Last week I posted some suggested lineup changes for the struggling Celtics. In response to that post Crow made some great suggestions. He pointed out teams play several hundred lineups during the season but after a while good teams have a few (say 5-15) lineups that have shown themselves to be very effective. Why not cobble a rotation out of those few effective lineups? I really think this approach would improve the peformance of most teams. The reason why this approach has not caught on is that coaches usually think they know the “right matchups” based on who the opposition has put on the court. Why not make the other team react to your team’s strength?
I still think playing lineups that have proved their effectiveness makes sense, so here is a suggested rotation of lineups for the Magic.
These 6 lineups have played 17.65, 16.29, 15.38, 56.23, 20.58, and 27.26 points better than average, respectively. These lineups involve 11 players (all but Bass). For the season as a whole the Magic play 7 points better than average. How can they not play better if they primarily stuck to these lineups?. There are even 2 lineups that rest Dwight Howard well. Also, with this sequencing nobody plays in more than 3 consecutive lineups, so rest is accounted for.
1. ORL 17.65 ( 14.63 4.88 -9.76 27.91 1.76 6.87) 335.71 minutes 111 appearances 1 A 13.30 93 $
Barnes Carter Howard Lewis Nelson 29.02 years 1610_ORL_2010
2. ORL 16.29 ( -2.00 1.52 3.52 -2.18 -0.22 5.97) 74.48 minutes 48 appearances 6 A 24.49 38 $
Anderson Gortat Pietrus Redick Williams 27.01 years 14369_ORL_2010
3. ORL 15.38 ( 11.33 3.57 -7.76 39.75 1.83 6.45) 31.80 minutes 27 appearances 2 C 12.07 8 $
Carter Howard Lewis Redick Williams 29.41 years 12872_ORL_2010
4. ORL 56.23 ( 20.21 12.19 -8.02 47.74 2.66 4.19) 63.29 minutes 26 appearances 7 A 55.36 73 $
Anderson Barnes Howard Nelson Redick 25.79
years 5187_ORL_2010
5. ORL 20.58 ( 12.58 4.31 -8.28 49.07 2.08 7.18) 28.99 minutes 21 appearances 3 C 18.21 11 $
Barnes Carter Howard A. Johnson Lewis 30.49 years 714_ORL_2010
6. ORL 27.26 ( -15.54 -6.73 8.82 -36.59 -2.10 5.45) 26.92 minutes 18 appearances 6 C 28.53 16 $
Gortat Lewis Pietrus Redick Williams 28.76 years 14880_ORL_2010
Looks like you are recommending a good set.
Why Orlando has chosen to spend over 75% of court-time on other lineups I don’t know. They are doing alright but most expected or want better.
Plenty of room between this approach and the current one. Don’t have to take it all the way but some greater lineup concentration would seem helpful. Will be interesting to see what lineup set and concentration level they run in the playoffs.
Comment by Crow — February 16, 2010 @ 11:32 am
Staggered transitions between lineups could be accommodated. They’d add lots more lineups but they don’t need to take much total time.
Comment by Crow — February 16, 2010 @ 11:37 am
Looks like 80-90% of their point differential is coming from that set of 6 lineups. The sum of the impact of the other nearly 200 lineups is just a bit over neutral.
Comment by Crow — February 16, 2010 @ 11:52 am
Another researcher has shown that if you sort lineups by minutes used for the entire league it is a pretty clearcut and smooth curve from the most used being on average best to lightly used on average being least productive. But coaches aren’t using the most used that much and there seems like plenty of room to use them more.
Comment by Crow — February 16, 2010 @ 1:04 pm
There are less than 1 lineup per team in league that has been used an average of 5 minutes (or about 10% of the time) per game for the season. Only 6 lineups are used 10 minutes a game. I’d think almost every other pro sport would beat these lineup concentration levels, excluding the pitching issue in baseball and maybe defense in the NFL.
Comment by Crow — February 16, 2010 @ 1:11 pm
How much of these great lineups can be caused by sample size? 31 minutes or 28 minutes isn’t really a good indicator of how it will play out over the entire time. Also, if these lineups are used more often, won’t more coaching and scouting and adjustments go into countering these lineups and making them less effective. While I wholly agree that coaches need to go beyond their typical thinking, these numbers are not going to be the same if they played them very exclusively.
Comment by djn123 — February 18, 2010 @ 4:06 pm
Until there is 200+ minutes I am not that confident of how good a lineup is, but if a small minute lineup is +15,+20 or more I think you can at least say there is little chance that it is actually a bad lineup.
Comment by Crow — February 20, 2010 @ 5:30 pm