APBRmetrics

Spurs had by far the best performing line up in playoffs on +/- (among those used 50+ minutes), +36 per 100 possessions, compared to the next best from the Heat at +20. However the Spurs used theirs less than 5 minutes per game. The Heat 10 but only in 8 games (starters with Lewis). Spurs’ second best lineup was +12 and third best was +5 but still that is an awfully light use of best lineup IMO, something that was true in regular season as well and was noted. Won in a general runaway so it didn't ultimately hurt them, but there was an early scare. Miami’s second best qualifying lineup was negative so they also might have gained considerably with more use of the best.

Thunder’s best lineup (starters with Jackson over Thabo) was 3rd best overall in playoffs. Second best for OKC was another lineup with Westbrook - Jackson. The traditional starting lineup was a distant 3rd in performance (+3) but had more minutes of use than the better 2 combined. Same old, same old IMO overly rigid lineup thinking led to not enough advantage cashing. Will RW-RJ pairing be in the starting lineup next season? Probably not. Should be final lineup at least but total time on court matters most.

Indiana had 4th best performing lineup. So top 4 spots went to conference finalists, in line with my past contention that having a top performing lineup matters.

GSW, ATL and LAC had an appearance in top 10 but none of the rest of the contenders did. They either need to use their best lineup(s) way more or find even better to go higher.

Overall only 5 playoff teams had a positive lineup used over 15 minutes per game. That seems a bit low to me for the best teams, with analytic resources, after a full season of trial and learning. I’d certainly feel more confident in a team with such than without. Only Indiana had that the 15+ minute level positive performer amongst the top 4 teams in the regular season. The other 3 did have a top 20 lineup used over 250 minutes.

GSW had the 2nd best regular season with Bogut and surely missed having that in the playoffs.

Clippers had 1st and 3rd best in regular season, featuring Crawford and Collison with the other 4 core players. So what does Doc do in the playoffs? Play Redick with those 4 core guys many times more minutes than the other 2 combinations (each of these used less than 50 minutes compared to almost 250 for the Redick lineup). With Redick was good but only to the tune of 9th best in playoffs (not enough) and just a bit better than half as good as those other 2 lineups were in regular season. Would it have mattered if he had played one or both of those lineups more? Maybe, maybe not.

Not everything goes to form for lineups in a dynamic game with match-up series that might not favor, but I continue to think that lineup performance data suggests that coaches (and organizations) are not as focused on and effective at lineup management as they ideally should be. Or at least it sorta looks that way to me from the stats.

Had to take a second look for those low minute collison and Crawford lineups. They did perform badly... In 31 and 13 minutes respectively. Did Doc react appropriately or over-react? Go by 31 and 13 minutes or the 250 plus in reg.season? It would tough to go ignore the immediate but his choice didn't work well enough.they probably could have done a lot of all three if he'd cut a lot of the other non productive in reg season and playoff lineups.

A brief addition on the topic of top lineups. I looked at the conference finalists for the last eight seasons to see where their best regular season lineup used over 250 minutes and best playoff lineup used over 50 minutes ranked. The title winner ranked on average 3.75 in regular season and the same in playoffs. The runner up hasn't off by much, ranking 5.4 in regular season and 4.6 in playoffs. The conference finals losers averaged 14.9 in reg. season but improved to 6.25 in playoffs.

The titlewinner never ranked below 6th in either time period. The runner up worse than 9th only once (Miami 16th in 2014) and never worse than 9th in playoffs. The conference finals losers were worse than 10th in regular season about 70% of time, worse than 20th 25% of time.But they were worse than 10th in playoffs only 3 times.

The titlewinner and runner up were close in even in who was better in both time periods. The conference finals losers were never the best of the 4 in reg season but there was one case in the playoffs (Spurs in 2012).

62.5% of the covered years at least one of the teams with a top 2 regular season lineup made the Finals. Once both did. 75% of the covered years, one of the teams with a top 2 lineup in playoffs made the finals but never both.

For the west, the Finalist averaged 4.4th best regular season lineup when they won or lost. For the east, the title winners averaged 2.7 while the losers averaged 6.0. For the west in the playoffs the winners averaged 3.4 while the losers averaged 4.3. The east averaged about 4.4 for winners and losers.

When we get further along I'll probably check to see who meets these marks and who doesn't. It is probably not the cause of success in itself but it seems like the recent trends are pretty distinct.

Great stuff Crow.

Overall only 5 playoff teams had a positive lineup used over 15 minutes per game. That seems a bit low to me for the best teams, with analytic resources, after a full season of trial and learning. I’d certainly feel more confident in a team with such than without. Only Indiana had that the 15+ minute level positive performer amongst the top 4 teams in the regular season. The other 3 did have a top 20 lineup used over 250 minutes.

Also surprising to me that so few teams had lineups that played more than 15 minutes. Average lineup plays together for 8.19 minutes a game since 2000, and only 7 lineups since 2000 have averaged more than 20 minutes per game.

Any theories on why that 15 minute mark is rarely exceeded?

A couple of points:

1) Strategy. On a defensive possession in GSW@POR, Kerr wanted to be able to switch everything on the perimeter, so he pulled Bogut for Barnes.

Kerr didn't play his best defensive lineup in favor of a match or strategy. Is this smart?

2) Positions. I mentioned something along the lines of "Why can't Thomas Robinson play the 5? Be my guest and post up Kendrick Perkins or Robert Sacre." While he rightly called me out for picking two terrible offensive Cs, I still don't think why it's a net loss to encourage CP-DAJ instead of CP-Blake pick and rolls, or encouraging Tiago Splitter to post up. But if Splitter nets 5 points in 3 possessions in a 2 minute span, coach is going to get killed.

3) Star preferences. Thought Duncan should've been a 5 a long time ago, but because he insisted to be a 4, Spurs always had to find 5s. If that means a happy Duncan willing to take a discount, then it's worth it.

4) SSS. Each lineup plays so few minutes together that any conclusions we draw from lineups are pretty much moot.

(Edit: first post. Go easy on me.)

Willgao, thanks for taking the plunge here.

Yes there can be valid strategic (or tactical) reasons to alter lineups to meet immediate needs and opportunities. But I wonder if this is overdone and also under-evaluated. How critically do coaches and their organizations analyzed their tactical lineup adjustments? Probably a lot but with really small sample sizes for these micro- choices it seems hard to learn from responsibly. A big reason for advocating larger minute lineups is to improve the significance of the sample and raise the credibility of the judgments from ultra-thin to daily thin or even modest.

There certainly are lineups that one might feel awkward about trying or have player issues with. But the reality is that coaches tend to maybe 25-40% of all available time on hundreds of little minute lineups. To me it would seem more logical to try to address the set of game scenarios with 20-30 lineups 85% of the time and get lineups results in triple digit minute levels instead of low double digit minutes.

BLI, in addition to the points willgao makes, injuries and trades do a lot to make "big" minutes rare. But these factors do not make 10-15 minute per game lineups
that difficult to employ. They are somewhat more common in playoffs but not much more.

I have criticized Popovich for not emphasizing best performing lineups much. The response is probably tied to belief in system and belief it can and should work the same regardless of who is on the floor. It worked for him this time and 4 other times but it could be considered a notable reason for many of the 10 non-title seasons. I tend to think that to beat the Spurs it would be a big benefit to have a lineup that gives you matchup or style advantages and then play the hell out of it.

Another factor that reduces the max you can or want to play a great lineup is the desire to have other pretty good to good moderate minute lineups within player playing time maxes. If you want 3-4 guys you like in most or all lineups you can't load up all the best on one. But you can give their best lineup preference in resources and time. Ideally you probably need at least 5-7 lineups to rotate thru but the goal should be to field the optimally performing set. That is likely to be the few best and then the best necessary supplements to get you thru.