0.5     1289    .95   14.6   .102   -0.1
23  24   49   1155   1.07   15.5   .099   -0.1     1020    .94   14.9   .094   -0.4

24  25   47   1304   1.04   15.7   .115    0.4     1275   1.05   15.6   .114    0.5
25  26   39   1139   1.02   15.2   .108   -0.4     1118   1.04   15.4   .112    0.0
26  27   30   1621   1.10   15.7   .117    1.1     1612   1.10   15.8   .116    1.0
27  28   26   1562   1.19   16.3   .117    0.8     1563   1.03   15.0   .106    0.1
28  29   24   1523   1.29   17.0   .123    1.6     1722   1.29   17.3   .128    1.7

Age to    #    Min   e484    PER   WS/48   BPM      Min   e484    PER   WS/48   BPM
29  30   18   1377    .87   14.2   .098   -0.2     1466    .86   14.0   .104   -0.5
30  31   23   1381   1.07   15.7   .114    0.4     1374    .99   15.2   .110   -0.1
31  32   12   1502    .84   13.9   .099   -0.1     1135   1.07   15.8   .120    1.1
32  33   13   1778   1.20   16.5   .116    1.3     1622   1.06   15.4   .104    0.4
33  34    7   1676   1.38   18.8   .145    2.5     1513   1.16   17.0   .133    1.2

34  35   15   1200   1.08   15.7   .107    1.1     1227   1.05   15.6   .108    1.1
35  36    4    898    .70   13.5   .113   -0.3      985   1.03   15.2   .161    2.0
36  37    5   1143    .61   11.8   .109    0.1     1140    .61   11.9   .119    0.2
37  39    4   1561   1.06   15.5   .116    1.7     1279   1.34   17.6   .128    3.2

Last year was unusually great for 23 yr olds -- Poole, Garland, Doncic, Quickley, Herro, Dort, Claxton, Simons... -- most of whom have dropped off this year.

https://www.espn.com/video/clip/_/id/39367251

[Crow]

How do the vast majority of players get worse on individual metrics overall while offensive efficiency is up at team level? Defensive metrics at individual and team level of course get worse but why does offense and defense not balance out? Are metrics in general or in particular cases systematically under-rewarding individual offense or over-penalizing defense?
Are the stat coefficients getting outdated? Not precise enough to begin with? Something with the team adjustment or specifically with the garbage time adjustment?

If these or broader age bands were used, how would the quartiles of them by performance quality look? Is virtually every quartile in every age band getting worse or is it 1 or 2 in particular, differently here and there among the age bands?


Is the individual drop-off mainly about a modest average age increase? How much of that age increase is due to increase retention, playing older guys now compared to past relatively more than young players or average age change among entering players?

[Mike G]

See glaring EDIT above.

An influx of great rookies could impact the veterans' stats, esp. if not so many good players are retiring.
In this case -- see the 2nd post in this thread -- it seems mostly due to better health / more minutes by a bunch of star players. Towns, Giannis, Davis, Durant, LeBron, Kawhi, Zion ... a lot of guys who missed a bunch of games last year but are staying on the court this year make life more difficult for their opponents.

Back when there were 10 teams, Wilt Chamberlain missed almost a whole season. Everyone else became a "better" rebounder that year, in the stat sheet. Really though, league rebounding was "weaker".
When Wilt returned next season, everyone else looked "worse".

[Crow]

So players on average get substantially better from 19 to 22 (among other things)... but the bias in drafting is to not take 22 years in favor of 19-21 year olds.

Individuals and averages are different things but there is a surface contradiction.

Improvement is probably based on age and NBA experience. A more accurate model would need to factor both variables.

[DSMok1]

So players on average get substantially better from 19 to 22 (among other things)... but the bias in drafting is to not take 22 years in favor of 19-21 year olds.

Individuals and averages are different things but there is a surface contradiction.

Improvement is probably based on age and NBA experience. A more accurate model would need to factor both variables.

I did an analysis on this a long while ago and found age dominated over NBA experience to the point that I never included NBA experience anymore after that analysis. Not sure where that study is, though.

[Crow]

Thanks for that information.