Early Celtics lineup data:
7 games in, 140 different lineups used. (League average is 107 lineups in 8.2 games. So Stevens' unique lineups used / games played ratio is about 50% above league average.)
Only 4 for 10 minutes or more total. All 4 very negative so far. (On average teams had 7, with almost 4 positive. All 4 Celtics lineups used over 10 minutes were bottom third performers compared to others meeting that criteria.)
Third straight year actual wins are less than expected wins, according to BR. Just by a little but you'd generally like to see actual wins exceed for "great" coaches.
7th easiest schedule to date according to Sagarin. 0-4 against his top 16.
Will check again later when there is more data.
Checking in on Celtics lineup usage (and other teams)
Re: Checking in on Celtics lineup usage (and other teams)
They're a fairly young team, and that seems to be typical.Crow wrote:..Third straight year actual wins are less than expected wins, according to BR...
Re: Checking in on Celtics lineup usage (and other teams)
That is a potential contributing factor.
About the same minutes weighted age as the Thunder though.
About the same minutes weighted age as the Thunder though.
Re: Checking in on Celtics lineup usage (and other teams)
Update:
4 of top 5 minute lineups negative. 3 very badly.
Only one lineup used for more than half the games. Only 6% of lineups used in more than 3 games so far.
Only 4 lineups used more than 11 minutes total after 11 games.
62% of the 186 lineups are non-positive on plus minus.
W-L is ok but behind many season projections. Actual wins - expected wins is at -2.
4 of top 5 minute lineups negative. 3 very badly.
Only one lineup used for more than half the games. Only 6% of lineups used in more than 3 games so far.
Only 4 lineups used more than 11 minutes total after 11 games.
62% of the 186 lineups are non-positive on plus minus.
W-L is ok but behind many season projections. Actual wins - expected wins is at -2.
Re: Checking in on Celtics lineup usage (and other teams)
Celtics were predicted by us around here to win 36 to 52 -- avg 45.1 -- and they're headed to 47.6, which is 3rd best (almost 2nd) in the East.
http://www.basketball-reference.com/fri ... f_prob.cgi
http://www.basketball-reference.com/fri ... f_prob.cgi
Re: Checking in on Celtics lineup usage (and other teams)
Ok, actual wins aren't much off the average prediction. I didn't check it. I guess some of the higher media predictions stuck out in my mind.
About 2/3rds of the predictions on kmedved's list predicted 45 wins or more. Simple 6-5 projects to 44 wins mathematically. So most were higher, by a little or more.
Last playoffs Stevens on average played a new lineup every 2 minutes and most never again. This season it is about every 3 minutes, very rarely to be used more than twice more so far.
About 2/3rds of the predictions on kmedved's list predicted 45 wins or more. Simple 6-5 projects to 44 wins mathematically. So most were higher, by a little or more.
Last playoffs Stevens on average played a new lineup every 2 minutes and most never again. This season it is about every 3 minutes, very rarely to be used more than twice more so far.
Re: Checking in on Celtics lineup usage (and other teams)
Now only 3% of lineups used in more than 4 games. Only about 16% lineups have gotten more than 5 minutes run total. Less than half have even gotten 2 minutes total.
Only 4 coaches (teams) who are -2 on actual wins - expected: Scott, Wittman, Brown, Stevens.
Only 4 coaches (teams) who are -2 on actual wins - expected: Scott, Wittman, Brown, Stevens.
Re: Checking in on Celtics lineup usage (and other teams)
Stan Van Gundy with a different style. The number of lineups used is just 45% of Stevens' total, including the by far most used lineup in league at almost 20 minutes per game. Top Pistons lineup has played as much Celtics top 13 lineups. Top 10, twice as much as Celtics top 10. Plus 1 on actual wins - expected.
Re: Checking in on Celtics lineup usage (and other teams)
Celtics starting lineup tonight -16 per 100 possessions in previous usage and 2 negative stints in previous game. First stint -5. Second stint -5. Third stint? Yep. -5. Two other lineups have good first stints of plus 5 and plus 8. Second stints? Nope. Best lineup in previous game got no second stint either. Too many other moves to try.
Re: Checking in on Celtics lineup usage (and other teams)
Starting lineup is now -25 per 100 possessions. Stick with it or turn to one of the positive "bigger minute" lineup alternatives with a very slim 10-15 minutes total of testing? Celtics not shining at picking decent sized minute strategic winners right now. Waste time on 200 rare to return to again dink lineups and you won't get as much data on the top 5-10-20 as you could have.
Non-positive lineups have crept up to 64% now. Can you be a tactical genius when nearly 2/3rds of your situational lineup choices produce non-positive results? Maybe, I guess, if a few really payoff. The pot of gold search strategy. But continually need big payoffs to make up for all the slot machine churn activity. Might help to play winning lineups combinations more, if they are real coach developed and identified winners and not just random payouts.
Last game, big game, with almost a month of testing to draw on... 12 brand new, not yet used lineups in the mix, it appears. Churn. Didn't work. Must be about a dozen brand new never used lineups every game appears to be typical for Stevens. 14 games, 210 lineups this season, 82 games, 845 lineups last season. Looking closer, Celtics in an average game so far use about 15 lineups that have been used before (for an average of a bit less than 3 times each) and 10 brand new ones. Took til Dec. 31 last season for Stevens to get to 238 lineups. So he has accelerated the lineup churn dramatically. Worth noticing, watching I think. To understand better.
Non-positive lineups have crept up to 64% now. Can you be a tactical genius when nearly 2/3rds of your situational lineup choices produce non-positive results? Maybe, I guess, if a few really payoff. The pot of gold search strategy. But continually need big payoffs to make up for all the slot machine churn activity. Might help to play winning lineups combinations more, if they are real coach developed and identified winners and not just random payouts.
Last game, big game, with almost a month of testing to draw on... 12 brand new, not yet used lineups in the mix, it appears. Churn. Didn't work. Must be about a dozen brand new never used lineups every game appears to be typical for Stevens. 14 games, 210 lineups this season, 82 games, 845 lineups last season. Looking closer, Celtics in an average game so far use about 15 lineups that have been used before (for an average of a bit less than 3 times each) and 10 brand new ones. Took til Dec. 31 last season for Stevens to get to 238 lineups. So he has accelerated the lineup churn dramatically. Worth noticing, watching I think. To understand better.
Re: Checking in on Celtics lineup usage (and other teams)
Pistons have the most used top lineup in league and it is quite successful. They follow it with lineups #2-4 being huge successes as well. How much is coaching insight vs. luck? Time will improve the reliability of the answer.
% of lineups used more than 4 games- 13% (more than quadruple the Celtics)
% of lineups used more than 5 minutes total- 26% (almost double Celtics)
% of lineups non-positive- 80%. (worse than Celtics)
Stan is a strategic lineup star, but doesn't have the touch with dink lineups.
Does anyone win a high % of total lineups? Not Atl, SAS, Chi, Dal, LAC, Mia, Cle, OKC on quick check. Even GSW only wins 40% of total lineups.
If you want to win, it appears you mainly need to do it with bigger minute lineups. Stevens may actually win on a higher % of total lineups than some others, but that test is not as significant as winning on bigger minute lineups.
% of lineups used more than 4 games- 13% (more than quadruple the Celtics)
% of lineups used more than 5 minutes total- 26% (almost double Celtics)
% of lineups non-positive- 80%. (worse than Celtics)
Stan is a strategic lineup star, but doesn't have the touch with dink lineups.
Does anyone win a high % of total lineups? Not Atl, SAS, Chi, Dal, LAC, Mia, Cle, OKC on quick check. Even GSW only wins 40% of total lineups.
If you want to win, it appears you mainly need to do it with bigger minute lineups. Stevens may actually win on a higher % of total lineups than some others, but that test is not as significant as winning on bigger minute lineups.
Re: Checking in on Celtics lineup usage (and other teams)
Leaguewide (before tonight) teams had used an average of 176 different lineups. Less than 4 for the equivalent of 2 minutes per game (for all games). So 97.7% of all lineups got used less than that pretty low threshold.
Averaging 1 lineup per team for 5 minutes or more per game. Just 3 teams (GSW, Dal, NYK) on pace to test 2 lineups for 500 minutes season. It will probably not happen for one or more. In all 3 cases it is just a one player change difference between these 2 lineups. That is really not much strategic "testing". One third are not on pace to test a single lineup 500 minutes and most will eventually fall short of this I think modest, reasonable goal.
Last season only 5 lineups got used 500 minutes. That was the lowest in the 16 seasons of BR data, by far in every case but one season. So instead of just failure to improve, we actually saw tremendous regression in fairly big minute testing last season. Just half to one third of what was typical. These lineups being used over 5 minutes per game are averaging close to plus 4 per 48 minutes. The minutes used by the other 5150 lineups are averaging close to -1. So there is a reason to prefer the top few and want to explore expansion of their use. Less than 1.5 per team got over 200 minutes total, or just 2.5 minutes per game. Lots and lots of dink lineups that can't be judged meaningfully. Almost none can be, even on a slim, too slim basis.
It does not have to be this way. Greater concentration in lineup usage, way beyond last season and beyond what used to be, might someday be a next great revolution. But it will probably be slow in coming and the concentrations increases and gains will be measured against those with greater dispersions. The concentrations will probably still be pretty low. You have to intend to break with that and be very disciplined.
Averaging 1 lineup per team for 5 minutes or more per game. Just 3 teams (GSW, Dal, NYK) on pace to test 2 lineups for 500 minutes season. It will probably not happen for one or more. In all 3 cases it is just a one player change difference between these 2 lineups. That is really not much strategic "testing". One third are not on pace to test a single lineup 500 minutes and most will eventually fall short of this I think modest, reasonable goal.
Last season only 5 lineups got used 500 minutes. That was the lowest in the 16 seasons of BR data, by far in every case but one season. So instead of just failure to improve, we actually saw tremendous regression in fairly big minute testing last season. Just half to one third of what was typical. These lineups being used over 5 minutes per game are averaging close to plus 4 per 48 minutes. The minutes used by the other 5150 lineups are averaging close to -1. So there is a reason to prefer the top few and want to explore expansion of their use. Less than 1.5 per team got over 200 minutes total, or just 2.5 minutes per game. Lots and lots of dink lineups that can't be judged meaningfully. Almost none can be, even on a slim, too slim basis.
It does not have to be this way. Greater concentration in lineup usage, way beyond last season and beyond what used to be, might someday be a next great revolution. But it will probably be slow in coming and the concentrations increases and gains will be measured against those with greater dispersions. The concentrations will probably still be pretty low. You have to intend to break with that and be very disciplined.
Re: Checking in on Celtics lineup usage (and other teams)
In wins the Celtics top 18 lineups used were all positive cumulatively including the starters moderately. In losses less than half positive and the starting lineup was on average horrendously bad per 100 possessions.
Still only 2 lineups used over 2 minutes per game for season and less than 3% used in more than 1/3rd of games.
Top 4 on minutes looking better with one nice, 2 meh and one bad.
Only 39 never before used lineups in last 5 games, so that introduction rate is down a bit but the churn thru novel dink lineups continues at a pretty high rate.
Net plus minus of best 5 performing lineups is about plus 90, or 90% of total team net points for season. Best 20 lineups are about plus 200, while the remaining 229 lineups are about minus 100. Dinking is on average a losing proposition.
Still only 2 lineups used over 2 minutes per game for season and less than 3% used in more than 1/3rd of games.
Top 4 on minutes looking better with one nice, 2 meh and one bad.
Only 39 never before used lineups in last 5 games, so that introduction rate is down a bit but the churn thru novel dink lineups continues at a pretty high rate.
Net plus minus of best 5 performing lineups is about plus 90, or 90% of total team net points for season. Best 20 lineups are about plus 200, while the remaining 229 lineups are about minus 100. Dinking is on average a losing proposition.
Re: Checking in on Celtics lineup usage (and other teams)
Last time I checked, lineup combos Houston with Lawson and New Orleans with Gordon had brutal splits.
What changes would you guys advise for their rotations? Any specific units worth featuring that they haven't been already?
What changes would you guys advise for their rotations? Any specific units worth featuring that they haven't been already?
Re: Checking in on Celtics lineup usage (and other teams)
The Warriors 5 most productive lineups (by cumulative net pts, not per 100 possessions) have won their time by 199 pts. Rest of lineups by 109. Spurs best 5 won by over 100 and rest by nearly 100 for 207 total. Of 7 other top contenders checked, 5 had best 5s who had a net over 100. But nobody else in this group was even positive overall after the best 5. If anyone is going to beat the Warriors or Spurs this will have to change (that is improve the selectivity and productivity of rest of lineups) and / or they better play the heck out of the best 5 lineups.