APBRmetrics

Horford - Embid, -1.6 pts per 100 possessions as noted earler. Simmons - Embid just +1. These are major disappointments, failures at one or more levels (Brown, Brand, analytics, players). Playoffs, who knows what happens. Major changes may come after.

Team strives when Embid is with Richardson. Simmons does well but less than half as well with him. The trio is moderately good but only 14th most used at 6-7 minutes per game.

Starters, very good; but used less than 4 minutes per game.

Some moderately negative Heat player pairs- Olynyk with Nunn or Jones. And Nunn with Dragic. Might want to review those further. Fix, de-emphasize or eliminate.

They have been good and broadly but probably need to be better to do something in playoffs. Against Pacers and Sixers this season, they generally shot better, but I dunno how much was real advantage vs. random.

Disclaimer:

Lineup and sub-lineup data is subject to change and will vary by matchup and situation. Playing the best to date isn't a guarantee of future success but it does in general suggest by results a somewhat better path (if anything does).

A team is either going heavily with strong performers or not enough. Too much previously bad or meh will likely hold them back.

Of course look at why lineups "work" or not and fix what you can. Have a system or strategy for specific lineups. But generally best performers to the front.

Celtics starters without Tatum have sucked (-21 pts per 100 possessions) though in limited minutes (99). Probably want to avoid that. Injury would be huge issue.

Based on pbpstats.com graphs, Toronto, Denver, Utah and Indiana were high on % of time given to playing all 5 starters in regular season (around 35-40%). Nuggets where only one of these 4 were this was done along with a high rate of 0 starter lineup minutes (over 10%)

Lakers, Bucks, Clippers and Sixers were contenders on the low end of this measure.

Last playoffs, Pacers were highest on usage of 5 starters. Rockets were high too. Warriors emphasized 4 starters time. Warriors and Bucks emphasized 3 starters more than other teams.

More could be done with this data of course.

Slightly less than half of teams used a lineup for over 300 minutes.

Of the 5 most used lineups, 3 were around neutral. Cavs and Nets fired their coaches, though this was not their worst offense. It might have been their best accomplishment.

Then there is Popovich. Big minute lineup, meh results. What will he change short-term with Aldridge out? Long-term? I'd be for major changes but I expecting less to way less.

After the biggest minutes 5, 9 of 10 were positive. Injuries limited a few here and there but pretty much all of them should have been used more. We'll see who ramps up lineup concentration most and best... in a few months. Nuggets and Jazz were leaders in biggest minutes lineups. Nut unless they increase them further, we will see at least some other teams catching up on concentration and yielding net pt gain, separately and comparatively.

Then we'll see how many teams lose / don't retain these lineups for next season and look for a replacement.

Heat have 3 lineups used over 1 minutes per game. All good but only 1 reliably "proven". One modestly tested, one very lightly tested.

After that it is dink lineups. The 17 biggest dinks are 7 good or better, 10 meh to bad. They do not really "know" from experience what they are going to get from these lineups. Way too much uncertainty for me. Could have reduced level of uncertainty with more disciplined testing. But hey, it is only a $100 plus million enterprise. Let's stretch guess, on one basis and / or another, on almost all of the time.

Of course there is always guessing and rarely enough information. But there is better and lesser real testing & guessing.

Of the 359 lineups used this season, probably 70-80% won't be used in playoffs. Try to make sense of / use the data including the sub-lineup data on the lineups actually used (and not) as best you can, after the scattershot season. The way they chose to do it.

I assume by comparison, soccer has far higher lineup concentration. Same for baseball minus the pitcher? How about football for the whole 11 man lineup or the non-line?

How does average NBA lineup concentration compare to averages for Euroleague, college and high school ball? Different length games, schedules, talent distributions and game point spreads but a curiosity. Even more relevant would be comparisons between NBA eras to the extent possible. I might look further into that at another time.

I am not going to do a big review here. But with respect to Sixers, they have performance issues with Embid with Simmons and Harris. On the other side, he does well with 2 or more bench players; so beyond starting and finishing, he might be the player to play most with the bench.

Of lineups used 300 plus minutes and positive, it appears that the Bucks are both best performing and highest in "cumulative average usage" for the 5 guys at in excess of 118.

That is helpful for that lineup and many one offs... with (and because of) Giannis.

Rockets, Nuggets, Griz, Blazers and an old Rockets lineup are in second tier fairly near 110.

Probably ought to see more big minute lineups designed with some usage experience surplus. Don't want to leave much time for usage starved lineups though. Ultimately it adds up to 100% on usage. To the extent you have those lineups you want them to be good on defense... and / or preferably play slow (unless they are still winning lineups)

How much thought do teams give to usage and pace matching within lineups? Not sure... except for what the data shows (with study for those things). Knowing player usage within specific lineups is a necessary step in the evaluation.

DiVincenzo with rest of starters instead of Bledsoe or Middleton is a modest reduction in average cumulative usage. Can do when ahead and / or when opponent strength on floor weakens (on offense or especially defense). Coming in for Matthews is a supercharger move- more often against tough defense or easy to exploit? Or does it look pretty random? Could check Bud's thinking with a little more effort. Maybe sometime in future.

High average cumulative usage should tend to help with spacing but not always. Can still be cramped if movement and shot geography overlaps or crowds too much.

Career average & max usage and efficiency by usage level would be worth knowing, matching up appropriately. If asked / listened to.

Proposed next season lineup rotation for Hawks (based on current assets):

1. 12 minutes Young Huerter Reddish Hunter Collins

2. 12 min Young Huerter Reddish Collins Capela

3. 6 min Young Huerter new wing Hunter Capela

4. 6 min Young Huerter new wing Hunter Fernando

5. 6 min backup PG new wing Reddish Hunter Collins

6. 6 min backup PG new wing Reddish Hunter Capela


If you can't get adequate new wings, substute Collins into lineup 4 in place of the new wing and pushing Hunter to SF.
Add Capela or Fernando to lineup 5, pushing Collins, Hunter and Reddish down one position each. Could do similar with lineup 6 if it helps.

Richardson - Scott - Embid is +12 / 100 p, about 50% more successful thsn any of the others in 20 most used. But is only 19th in minutes. Add Korkmaz and Simmons and that lineup is +29 / 100 p (best in top 20) but only in 33 minutes.

This is affected by missed games but under-done even when available. Will they emphasize these 3 an 5 man lineups enough in playoffs? Probably not.

Among other things, increase time for Harris with Embid and Scott. Reduce time without. Do same with Thybulle (reducing time with Harris and Simmons). Reduce time for Horford with Richardson and greatly reduce time with Embid in lineups with negative track records. Simmons time should emphasize Horford and Richardson. Reduce Korkmaz time with Harris and Simmons.

My first draft playoff lineup set is:

Simmons Richardson Harris Horford Embid 16 minutes

Simmons Korkmaz Harris Horford Embid 8

Simmons Richardson Scott Harris Horford 12

Simmons Richardson Korkmaz Scott Embid 4

Burk Richardson Korkmaz Scott Embid 8

That makes

Simmons 40
Burke 8
Richardson 40
Korkmaz 20
Scott 24
Harris 36
Horford 36
Embid 36

Sprinkle in Thybulle a bit in place of Simmons, Richardson, Korkmaz and Scott where rest is needed or you can get away with it or need it for D.

This lineup set only achieves half or a bit less of the things emphasized in last post. Possibly some changes could be made with more time and motivation. But all these lineups were tried modestly with moderate to big success so it may not absolutely be necessary to achieve all those sub-lineup goals.

Let's see what they do, how it does and how the lineup set compares to this one and how this lineup set might have done overall, on surface hypothetically.

This is a bit rough, but first cut for Thunder:

Paul Schroder SGA Gallinari Adams 20

Paul Schroder SGA Gallinari Noel 10

Paul Dort Diallo Gallinari Noel 6

Paul Schroder Diallo Bazley Muscala 2

Schroder SGA Diallo Bazley Noel 4

Schroder SGA Dort Diallo Adams 4

Schroder Ferguson Diallo Nader Adams 4

Paul 36
Schroder 40
SGA 36
Gallinari 34
Adams 28
Noel 18
Muscala 2
Dort 12
Diallo 20
Bazley 6
Ferguson 4
Nader 4

If Dort or Diallo falter, play more Ferguson, Nader, Bazley. If Bazley does well reduce others.

So it sounds like Brett Brown is planning to make Sinmons "exclusively" a PF. So perhaps most or all of the season data will be dated / no longer as strong a guide.

If the starters include Milton and Thybulle and not Horford, they have a whopping 13 minutes of data and a moderately negative result.

Either this is the hail Mary to possible breakthru or it is going to be rough and questioned.

Why now vs. earlier to way earlier? Does he actively fear being fired for the first time? Is it a brilliant changeup that might be hard for the opponents to figure out fast enough? Maybe it won't be great or terrible or that different. Have to see.

Much or all of my proposed lineup set seems unlikely. Even more than previously.