'22-23 Commentary
Re: '22-23 Commentary
In a contest elsewhere regarding Western Conference ranks (not Wins) with 55 entries, I was the winner at the wire.
Re: '22-23 Commentary
Kings, most used lineup in league at over 900 minutes. Part of their success story.
Nuggets' top lineup 3rd biggest, Sixers 8th, Celtics 10th. All huge parts of their success.
Cavs 13th. Warriors 17th and by far better than any with bigger minutes and better than any in the biggest 50 by minutes. Best still possible in top 153 actually and 2nd best too.
154th on minutes but better than GSW, the new Lakers starters.
Better still but no longer possible, Thunder with Muscala. But nah, Coach D and Presti not much interested in further testing / using that +26pts / 100p. Preferred to put out crap starting lineups.
Houston with Eason (and Martin) as a starter even better in too small a test. Wizards without Beal even better in too small a test.
What is small sample variance and what is great? If you didn't test enough, you don't know and made a bad management error.
Quads tell much the same stories but with bigger samples.
But almost every one in media completely ignores lineup data or might mention it on very rare occasion. Followers / fans usually ignore it too with a few exceptions.
Have to wonder how much time and credence NBA coaches, management and analytic staff give lineup data the way lineup data looks. Sure doesn't look the way I'd want it to look with respect to concentration, far greater concentration on what actually is working. Even the few best team cases could be better in concentration and specific details.
If any team wanted to increase focus on that... they should and I could help.
A 100 lineup season would be a worthy plan. Probably more to far more than my true ideal though. But unlikely to ever happen. 300 is certainly better imo than 500 - 700 (or 985) but still bloated, inefficient and not very wise.
100 lineups would be enough to test almost every combination of 2 guys at two positions and 3 guys at the other 3. Could try 13 total or less with multiple position assignments. Need or want 3, 5, 7 or 10 times as much? I think not.
If you are a good team wanting to be great, this approach is 1 possible way to get there. Average wanting to be good, weak wanting to no longer be weak? Possible way. What do you have with more hope for improvement? And why did that / they not fully satisfy in most cases this season or the last one or the one before that?
Try greater / super concentration and it doesn't work? Fine, quit or put someone new in the process. But don't try it? Then you strike me as rigid and not that wise.
Nuggets' top lineup 3rd biggest, Sixers 8th, Celtics 10th. All huge parts of their success.
Cavs 13th. Warriors 17th and by far better than any with bigger minutes and better than any in the biggest 50 by minutes. Best still possible in top 153 actually and 2nd best too.
154th on minutes but better than GSW, the new Lakers starters.
Better still but no longer possible, Thunder with Muscala. But nah, Coach D and Presti not much interested in further testing / using that +26pts / 100p. Preferred to put out crap starting lineups.
Houston with Eason (and Martin) as a starter even better in too small a test. Wizards without Beal even better in too small a test.
What is small sample variance and what is great? If you didn't test enough, you don't know and made a bad management error.
Quads tell much the same stories but with bigger samples.
But almost every one in media completely ignores lineup data or might mention it on very rare occasion. Followers / fans usually ignore it too with a few exceptions.
Have to wonder how much time and credence NBA coaches, management and analytic staff give lineup data the way lineup data looks. Sure doesn't look the way I'd want it to look with respect to concentration, far greater concentration on what actually is working. Even the few best team cases could be better in concentration and specific details.
If any team wanted to increase focus on that... they should and I could help.
A 100 lineup season would be a worthy plan. Probably more to far more than my true ideal though. But unlikely to ever happen. 300 is certainly better imo than 500 - 700 (or 985) but still bloated, inefficient and not very wise.
100 lineups would be enough to test almost every combination of 2 guys at two positions and 3 guys at the other 3. Could try 13 total or less with multiple position assignments. Need or want 3, 5, 7 or 10 times as much? I think not.
If you are a good team wanting to be great, this approach is 1 possible way to get there. Average wanting to be good, weak wanting to no longer be weak? Possible way. What do you have with more hope for improvement? And why did that / they not fully satisfy in most cases this season or the last one or the one before that?
Try greater / super concentration and it doesn't work? Fine, quit or put someone new in the process. But don't try it? Then you strike me as rigid and not that wise.
Re: '22-23 Commentary
Atlantic and Pacific divisions each had 5 in post season action, as much combined as the other 4 divisions.
How much is willingness to spend in big markets and how much is player location preference? When does the Southeast join them on either or both criteria?
How much is ownership / management?
How much is willingness to spend in big markets and how much is player location preference? When does the Southeast join them on either or both criteria?
How much is ownership / management?
Re: '22-23 Commentary
Sheppard out in Washington.
My most detailed recent comments on page 12 of this thread.
My most detailed recent comments on page 12 of this thread.
Re: '22-23 Commentary
Crow wrote: ↑Sun Mar 05, 2023 9:23 pm So Jordan Goodwin was a big rebounding guard in college who is nothing special rebounding in the NBA for Wizards. And Johnny Davis is the same. And now the Wizards get another big rebounding guard out of Australia. Will he rebound above average, for team impact? Is the team rebounding concern going to be fixed by a guard?
And the Wizards have Avdija, a pretty good rebounder, better than the first 2 at least, who some claim is or can be a guard but rarely plays there for Wizards?
And with current roster, is all this just about who gets minutes as 4th or 5th or 6th guard / replaces K Nunn next season?
Or is it bigger than that?
If Kuzma leaves, does Avdija take his place and Kispert take Avdija's place, creating room for more bench guard play?
Are both Morris and Wright going to be back?
Getting to FT line is also a pretty big issue. Morris, Wright, Goodwin, Davis are all below average on FT rate. Cooks got to FT line more this season but only hit 54% of time. Avdija is above average on FT rate but is fairly low usage. Not likely for team to improve there with these guards.
Goodwin, Davis remain weak rebounders. X Cooks looks very good on offensive glass but just near average on defensive boards if playing PF. Morris, Wright and Barton were weak. And Kispert and Nunn and others.
Overall a weakness.
Some defensive factor strength but not near enough for an average defense or average SRS.
Re-evaluate everything.
Going to pick a GM that believes in greater lineup concentration? Will forcefully encourage lineup concentration? Will insist on lineup concentration to the point of immediate firing for non-compliance? Who will insist the analytic staff make it a top, constant work priority to guide lineup concentration? Who will announce greater lineup concentration to the media, public and peers as an obvious, essential permanent organizational value?
Re: '22-23 Commentary
N Nurse out.
Poeltl with starters really worked. Used 10 minutes per game. Did he / they consider going all in and making it 20 plus min / gm? Probably not. Probably should have, though that would have more clearly indicated a move from / rejection of the heavy "6-9" strategy of past.
Raptors 4 - 13 on Fridays. Probably random (nothing consistent in prior 2 years) but could / should be looked at. Opponents fg% jumped up over average.
Poeltl with starters really worked. Used 10 minutes per game. Did he / they consider going all in and making it 20 plus min / gm? Probably not. Probably should have, though that would have more clearly indicated a move from / rejection of the heavy "6-9" strategy of past.
Raptors 4 - 13 on Fridays. Probably random (nothing consistent in prior 2 years) but could / should be looked at. Opponents fg% jumped up over average.
Re: '22-23 Commentary
Lineup commentary.
https://twitter.com/bballstrategy/statu ... 67233?s=20
In general and on Warriors.
https://twitter.com/bballstrategy/statu ... 67233?s=20
In general and on Warriors.
Re: '22-23 Commentary
Udoka last season had Celtics -8 wins in regular season, actual - expected. Mazzulla this season, even.
Maybe say more about style and luck than coaching quality but that is a huge difference.
Playoff outcome this time for Mazzula will be interesting.
Next year Udoka in Houston will be interesting as well. How good?
Maybe say more about style and luck than coaching quality but that is a huge difference.
Playoff outcome this time for Mazzula will be interesting.
Next year Udoka in Houston will be interesting as well. How good?
Re: '22-23 Commentary
Crafted PM (based on 6 metrics) has 38 players under 24 total who are rated at or above neutral. 25 of those are at or above +1.
So the average has about one player at neutral or beyond.
122 players under 24 according to crafted nba. Or about 4 per team.
About 180 by BRef with about 35 at or above neutral. Or about 6 total per team but only one above the performance threshold. (The count difference is probably related to 10 day contracts, possibly 2ways.)
So about 20-30% of such players are not hurting team results by these metrics. 70-80% are, to some degree.
Young players are not usually performance assets in their youth. Young players are usually performance burdens and sometimes financial burdens.
The draft as a main source of such players is not that much of an immediate value generator.
Unless you have top draft picks, I can see why some teams (especially now teams) de-emphasize the draft as a source of players.
So the average has about one player at neutral or beyond.
122 players under 24 according to crafted nba. Or about 4 per team.
About 180 by BRef with about 35 at or above neutral. Or about 6 total per team but only one above the performance threshold. (The count difference is probably related to 10 day contracts, possibly 2ways.)
So about 20-30% of such players are not hurting team results by these metrics. 70-80% are, to some degree.
Young players are not usually performance assets in their youth. Young players are usually performance burdens and sometimes financial burdens.
The draft as a main source of such players is not that much of an immediate value generator.
Unless you have top draft picks, I can see why some teams (especially now teams) de-emphasize the draft as a source of players.
Re: '22-23 Commentary
6 of 8 oldest teams by playoff minutes are leading or have won first round, though 2 oldest are out. 4 of 5 youngest are behind or out.
5 of 6 best offenses in playoffs are leading or have won. 3 of 5 weakest offenses are out, with one still facing 2 more elimination games.
4 of 5 best defenses are leading or have won. 3 weakest defenses are out.
The 4 teams above average on both sides of the ball are Sixers, Nuggets, Celtics and Warriors.
5 of 6 best offenses in playoffs are leading or have won. 3 of 5 weakest offenses are out, with one still facing 2 more elimination games.
4 of 5 best defenses are leading or have won. 3 weakest defenses are out.
The 4 teams above average on both sides of the ball are Sixers, Nuggets, Celtics and Warriors.
Re: '22-23 Commentary
Knicks most used playoff lineup is -11 pts / 100p. Was -14 in regular season and just 23rd most used. So far this elevation is a very bad call. May look right on paper based on individuals but not together in average reality.
Re: '22-23 Commentary
Griz most used playoff lineup was awful. Mild positive in a half decent 142 regular season minutes but that was not quite a reliable test. Could have been more.
Re: '22-23 Commentary
Hawks in playoffs, top 3 Synder lineups all sucked. Most used was modest positive in big regular season test but even that can fail. Modest positive is not as convincing as strong or super strong.
2nd and 3rd most used were not that in regular season. One was a superstar in a way too small 48 minute rs test. The other was an even smaller test with negative results to boot. Not acceptable testing and bad results. Coaching guesses got wrong.
2nd and 3rd most used were not that in regular season. One was a superstar in a way too small 48 minute rs test. The other was an even smaller test with negative results to boot. Not acceptable testing and bad results. Coaching guesses got wrong.
Re: '22-23 Commentary
Raw +/- at game level is certainly problematic and should probably cite it less even if I still check it.
There would be opportunity to build an alternative weighting individual stat performance with teammates and opponents generally and at matchup level. Maybe something like 25% own, 25% matchup, 25% teammates on court with, 25% other opponents. Better than simple raw, at some computational time & effort.
Game level RAPM would be another alternative. Anybody compute that?
Game level BPM is in BRef advanced boxscores and I should check them but have not been in habit of doing so.
Game Score is also available but not particularly desirable beyond being available.
There would be opportunity to build an alternative weighting individual stat performance with teammates and opponents generally and at matchup level. Maybe something like 25% own, 25% matchup, 25% teammates on court with, 25% other opponents. Better than simple raw, at some computational time & effort.
Game level RAPM would be another alternative. Anybody compute that?
Game level BPM is in BRef advanced boxscores and I should check them but have not been in habit of doing so.
Game Score is also available but not particularly desirable beyond being available.
Re: '22-23 Commentary
I wonder if somebody could build player tracking data informed game level RAPM and spend time fine-tuning it. Or maybe season level complete - season level without the last game? The Drip team could probably do that. Do they? Or analytic gamblers? 5-10 game samples or moving averages?
Play level RAPM might be absurd... unless it was any help for new learning. However many current data points there are now and the exponential increase in the next wave of player tracking, there should be attempts to use all of them as a set in addition to whatever subsets are examined (and perhaps used to some degree?).
Play level RAPM might be absurd... unless it was any help for new learning. However many current data points there are now and the exponential increase in the next wave of player tracking, there should be attempts to use all of them as a set in addition to whatever subsets are examined (and perhaps used to some degree?).