What hasn't been in public basketball analytics to my knowledge?
A lot on lineup management detail.
I have done more than I see elsewhere but there is more that could be done.
Look at a Coach's lineup management:
against opposing lineup types in general and by specific main lineup and by main lineups as a group vs dinks
by quarter, home / road, stronger/ weaker opponent, ahead / behind, first time / next time, within a series
for offensive and defensive results, by factor, for specific stats
identify habits, best and worst. identify best and worst intuitive calls / guesses and their basis. identify where staff or players influenced and results.
LIneup trends around the league
Re: LIneup trends around the league
Da Mitchell, D Smith, Highsmith, Jovic, Ware had a lot of good pairs this season. Time together as a lineup? Zero minutes. I'd want 200+ minutes as soon as possible next season. And variations with 4 of the 5. And 3.
Mitchell Highsmith Ware had the best bigger minute test at 173 minutes for +13.2 / 100p. Highsmith Jovic Ware was next at 78 minutes for +9.6 / 100p.
I'd trade anybody on Heat for a more desirable return and future fit but I'd prefer to keep these pieces and see what they can do with others.
I'd start blowing it up immediately but wait til trade deadline or next summer if they want.
Mitchell Highsmith Ware had the best bigger minute test at 173 minutes for +13.2 / 100p. Highsmith Jovic Ware was next at 78 minutes for +9.6 / 100p.
I'd trade anybody on Heat for a more desirable return and future fit but I'd prefer to keep these pieces and see what they can do with others.
I'd start blowing it up immediately but wait til trade deadline or next summer if they want.
Re: LIneup trends around the league
Best and worst bigger minute playoff lineups?
For 11 most used, Lakers starters with DFS were worst. Clippers 2nd worst. Followed by the dream design of the Knicks.
Best were from the Timberwolves, Pacers and Nuggets.
Thunder with 2 bigs was meh, with Caruso slightly negative, with C Wallace 13th most used but far better than any bigger. A non-SGA lineup was 14th most used and 3rd best to that level.
Rockets had a strong big minute playoff lineup and blew it up. Probably right but will they actually do better next season?
Pistons had a very good lineup with 12th most minutes. All parts returning? Depends on Hardaway.
Warriors starters very good in 16th most use.
Knicks 2nd and 3rd most used lineups also negative. Should re-examine design and assumptions. 4th most used with McBride and Robinson over expensive acquisitions of Towns and Bridges was really great... in 40 minutes. 5th most used modifying the 4th with Bridges over Hart was even better.
Celtics starters even with Tatum performed awful in 25th most minutes.
Lakers starters with Hayes was 2nd worst among 50 most used. TWolves starters with DiVincenzo was the worst. Switch out Gobert for Reid and it became one of the best.
Small sample sizes and particular matchups influence results but these are the results, to be considered or basically ignored.
Playoffs generally won or lost by the results of the 5 most used lineups. Sometimes it is one lineup, sometimes it is the grab bag of dink lineups. But the set of lineup results is the result and are important, whether carefully planned or picked in spur of moment.
Of the 11 playoff lineups with 50+ minutes and +5 or better results per 100p, they mostly shot the 3 well but less often than playoff average and mostly got to FT line less often. Success was not much based on Moreyball. Rebounding and assist distributions were pretty average. Steal rates below playoff average for all such lineups except Thunder with Caruso and Wallace.
What is notable about the lineup designs? None involved 2 true traditional bigs. Only a few had 2+ players with 3 yrs or less experience. All but one had a PG with more than 3 yrs experience. A majority had a big scoring wing. Own turnovers were elevated for only a few.
For 11 most used, Lakers starters with DFS were worst. Clippers 2nd worst. Followed by the dream design of the Knicks.
Best were from the Timberwolves, Pacers and Nuggets.
Thunder with 2 bigs was meh, with Caruso slightly negative, with C Wallace 13th most used but far better than any bigger. A non-SGA lineup was 14th most used and 3rd best to that level.
Rockets had a strong big minute playoff lineup and blew it up. Probably right but will they actually do better next season?
Pistons had a very good lineup with 12th most minutes. All parts returning? Depends on Hardaway.
Warriors starters very good in 16th most use.
Knicks 2nd and 3rd most used lineups also negative. Should re-examine design and assumptions. 4th most used with McBride and Robinson over expensive acquisitions of Towns and Bridges was really great... in 40 minutes. 5th most used modifying the 4th with Bridges over Hart was even better.
Celtics starters even with Tatum performed awful in 25th most minutes.
Lakers starters with Hayes was 2nd worst among 50 most used. TWolves starters with DiVincenzo was the worst. Switch out Gobert for Reid and it became one of the best.
Small sample sizes and particular matchups influence results but these are the results, to be considered or basically ignored.
Playoffs generally won or lost by the results of the 5 most used lineups. Sometimes it is one lineup, sometimes it is the grab bag of dink lineups. But the set of lineup results is the result and are important, whether carefully planned or picked in spur of moment.
Of the 11 playoff lineups with 50+ minutes and +5 or better results per 100p, they mostly shot the 3 well but less often than playoff average and mostly got to FT line less often. Success was not much based on Moreyball. Rebounding and assist distributions were pretty average. Steal rates below playoff average for all such lineups except Thunder with Caruso and Wallace.
What is notable about the lineup designs? None involved 2 true traditional bigs. Only a few had 2+ players with 3 yrs or less experience. All but one had a PG with more than 3 yrs experience. A majority had a big scoring wing. Own turnovers were elevated for only a few.
Re: LIneup trends around the league
4 of the 10 most used playoff lineups are already no longer possible and a 5th is likely.
14 of 20 most used regular season lineups are no longer possible.
The 30th biggest regular season lineup is the equivalent of 2 minutes per game. The 60th biggest regular season lineup (essentially about the smallest 2nd biggest lineup for a team) averages the equivalent of about 1.5 minutes per game across the season. By the time you get to the smallest of the 5th most used lineups, the time of use falls to 1 minute per game. 10th and it is barely half a minute per game.
The 10 most used lineups will cumulatively receive near 1/3rd of available regular season time, give or take some by specific team.
14 of 20 most used regular season lineups are no longer possible.
The 30th biggest regular season lineup is the equivalent of 2 minutes per game. The 60th biggest regular season lineup (essentially about the smallest 2nd biggest lineup for a team) averages the equivalent of about 1.5 minutes per game across the season. By the time you get to the smallest of the 5th most used lineups, the time of use falls to 1 minute per game. 10th and it is barely half a minute per game.
The 10 most used lineups will cumulatively receive near 1/3rd of available regular season time, give or take some by specific team.
Re: LIneup trends around the league
Average playoff team apparently had 3.5 guys performing at +1 or better on BPM, another one over neutral and another 1.5 over -1. Generally a core 6, though varied by team.
Among those who performed below this standard: Conley, Alexander-Walker, Jalen Green, Brooks, Westbrook, M Porter, Garland, Finney Smith, Dort, Draymond Green, KCP, Jaren Jackson, Reaves, Porzingis, Hardaway, Andrew Wiggins and Kuzma. Half are already elsewhere.
Among those who performed below this standard: Conley, Alexander-Walker, Jalen Green, Brooks, Westbrook, M Porter, Garland, Finney Smith, Dort, Draymond Green, KCP, Jaren Jackson, Reaves, Porzingis, Hardaway, Andrew Wiggins and Kuzma. Half are already elsewhere.
Re: LIneup trends around the league
Micro-managing lineups for specific player matchups might make sense if there is past data to support going for that matchup and getting usage & results or having such data against a general player type. But coaches guided by actual solid data or memory & hunches?
Data is not readily & reliably available in public but could be assembled with video work by team or consultants.
Apart from specific data analysis, can check coaches on percent of dink lineups won. Need to find the average performance and / or a high standard to compare against.
In a quick check of 17 teams, Thunder and Celtics lead with near 50% dink winners. Playoff teams appear to be over 40%, lottery under. This of course is a function of player talent and lineup construction. Wizards were lowest of those checked at 33% winners. Weak team tens to lose the dink game more and probably ought to emphasize concentration more than usual. But probably don't.
Data is not readily & reliably available in public but could be assembled with video work by team or consultants.
Apart from specific data analysis, can check coaches on percent of dink lineups won. Need to find the average performance and / or a high standard to compare against.
In a quick check of 17 teams, Thunder and Celtics lead with near 50% dink winners. Playoff teams appear to be over 40%, lottery under. This of course is a function of player talent and lineup construction. Wizards were lowest of those checked at 33% winners. Weak team tens to lose the dink game more and probably ought to emphasize concentration more than usual. But probably don't.
Re: LIneup trends around the league
Will want to see how deep teams play into their depth. Wonder if there is a recent trend or notable movement next season.
Re: LIneup trends around the league
If interested:
In this playoffs, 1210 lineups were used. On average for 7 minutes total.
On average teams played 11 games and 76 lineups. Or 7 brand new in the playoffs every game. Slightly more than half on average had 1 lineup used over 100 minutes. Then only 5 of those 9 got used another 100+ minutes after the modest first 100 minutes "test". They had less than 4 over 20 minutes or 5% of total lineups. 88% of lineups were used 10 minutes or less total. 74% for 5 minutes or less.
Playoffs almost entirely dink on average vs. dink on average lineup (though not necessarily dink for particular selected games). When it wasn't, it almost always involved near dinks on average. Teams averaged 7 of 76 lineups used in more than 5 games (or less than half of average games played).
54% of 11 most used were positive. 50% of 50 most used. In these groups, none were neutral. Biggest 50 coaching selections did not beat a random division of positives & negatives. For all lineups, 39.5% were positive, 14.5% neutral, 46% negative. The biggest 50 couldn't beat random division but it did beat the average performance of the 1160 lineups used under 24 minutes total or barely 2 minutes per game on average across average amount of games. Bigger, more concentrated lineup use was better.
Thunder remained close to the typical 7 brand new lineups per game. Just slightly less than 50% of lineups were positive. Good enough that way. Barely. 2 lineups clearly beyond 3 minutes per game on average. Just 6 of 155 lineups used in half or more of games. Not much more than 20% were used in more than 3 games. 96% of lineups qualify as dink on average (less than 1 min / gm across all games). More than average.
That's the way they do it.
Plenty could / "should imo" have been different to emphasize bigger / better performing lineups more. But nah, pretty much just 'dink it' with a few exceptions.
In this playoffs, 1210 lineups were used. On average for 7 minutes total.
On average teams played 11 games and 76 lineups. Or 7 brand new in the playoffs every game. Slightly more than half on average had 1 lineup used over 100 minutes. Then only 5 of those 9 got used another 100+ minutes after the modest first 100 minutes "test". They had less than 4 over 20 minutes or 5% of total lineups. 88% of lineups were used 10 minutes or less total. 74% for 5 minutes or less.
Playoffs almost entirely dink on average vs. dink on average lineup (though not necessarily dink for particular selected games). When it wasn't, it almost always involved near dinks on average. Teams averaged 7 of 76 lineups used in more than 5 games (or less than half of average games played).
54% of 11 most used were positive. 50% of 50 most used. In these groups, none were neutral. Biggest 50 coaching selections did not beat a random division of positives & negatives. For all lineups, 39.5% were positive, 14.5% neutral, 46% negative. The biggest 50 couldn't beat random division but it did beat the average performance of the 1160 lineups used under 24 minutes total or barely 2 minutes per game on average across average amount of games. Bigger, more concentrated lineup use was better.
Thunder remained close to the typical 7 brand new lineups per game. Just slightly less than 50% of lineups were positive. Good enough that way. Barely. 2 lineups clearly beyond 3 minutes per game on average. Just 6 of 155 lineups used in half or more of games. Not much more than 20% were used in more than 3 games. 96% of lineups qualify as dink on average (less than 1 min / gm across all games). More than average.
That's the way they do it.
Plenty could / "should imo" have been different to emphasize bigger / better performing lineups more. But nah, pretty much just 'dink it' with a few exceptions.
Re: LIneup trends around the league
Coaches and lineup management. Coach Carlisle.
https://x.com/bballstrategy/status/1952742086270435757
https://x.com/bballstrategy/status/1952742086270435757
Re: LIneup trends around the league
EVERY negative player pair used 400+ minutes in playoffs was for the Knicks. There were 9 of them. The only positive Knicks player pair of that size was their 10th biggest, Hart - Towns at +3.7.
Have to get to at least 2nd round to make this list but Thibodeau generally picked / managed ultimately mild negatives. Bridges in the 2 worst in 50 most used across the playoffs beside Hart and Anunoby.
Against Pacers only 2 of 16 most used pairs were positive. Towns with Anunoby and Hart. All 7 Brunson pairs over 60 minutes were negative. 3 of the 4 worst of that size involved Hart and same for Robinson.
Knicks probably need massive overhaul of lineup management and probably of roster to go further. Thibodeau gone, roster remains mostly intact (with the addition of Yabusele). For now. What will Brown do to lineup management?
Looking at the playoff data suggests more of Towns with 2 of Brunson, Hart and Anunoby but not all 3. Towns Hart and Brunson or Bridges were best in 11 most used but only near neutral.
Some Robinson and McBride trios did well but they also had some of the worst. Hart - Bridges was in 5 of 6 worst quads in 20 most used. Brunson had some good quads with McBride, Robinson or both. Hart Bridges Anunoby was the worst trio among 12 most used.
Optimal lineups and rotation based on regular season data, pre Pacers series and retrospective would take more work, might be different but should be reviewed and considered. Bottomline, probably very different from actual.
Towns generally looks comparatively favorable in the above notes but the 2 by far best lineups in 7 most used replaced him with Robinson and Bridges or Hart with McBride. Everything but the 5 starters together was small sample, so the actuals may not be predictive but that is what is available to consider.
Spend 5 hours studying or 50+? It "should" be the latter. Knicks have had almost 6 weeks to evaluate. What have they concluded? This is what I found in first hour.
Found 1 podcast suggest Towns - Robinson would be a big deal going forward. 9 minutes of that pair per game in playoffs. It worked. Became part of starting lineup in game 3 against Pacers.
Barely tested in playoffs before then and did poorly. Against Pacers it was slightly negative. Other lineups with that pair did better.
Will wait n see how important / good the par and lineup versions are under Brown.
Overall it was Robinson's 2nd best playoff pair. 3rd best for Towns.
Have to get to at least 2nd round to make this list but Thibodeau generally picked / managed ultimately mild negatives. Bridges in the 2 worst in 50 most used across the playoffs beside Hart and Anunoby.
Against Pacers only 2 of 16 most used pairs were positive. Towns with Anunoby and Hart. All 7 Brunson pairs over 60 minutes were negative. 3 of the 4 worst of that size involved Hart and same for Robinson.
Knicks probably need massive overhaul of lineup management and probably of roster to go further. Thibodeau gone, roster remains mostly intact (with the addition of Yabusele). For now. What will Brown do to lineup management?
Looking at the playoff data suggests more of Towns with 2 of Brunson, Hart and Anunoby but not all 3. Towns Hart and Brunson or Bridges were best in 11 most used but only near neutral.
Some Robinson and McBride trios did well but they also had some of the worst. Hart - Bridges was in 5 of 6 worst quads in 20 most used. Brunson had some good quads with McBride, Robinson or both. Hart Bridges Anunoby was the worst trio among 12 most used.
Optimal lineups and rotation based on regular season data, pre Pacers series and retrospective would take more work, might be different but should be reviewed and considered. Bottomline, probably very different from actual.
Towns generally looks comparatively favorable in the above notes but the 2 by far best lineups in 7 most used replaced him with Robinson and Bridges or Hart with McBride. Everything but the 5 starters together was small sample, so the actuals may not be predictive but that is what is available to consider.
Spend 5 hours studying or 50+? It "should" be the latter. Knicks have had almost 6 weeks to evaluate. What have they concluded? This is what I found in first hour.
Found 1 podcast suggest Towns - Robinson would be a big deal going forward. 9 minutes of that pair per game in playoffs. It worked. Became part of starting lineup in game 3 against Pacers.
Barely tested in playoffs before then and did poorly. Against Pacers it was slightly negative. Other lineups with that pair did better.
Will wait n see how important / good the par and lineup versions are under Brown.
Overall it was Robinson's 2nd best playoff pair. 3rd best for Towns.
Re: LIneup trends around the league
Thunder played as much as anyone and far more than anyone beyond Pacers; but only had 9th most used lineup and it wasn't the typical starting unit and it wasn't positive. Not an impressive lineup management note.
They did have the 3 most used pairs though and they were 3 of the 4 best in the 6 most used. SGA with Dort, Williams and Holmgren. Very strong with Caruso and Wallace. With Hartenstein was only 37th most used and the tiniest bit negative. Reached the goal but this pair go5 and deserves further reflection.
Of 20 most used pairs, 12 were positive, 8 negative. There is reason to expect close to even but still not that impressive.
Centers slightly under-represented, PGs mildly over-represented. SGA - Holmgren was the best in group, a PG - Center pairing. At least 5 of best 6 involved a PG or Center. Nembhard is in the modestly ambiguous case.
12 most used trios all involved PGs but only 7 were positive. 9 of 15 most used had a center. 6 positive, 3 negatives were all for Knicks. 7 of 12 most used had both. 5 of 7 positive. 3 of 4 conference finalists had such a combo. Timberwolves did not. Centers a strength for Timberwolves in some perspectives and they have some positives with one or other of PG & C in 50 most used, but none with both and none in 20 most used. Will they change at PG or find a trio with both that works?
They did have the 3 most used pairs though and they were 3 of the 4 best in the 6 most used. SGA with Dort, Williams and Holmgren. Very strong with Caruso and Wallace. With Hartenstein was only 37th most used and the tiniest bit negative. Reached the goal but this pair go5 and deserves further reflection.
Of 20 most used pairs, 12 were positive, 8 negative. There is reason to expect close to even but still not that impressive.
Centers slightly under-represented, PGs mildly over-represented. SGA - Holmgren was the best in group, a PG - Center pairing. At least 5 of best 6 involved a PG or Center. Nembhard is in the modestly ambiguous case.
12 most used trios all involved PGs but only 7 were positive. 9 of 15 most used had a center. 6 positive, 3 negatives were all for Knicks. 7 of 12 most used had both. 5 of 7 positive. 3 of 4 conference finalists had such a combo. Timberwolves did not. Centers a strength for Timberwolves in some perspectives and they have some positives with one or other of PG & C in 50 most used, but none with both and none in 20 most used. Will they change at PG or find a trio with both that works?
Re: LIneup trends around the league
I have a first rough draft of a 9 lineup rotation for Knicks. Will finetune in fall.
Could be expanded to 15-25 main lineups then whittled down. Some dinks might fit circumstances but I certainly would not go with 500+.
Could be expanded to 15-25 main lineups then whittled down. Some dinks might fit circumstances but I certainly would not go with 500+.
Re: LIneup trends around the league
Knicks traditional starters were almost +4 / 100p in regular season but just +1.4 against Pacers. Then -4 in first two rounds of playoffs. But Thibodeau still went for those starters for 26 minutes in game 1 against Pacers and got crushed by -30 / 100p. Would that be enough to bring change? Nope. Game 2 it gets crushed by -37 / 100p. Minutes reduced but too late. Change in Game 3 is typical, typically too late.
Things were leaning against that lineup before the conference finals. Unlikely to change Thibodeau before even he acknowledged it wasn't working but that was the task. And likely a big part of the reason he was fired.
Will front office be more hands on / demanding about lineup management with Brown? Time will slowly tell
Things were leaning against that lineup before the conference finals. Unlikely to change Thibodeau before even he acknowledged it wasn't working but that was the task. And likely a big part of the reason he was fired.
Will front office be more hands on / demanding about lineup management with Brown? Time will slowly tell