Examining Lineup Data – the Value of Continuity

*Note – this column was written in advance of the Pelicans’ November 10th victory against the Bucks*

Key Takeaway: Pelicans lineups that have played more than 5 minutes together this season (making up ~65% of the team’s total minutes played) have outscored their opponents by 6.6 points per 100 possessions (roughly the Thunder’s net rating last year), but the 35% of minutes using lineups that have played 5 or fewer minutes together have been outscored by a historically bad -31.4 points per 100 possessions.


So, first things first – the Pelicans are a bad basketball team right now. There is no other way to explain a start of 0-8. But, some good news – they are not as bad as their record would seem to indicate. The other winless team, the 76ers, has a net rating over twice as bad as New Orleans (-6.8 vs. -14.4!), and they’re only one point per 100 possessions worse than the 4-4 Grizzlies. So, what’s going on? We’ve already had a few great pieces on the site digging into this question – check them out here, here, and here – but I want to take a different approach in regards to something that really surprised me.

Curious to see what has worked and what has not from a lineup perspective so far, I dove into the NBA stats website to see what kind of conclusions could be drawn. Admittedly, it’s a little early to make any concrete declarations based on individual lineups, as the sample size is still too small to be terribly significant. However, on aggregate, something caught my eye pretty much instantly. Take a look:


If you’re like me, your first reaction is “that sure doesn’t look like an 0-8 team.” The top three lineups are all positive to varying degrees, and the others below are a relatively even split of good and bad. So, I went a little further – and the results were pretty eye-opening. Take a look:


Yes, you’re reading that correctly – Pelicans lineups that have played more than 5 minutes together this season (making up ~65% of the team’s total minutes played) have outscored their opponents by 6.6 points per 100 possessions, a number that would rank among the best in the NBA. Conversely, the lineups that have hardly seen any time together are outscored by an absolutely awful -31.4 points per 100 possessions, which would be the worst net rating in NBA history by no small margin. Naturally, you’d expect a difference here, but a difference of this magnitude is pretty absurd. The craziest part? It has nothing to do with Anthony Davis. Here is the proof:


80% of minutes from the “common” lineups feature Davis and 68% of “uncommon” lineups do, so no significant difference there. In fact, the most jarring number is probably the Pelicans’ “common” lineups being a +22 in 52 minutes with AD off the court vs. only +11 in 203 minutes with him on the court. There is a lot of noise in those two samples, to be sure, but that is certainly unexpected. So, what possible explanations can we draw from this data?

  1. Coaching. Gentry is learning which lineups work best together, and abandoning those that don’t. We can also put a negative spin on this theory, though – that Gentry hasn’t gotten his team ready to fill whatever role is necessary in a given lineup.
  2. Rhythm. Lineups that only play together for a very short period of time don’t have enough of a chance to establish any sort of cohesiveness, and the results are horrendous as a result.
  3. Injuries. The injuries to Holiday, Evans, Pondexter, and even Stephenson force the Pelicans to go deeper down the bench for meaningful minutes, and a casualty of that reality is short periods of time where lineups that would otherwise never play together end up playing together here and there, leading to disastrous results.
  4. Randomness. It’s still early, and we should expect infrequently used lineups to be less successful than commonly used lineups. The degree of difference at present is far larger than what would be considered normal, but it could be just due to a small sample size.

Thoughts from our readers? I wanted to write this more to generate discussion and see what others think. Is the fact that commonly used Pelicans lineups have an overwhelmingly positive net rating a positive indicator of things to come, or is it just random?

3 responses to “Examining Lineup Data – the Value of Continuity”

  1. Continuity is important, but it’s also instructive to look at individual players and combinations as weak vs strong lineup links at this stage.
    Let’s assume the set of lineups that have played less than 5 minutes together are too spurious to derive meaningful information.
    Of the remaining 21 ineups:
    Terrence Jones is in 8 of the bottom 10 lineups and only 3 of the top 11 by net rating. This team is bad at rebounding so it needs Davis and/or Asik to be effective. When Terrence Jones is on the court the team has its lowest REB% and when Jones is off the court the team REB% is at its highest. The team is in desperate need of a backup big that can rebound or else they can’t end possessions.
    A backcourt of Buddy Hield and Langston Galloway is in 6 of the bottom 7 lineups and 3 of the top 14. Hield and Galloway have the 2nd and 3rd highest usage and the two worst true shooting percentages on the team (not including Diallo). A backcourt of those two is basketball poison, and Jrue can’t return quickly enough.
    A combo of Davis and Frazier is in only 1 of the bottom 7 lineups and 10 of the top 14. When those two are on the court the team’s turnover ratio is at its lowest and the team’s true shooting percentage is at its highest. The NBA often comes down to the 2 man game, and we’re at our best with a decent scoring threat that also knows how to get the ball to Davis.

  2. stiegosaur excellent insights – I think that we agree on all counts. My take was more to suggest that it’s probably more than one thing causing the discrepancy between the two groups of minutes, and to call attention to how drastic the difference is.

  3. MasonGinsberg stiegosaur yeah, it is a pretty stark difference. I’m hopeful that part of it is just Gentry trying out different lineups, but if he has any basketball coaching talent, shouldn’t he know that an Ajinca-Cunningham-Galloway-Hield-Lance lineup is never going to work? Granted you play with what’s on the roster, but really?

Leave a Reply

Your email address will not be published.