APBRmetrics

Crow wrote: ↑Fri Nov 27, 2020 5:42 am Here is a first cut (5 minutes of thought) lineup design for the Warriors:

Curry Bazemore Oubre Green Wiseman 16 min / g

Curry Wiggins Oubre Paschall Wiseman 8

Wanamaker Lee Wiggins Green Chriss 6

Curry Bazemore Wiggins Green Chriss 6

Wanamaker Lee Wiggins Oubre Paschall 6

Curry Bazemore Wiggins Green Chriss 6

Order of usage is not specified beyond the initial lineup, which probably starts and maybe closes initially. A death lineup of Curry Bazemore Wiggins Oubre Green might takeover a spot, especially in playoffs.

Minutes / game:

Curry 36
Wanamaker 12
Bazemore 28
Oubre 30
Wiggins 32
Green 34
Paschall 14
Wiseman 24
Chriss 18

Change as appropriate as weeks / months of data come in.

Use others in case of injury, rest, blowout and special situations but max trial of main lineups and minimize random, mostly meaningless dink lineups.

Unless I missed it in reviewing the line up data, Kerr has not uses my recommended biggest minute starting lineup
for a single minute.

Probably a mistake to not even try that while trying so many other lineups. Can't prove claim it is probably worth playing til it is.

2nd lineup above, also totally unused.

3rd, 4th & 6th light to none with absence of Chriss.

5th lineup, great... in 7 minutes.

Crow wrote: ↑Mon Nov 30, 2020 7:20 pm Draft Blazers rotation:

https://twitter.com/bballstrategy/statu ... 10240?s=20

My proposed starters used just 5 minutes.

My 2nd lineup used as starters and good.

3rd and 5th not yet available with Collins' absence.

4th unused.

Crow wrote: ↑Wed Dec 09, 2020 4:25 am Lots of options for Wizards but they will need to dial in the lineup rotation to something decent pretty quick to avoid too much chaos and hopefully be effective enough to stay in playoff hunt.

I'd start with something like this line up set probably. A bunch of deep bench players I don't know. I didnt really consider them but maybe as time goes on.

Westbrook Beal Brown Bertans Bryant 16 minutes

Westbrook Brown Bonga Bertans Lopez 8

backup PG Beal Bonga Hachimura Bryant 8

backup PG Beal Avdija Hachimura Pasecniks 8

Westbrook Brown Avdija Bertans Wagner 8

In no particular order of appearance (beyond first line up of starters / probably closers). Can be more than one stint.

Average minutes

Westbrook 32
Beal 32
Brown 32
Bertans 32
Bryant 24
backup PG(s) 16
Bonga 16
Avdija 16
Hachimura 16
Lopez 8
Wagner 8
Pasecniks 8

Try 10-20 games and adjust. Vary from this is there is a good reason but I'd suggest they not do the typical 700 lineup thing. At least one big shakes out. Wing priority could vary based on results.

Proposed starters here got just 3 minutes before Bryant went down.

2nd lineup available, used 4 minutes.

5th lineup 0-1 minute of use.

Crow wrote: ↑Wed Dec 09, 2020 6:40 am Minutes summary (starting point)

PG
Jamal Murray 32
Monte Morris 12
Facundo Campazzo 8
Markus Howard 8

SG
Gary Harris 24
P.J. Dozier 4
RJ Hampton 8

SF
Michael Porter Jr. 24
Will Barton 20
Vlatko Cancar 4
Tyler Cook 4

PF
Paul Millsap 20
JaMychal Green 24

C
Nikola Jokic 32
Bol Bol 4
Zeke Nnaji 8
Isaiah Hartenstein 4

Malone not playing the deep bench anywhere near as much as I thought. Not that I really care about playing the deep bench.

Crow wrote: ↑Wed Dec 23, 2020 10:15 pm Suggested Sixers base lineup.

https://twitter.com/bballstrategy/statu ... 13856?s=20

There might be others on my Twitter that I forgot to link to.

Starters doing great in pretty big minutes. I suggested using them way more. Still would, especially in playoffs.

Little to no use of the proposed Scott lineups.

One of the Tybuille lineups great in 9 minutes, the other used 0-4 minutes.

Crow wrote: ↑Mon Dec 28, 2020 7:16 am First cut Bulls rotation. I'd probably change even more radically but this is a compromise:

Sat LV OP PW WC 8 min
Sat GT CH OP M 6
CW GT Sat PW M 10
CW LV CH PW M 6
Sat GT CH OP M 6
CW LV CH PW M 6
CW LV OP M C 6

Probably in that order this time.

Satoranksy 30
White 28
Temple 22
LaVine 26
Hutchison 18
Williams 30
Porter 32
Markkannen 34
Carter 20

First lineup used 3 minutes.
Second / fifth, great in 7 minutes.

All the White lineups, 0-10.


In general my lineup sets have not been tested hardly at all.

Crow wrote: ↑Mon Jan 04, 2021 11:38 pm Pretty obvious imo that the Knicks should try Payton Burks Barrett Randle Robinson more when Burks returns.

Not obvious apparently to Thibodeau / Knicks, yet. 0-1 minute use.

Crow wrote: ↑Thu Jan 14, 2021 5:04 am Rockets absolutely need to remove Nwaba from starting lineup.

Try Wall Oladipo Tucker Wood with Gordon and with Tate. See which does best.

Probably start with Tate and close with Gordon.

I'd play several bench lineups with 0-1 starters with the potential of picking & amping one as a main rotation lineup.

May review more fully later.

Wall Oladipo Tucker Wood with Gordon and with Tate are 2nd and 3rdmost used. One moderately positive, one moderately negative.

They haven't picked / used a 0-1 starter lineup in major way.

Nets lineup analysis on twitter recently.

80% of Wizards' most used 20 pairs are negative, more than half badly including Westbrook - Beal at -11.4 pts / 100p. Beal negative with everyone in main rotation but Bonga. Westbrook negative with all except I Smith.

I'd play Beal without Westbrook as much as they are willing. Smith Beal Bonga Avdija Wagner or throw in 1 other.

Westbrook with Smith, Bertans, Bonga and Lopez or Wagner.

I'd trade both and more including Bertans. Going nowhere as is.

I assume that Brooks and most to all of his staff are headed to being gone this summer. 10 man coaching staff and not much to show for it. Good at protecting ball and getting to ft line. Ok shot distribution but poor shooting. In bottom 11 on 5 factors.

Hachimura found 3ball, better at getting to line and perhaps better at boxing out; but team results on court and on / off even worse than last season. Probably not being used optimally. Greatly missing Bryant. Probably should play a lot with Smith or Neto, Mathews and probably Wagner.


239 lineups used. 38% cumulatively positive but 119 were worse than -20pts / 100p. Play with pretty random dink lineups, it usually burns overall.

Crow wrote: ↑Thu Feb 04, 2021 7:52 pm

Crow wrote: ↑Mon Jan 04, 2021 11:38 pm Pretty obvious imo that the Knicks should try Payton Burks Barrett Randle Robinson more when Burks returns.

Not obvious apparently to Thibodeau / Knicks, yet. 0-1 minute use.

Now it is Burks' biggest lineup. 87 minutes is only 18% of his minutes though and only 6% of total team minutes. Mildly negative so far. Decent enough to keep testing. More. With and without Robinson. More without... because of way better results.

Other lineup analysis at my twitter handle recently about Warriors, Wolves, Rockets, Wizards and perhaps others.

Here is comparative analysis of some good and supposed to be good teams in East along with #15 seed Pistons. It includes some data I don't often report and some I have not previously calculated, nor has anyone else in public to my knowledge.

But first, here is some data I have reportedly in past: total number of lineups used and percent cumulatively positive.

Celtics 341 lineups, 37.5% positive.

Bucks 216, 43%.

Sixers 222, 39.6%.

Heat 283, 39.6%.

Nets 254, 45.7%.

Raptors 243, 41.6%.

Pistons 225, 38.7%.

Celtics, by far most lineups used and lowest % positive. Bucks fewest lineups used and 2nd highest % positive. Nets highest percent positive.

Now moving toward the new stuff.

I wanted to compare degree of positive bigger minute 5 man lineups (over 20 minutes for season, a ridiculously low bar meet by few lineups given industry practices) to percent of 20 most used player pairs that are positive.

Of these teams, Sixers and Heat had lowest percent positive of lineups used 20 plus minutes (38%) followed by Celtics (44%). The highest were Nets and Raptors (73%) followed by Bucks at 64%. Pistons were median (58%).

On % positive most used 20 player pairs it was as follows:

Celtics 70%
Bucks 100
Sixers 80
Heat 50
Nets 85
Raptors 85
Pistons 30

Now the new ratio of % 5 man lineups positive at or over 20 minutes ÷ % positive of 20 most used player pairs. In effect a conversion rate of positive pairs into positive 5 man lineups.

Sixers had the lowest conversion rate at .475. Celtics 2nd lowest at .63, then Bucks 3rd lowest at .64. Heat at .76, Nets and Raptors at .86. Pistons with the by far highest conversion rate at 1.93. Doing relatively best at conversion, at least for very biggest lineups (only a small part of the total minutes) with least % positive pairs.

Of course pairs are used simultaneously in pairs and lineups. So they affect / make each other rather than just one way.

And of course not every positive is the same magnitude of positive or same level of use, so this is fairly crude method. But you can see a little bit more about relative lineup performance.

Finally, one more piece of data: % positive of 50 most used player pairs:

Celtics 60%
Bucks 72
Sixers 50
Heat 38
Nets 70
Raptors 72
Pistons 26

Most lineups beyond the ultra dink will draw on the 50 most used player pairs.

Overall across this data, I think it is fair to say Celtics and Sixers generally do worse than most others. Bucks good except on conversion rate. Heat mostly weak except for conversion rate. Nets and Raptors generally good. Pistons weak on everything but conversion rate.

Other conversion rates are possible. Will consider variations for possible future use.

Most all the numbers say something each about talent, fit and coaching / analytics. Some say a bit more perhaps about coaching / analytics.

Same type study for 5 western teams:

% of lineups 20 plus minutes that are positive

Lakers 71%
Clips 77
Jazz 71
Nuggets 53
Thunder 18

% of 20 most used player pairs positive

Lakers 90%
Clips 80
Jazz 100
Nuggets 95
Thunder 5

Conversion rate (positive pair rate ÷ positive 5 man rate with above data)

Lakers 79%
Clips 96
Jazz 71
Nuggets 56
Thunder 3.6x

Total lineups / % positive
Lakers 184, 47%
Clips 240, 41
Jazz 93, 43
Nuggets 203, 38
Thunder 199, 38

% of most used 50 player pairs positive

Lakers 78%
Clips 70
Jazz 80
Nuggets 74
Thunder 18

Pretty tight race between top 3. Nuggets generally behind. Thunder way behind except on conversion rate for a few bigger minutes lineups compared to incredibly weak 20 most used player pairs.

Jazz with by far fewest lineups used of any team studied so far.

Could weight all these data reports by unit minutes.

That and maybe other enhancements if any team showed visible interest.