APBRmetrics

FWIW, Lakers have the 6th easiest SoS so far in my power ratings which adjust for HCA. BR's SRS ratings don't adjust for HCA.

Does a coach have to have more than his 15 most used player pairs negative, 90% of his most used trios and 85% of his top quads to be fairly critized as not having figured it out?

If neutral is too high of a standard (and ok, maybe it is) then I see 11 of his 15 most used player pairs are worse than -4 per 100 possessions and 10 of the 16 top trios. That standard is a bad team standard and most his FIRST AND BIGGEST sub lineup choices fail to beat that standard.

Has he figured out right pace? Unclear or probably not. How to make Ball work as an individual or team general? Heck no. Whether to keep KCP? Who to start at SF? Whether to keep Randle or Lopez? How to get and hit more 3s? What has he figured out about the now or the direction? Kuzma, Nance. Maybe more but they'll probably throw some of the other ok stuff anyway, so I hesitate to give much credit for them.

To get to .500 they need Ball and Ingram to play like average / veteran starters. How long will that take? 2-3 months, 2-3 years, longer or does it ever happen?

There is usually 2 sides to a story. Sometimes I try to tell and consider both. Other times I may simplify and mainly tell one, the dominant one. I'll read points on the other side, maybe get around to making them myself if no one else does. But I start somewhere. As do the mostly or all positive puff pieces that I react to and try to counterbalance.

74% of all player pairs used over 50 minutes are negative. 65% of all over 25 minutes. Both pretty bad. And that also says the lineups used 25-50 minutes are more positive on average than the ones used over 50 minutes. Maybe that is expected for small and less small samples of a bad team if everything is bad. But also does not look figured out or anywhere close to me.

Sixers, first 20 games: 4 of top 9 most used lineups (used over 1.5 minutes per game) were great, 3 in between, 2 terrible. Since then: 4 great, 2 in between, 3 terrible. First 20 games: 33% of lineups used cumulatively positive. Since then: 30%. So slippage at both bigger minute and dink lineup levels. Concentration on most used 3 lineups has gone down from first time period to second. Lots to examine / probably change there.

Irving-Brown-Tatum-Horford with Smart in first 25 games was almost plus 10 per 100 possessions. Since then -21 / 100. Minutes per game down some, but lineup is costing them 1 pt per game in recent period.

Spurs, first 25 games: 2 of top 6 most used lineups bad. Since then: 4 of the top 6 horrible.

Heat, first 25 games: 2 of top 6 most used lineups horrible. Since then, just 1.

Magic, 5 of top 6 negative in both periods.

Surely partly random but 1 slipped, 1 improved and 1 stayed the same (by these criteria).

Team with top 4 most used lineups negative and 7 of top 9 negative. Coach, well regarded, called "great" on a podcast recently. I'm pretty sure it was Jeff Van Gundy. He definitely said it last season. Team in 10th place in conference. 40% win for season and 30% in last 10. Who is it? The Jazz and Quinn Snyder.

Are all 20 of the coaches of teams ahead of him great or better than great? Are the coaches only a few games behind good or very good? Is any coach average or worse than average if Snyder is great?

Maybe they are nearly all good, compared to the alternatives? They are not in comparison, on a performance scale of just them. And coaches come & go that behave similarly and post similar numbers. So probably not all good in that comparison of those that get the opportunity and maybe not for those outside either?

Kings: 20 of 20 most used player pairs negative (at minimum, not sure how far it goes), almost all bad to horrendous. 18 of the top 20 trios, same. 16 of the quads. 12 of the top 20 lineups negative, including 5 of top 6. Not much management, coaching or analytics can point to that is working. Few that are not getting much time. Dink, dink, dink, disaster.

368 lineups used. "Coaching" Less than 30% cumulatively positive. Coaching. Less than 14 minutes per game are played with lineups over the ridiculously small threshold of 1 minute per game. 34 minutes per game spent on low success dink and super dink lineups. Coaching madness unchecked.

Hawks, 100% of top 20 most used player pairs are negative, but many mildly. 95% of trios, about 75% of quads and 5 man units. Best big minute lineup being used / tested at less than 2 minutes per game. Is it good? Don't know, don't look they really care to know. Odd way to go into future.

Good news Nets: top 2 most used lineups doing very well. Interestingly, no common player. Two flights. Might be a good 10 man rotation? Maybe, but about 85% of total minutes going to other stuff. Insufficient testing.

Why focus when you can goof around with 420 lineups? Why try to have hardly any non-tiny sample sizes? We can just decide what we like with virtually no data or just keep playing randomly for awhile further.

Memphis: 80% of top 20 most used player pairs are negative, including every one involving Gasol. But nah we apparently reject the idea of major change.


Two bigger minute lineups doing well. But only getting about 10% of total time. Sounds familiar. Why test rigorously? Guessing with 388 lineups is considered advanced coaching. If they don't start working better, we can try 300-400 more new ones.

Lineups in league used over 200 minutes this season: 71% are cumulative winners. Average of less than 1 per team.

Lineups in league used 100 - 200 minutes this season: Less than 40% are cumulative winners. About 1.4 per team.

Lineups in league used 50 -100 minutes this season: 46% are cumulative winners. About 3.6 per team.

Lineups in league used 10 -50 minutes this season: 52% are cumulative winners. Average of 40.5 per team.

Lineups in league used less than 10 minutes this season: Less than 29% are cumulative winners. Average of 302 such lineups per team.

Finding good big minute lineups and using them more til issues arise / performance drops makes the most empirical sense. Just small minute lineups not near as good, but maybe better than earlier in season. The real issue is all the horrible on average super dink "I am (micro) coaching" lineups.

So the average NBA coaching lineup record to date is 6 mostly small non-dink lineups, about 40 dink lineups and about 300 super dink lineups. Does anyone think this distribution is optimal or even logical? My impression is that most media and fans are not aware that it is this splintered. It does not seem that coachs & GM care or maybe hardly consider this. If they did I really wonder why they don't change, moderately or radically.

10 main non-dink lineups, mostly at least twice as big as current, 10-20 more situational, fairly large dink lineups and 50-150 super dink lineups would seem possible, more disciplined and probably better performing than the current guessing with way more, overwhelming tiny and poor historically super dink lineups. I'd probably want to concentrate even further but this is a first call that almost no team or no team at all are likely to answer. If it happens, I think it will require an analytic department recommendation that the GM sells or insists upon with the Coach. A few coaches in history have shown starting lineup concentration beyond the norm. Probably should look further at overall concentration beyond that.

These distributions will get 50-100% larger in lineup counts by season end. Will the minutes get more diffuse or improve in concentration? More to check but the concentration has almost never been enough imo. How does it compare to average lineup concentrations of top 5-7 offensive players in football or the top 5-7 in baseball lineups? I'm not going that far but maybe someone could if interested. Maybe they've tried concentration more or better? If I was an insider, I'd want to know, try.

2 minutes looking at Jazz lineup and sub-lineup data and this lineup screams out for much, much, much more testing: Rubio Mitchell Ingles Jerebko Gobert. Actually tested only a bit more than 1 minute per game. Good lineup results and strong overlapping trio results and nothing really competing. I'd test this 5-10 minutes per game and with Crowder instead of Ingles or Jerebko too.

Celtics 6-7 over last 13 games. No lineup used 5 minutes per game. 7 used over 1 minute per game. 70% of all minutes over this stretch used by about 165 dink lineups (less than 1 minute per game). Not terrible but not that good results either.

In basketball media, you will occasionally see mention of best performing lineups in league, performance of a team's starters and occasionally performance of a particular players top lineup or the choice between a few. I have yet to see anyone showing independent interest in overall league or team lineup trends by lineup type, the size of top lineup utilizations, the vast quantity of dink lineups, their average poor performance, the % cumulatively positive, etc. I don't get the lack of interest and work on the subject. There is very little public work on sub-lineup data. There is plenty more detailed and advanced work to be done; but with so little apparent interest in the basics (including from former and current insiders to my eye & ear reading & chatting with a few of them), I wonder whether to go further. Thanks to the 2 people who commented in this thread and the handful who did in prior year or two of work on the topic.

If a team insider has been influenced by this, a private message would be welcome. If a team insider wants to pursue more on lineups, say something. To me or another person of your choice, if you aren't getting enough / using enough right now.