Most used lineup analysis, leaguewide 2016-17

[Crow]

25 lineups used 40 minutes or more. 40% are better than +5 per 100 possessions. 40% are between +5 and -5. 20% worse than than -5. To repeat that a bit differently, of the top 2% most used lineups, the lineups that coaches most trusted, only 40% were a moderate positive, not that much better than a completely random and equal distribution of results.

Average of 4 lineups used 10+ minutes total per team. On average teams have used almost 70 lineups. # games varies of course. Only 30% have yielded positive net points.

Boston at 38% positive results on 86 lineups used (about 23% more than average). Cavs- 38% positive, 73 used. Warriors- 33% positive, 81 used. Spurs- 37% positive, 98 used. Thunder- 35% positive, 83 used.

+10 per 100 possessions and used over 20 minutes? 17 lineups. Miami has 3 of them, Kings 2, nobody else more than 1. http://bkref.com/tiny/VZHQp

[Crow]

Coaches probably use their most used lineups a lot against each other and so leaguewide the results look meh. How deeply have teams studied using your best against their best, middling or worse (might be hard but did you try?)? Is the coach being conventional and / or optimal with that choice?

[Crow]

LAC starting lineup is runaway best lineup amongst those used over 100 minutes. Most used by a healthy margin as well.

The next 4 biggest minute lineups in league are all negative on net pts. IND, DET, OKC and ORL. How long do they stick with them? Probably a long time for most. Easier that way. Coaches don't seem pushed that hard to re-think their strategy. Their turf, their call.

[Crow]

Some are doing well. Doc Rivers has given his 5 biggest lineups 435 minutes (almost 76% of the total) and got 107% of his net positive point margin from them. This can be done with health, good roster construction... and discipline / faith.

[Crow]

Only 5 lineups in entire league used over 200 minutes and performing at more than a small positive rate.

http://bkref.com/tiny/q9yWV

3 for the 3 best teams, then 2 lesser teams.

25 coaches and front offices haven't created one yet.

[Crow]

Half the league has found a good lineup playing over 5 minutes per game, or barely 10% of the game or more. 1. That is something. Not that much (scattershot for the other 80-90% of time) but...
Only Houston has two. May be a lot about fluke finds / usage.

Might be worth looking at who does or doesn't discontinue bad performing big minute lineups and how quickly.


Of the 5 best over 200 minutes, 4 had 3 good 3 pt shooters in the lineup. 4 also had more FTM. Essentially Moreyball on offense.

[Crow]

Team lineups over 200 minutes

http://bkref.com/tiny/0uGwl

Worst big minute lineups for playoff teams are for Toronto, Detroit and Atlanta. (for 2 of the Coach / "GMs"). Toronto has a good lineup that just missed the minute cutoff. Detroit and Atlanta don't and it is nowhere close.

Utah and Memphis don't have a qualifier and it is nowhere close.

Boston's qualifier (at +6) is about half as good as the least good of the top 6 teams.

[Rd11490]

Worst big minute lineups for playoff teams are for Toronto, Detroit and Atlanta. (for 2 of the Coach / "GMs"). Toronto has a good lineup that just missed the minute cutoff. Detroit and Atlanta don't and it is nowhere close.

Maybe I don't understand what you are trying to say, but Atlanta's most used lineup is +6.2.

[Crow]

Huh. Perhaps I looked too fast / misread. The plus and minus signs are tiny and it appears I missed on that one (or the data displayed differently on my query yesterday?) Thanks for the correction.

[Crow]

About one-third of teams have a lineup used over 400 minutes. Minn., Det. and Ind. have two. Among top contenders GSW, HOU, LAC and WASH have top preforming ones. Cle, Tor, Bos, SA? Nope, not at this minute level.

At 250 plus minutes, Boston has 7th best performer, SAS 9th, Cle, 12th. Toronto? 30th best at -10.6 per 100 possessions. None rival the Warriors best at +24.2.

Utah doesn't have a big minute lineup to rank until you go to 150 minutes. Then they get 40th place at flat neutral. Problem.

Detroit with 2 of the worst 6 over 150 minutes. Roster construction and / or coaching and team effort. NYK and PHO also have 2 of the worst 10 in group. Memphis with 11th and 13th worst. Days of a great big minute lineup disappeared for them. At 150 minute criteria Cleveland has the best lineup at +32. Liggins in it.

5 of top 8 most used Raptors lineups and 11 of top 20 are negative. How do you tolerate that or advance with that set of results?

[Mike G]

5 of top 8 most used Raptors lineups and 11 of top 20 are negative...

How do they win games like this? Was their early season success due to gimmick lineups? Is their lack of actual depth catching up with them, as opponents have scouted their trickery?

[Crow]

You can slice stuff different ways. 7 of their top 13 lineups are very strong winners. It is not this simple but maybe try playing the big winners more Dwayne and cut back on some of the big losers?

D. Carroll | D. DeRozan | K. Lowry | P. Siakam | J. Valanciunas has cost them about 2.5 pts per game when used at about 11 minutes per game. That lineup at least disappeared after February.

430 lineups used. Not extreme for league but too many imo. Wasted opportunity to target research. 40% positive is probably ok in league, maybe even good.

Missing Lowry but only 6 of his top 10 most used lineups are positive. DeRozan only 5. Lineup construction seems to matter. Maybe not well founded, though randomness plays a role.

Lowry's top 8 most used player pairs are all positive. DeRozan, only 4. Fine together.

[Crow]

Cavaliers in March, 5 of top 7 most used lineups were negative. 3 very badly. For full season, top 7 were all positive. Pretty big change. Less K Love is part of reason.

[Crow]

For season, teams average 10-11 lineups used 50 plus minutes. About 2/3rds are positive.

Warriors only have 8. Cavaliers only 6. Two of the lowest totals out there. Injuries are a factor but that is very light on concentration of testing. Maybe somewhat intentional hiding. Does hiding have a history of working? Is the confidence level in the eventual dominant playoff lineups high and is that degree of confidence warranted?

Average team has used 518 lineups and will push closer to 600 by the end. The average lineup has been used about 7 minutes for the entire season. Only just a bit over 30% had positive net results. Crazy patterns to me. Only 2% tested over 50 minutes and 50 minutes of testing is too low. Need more over 100-200. It is still guessing but somewhat better guessing if you test longer.

My impression is that we are still generally in the dark ages when it comes to lineup testing and probably in lineup optimization. Very dark. Maybe that is too severe but there sure are weird, glaring on the surface issues.

[Crow]

75% of the top 20 most used lineups from last season are no longer possible due to player movement. 65% of the 20 best performing lineups over 200 minutes are gone for the same reason. New season, most of the semi-proven, gone. Begin process again.