Advanced stats and analytics have changed the game of basketball in a myriad of ways as coaches try to get the best out of their team’s talent and ultimately make for the most efficient basketball on either side of the court.
The pace of play? Accelerating as teams learn they can score easier in transition as opposed to a set defense.
Post ups? Seldom used, and only in high usage by the best of the best.
Difficult midrange twos? Only in a desperation late clock scenario, and even then it’s not usually the most advantageous look.
These changes are mostly noticeable across all of college basketball whether you’re watching blue bloods and powerhouses compete at the highest level all the way down to the low major ranks. What used to be an analytical edge one team could have over another is now common knowledge that is closer to gospel than just one’s statistically backed opinion.
One of the newest pieces of advanced stats that are starting to permeate college basketball is lineup data. Breaking down play by play data into which groupings of players are on the court together at that time means you can see what combinations of players work, and which don’t. It’s not even just five man combinations–you can see if a pair of backcourt players jive with each other, or how that wing trio from the shooting guard, small forward, and power forward have worked together. Really, the possibilities are somewhat endless.
Why does lineup data matter? Well, oftentimes even a minor change can have a huge impact. Basketball chemistry can be fickle, and is somewhat mysterious. Oftentimes you don’t know what lineups are going to work with each other until you see them in live action, and it’s not at all uncommon for a grouping of players that looks great in practice actually struggles mightily in a real game.
Let’s use a real example from just a few days ago.
For starters, the way we evaluate a lineup is by a stat called net rating. Net rating is a lineup’s offensive rating minus their defensive rating. Both offensive and defensive rating are essentially an efficiency rating prorated to a per-100 possessions basis.
That might sound a bit foreign, essentially net rating is this–if the lineup is positive it’s good, if it’s negative it’s bad. Makes enough sense, right?
Let’s get to that example from the other day. Florida was playing South Carolina, and it was tight for most of the game. Then, with only a few minutes remaining, the Gators hemorrhaged a 10-3 run in less than two minutes. They weren’t able to climb out of that hole, and they lost.
Those two minutes, only a few possessions, were disastrous for the Gators and it cost them the game. Runs are a part of basketball, so was this just a regular swing of momentum that happens from time to time?
Let’s take it back to the stoppage before the run happened.
South Carolina put out a closing group that featured AJ Lawson, Seventh Woods, and Justin Minaya that has had success in the last few games. This lineup had a +6 net rating on the season–not a killer lineup, but one that has been quite successful. Any time a lineup has a +6 net rating, it’s unquestionably a solid one.
Then the Gators had a choice of who to put in the game. They chose this combination:
Tre Mann
Tyree Appleby
Noah Locke
Scottie Lewis
Colin Castleton
Looking at the lineup, it’s a group of arguably Florida’s most talented players.
Makes sense to put out, right?
Well…maybe not so much. That lineup had a -7 net rating entering the game, a poor number that has shown just how much they have struggled together this season.
So, if you’re keeping track at home, the Gators put out a -7 lineup, one that has struggled this year, against a +6 South Carolina lineup that has been solid recently.
What would you expect to happen in that situation?
Well, what you expected probably occurred. In only a few possessions South Carolina imposed their will on both sides of the floor and went on a 10-3 run in the clutch, ultimately winning the game.
Could this have been predicted? Maybe. Yes, basketball isn’t played on a spreadsheet and these are real people competing on the floor, but when things go exactly as the data would have expected them to go… it speaks to why that data is important.
Here is another example of why lineup data is so intriguing and why it is so important to modern basketball.
So we know the lineup of:
Tre Mann
Tyree Appleby
Noah Locke
Scottie Lewis
Colin Castleton
…hasn’t been good this season. But what if we made just a tiny change to the combination, taking out Scottie Lewis and instead putting in Anthony Duruji. That would look like this:
Tre Mann
Tyree Appleby
Noah Locke
Anthony Duruji
Colin Castleton
So, we know those same four players with Scottie Lewis have been bad, so how much can the change of one player really matter?
Well, a lot. This lineup has a +16 net rating and has hammered teams whenever they’ve been on the floor together. That right there shows just how much lineups can matter and how the simple decision of who to put in at power forward can win or lose you a game.
Would Anthony Duruji being on the floor in place of Scottie Lewis swung the game in Florida’s favor? I guess we’ll never know, but what we saw on the court was what data predicted, and data would certainly suggest having Duruji in there instead of Lewis would have been a wiser move.
Here’s another subtle lineup change that might surprise you. Let’s make one tweak from this lineup that has been effective, the +16 lineup we just mentioned.
For a number of games Florida’s staff took Noah Locke out of that same group, putting Scottie Lewis in that spot:
Tre Mann
Tyree Appleby
Scottie Lewis
Anthony Duruji
Colin Castleton
That lineup, with just a subtle change from the lineup that was +16 by taking out Noah Locke and putting in Scottie Lewis?
-10.
What makes that -10 net rating so crazy is that not only is the number bad, but the Gators have used that lineup a ton. It’s Florida’s second-most used lineup on the season, which is awfully concerning for a couple of reasons. First, with it being Florida’s second-most used lineup, it’s a significant sample size and that means the poor performance by this group isn’t an anomaly. Second, the fact that this lineup has gotten heavy usage despite bad performance means that the coaching staff is either unaware of this lineup data, or unwilling to let it dictate or assist their lineup choices.
Here’s another crazy number following a subtle change.
Let’s make one change from that lineup that’s -10. This time, we’re taking out Tyree Appleby and putting Scottie Lewis back in:
Tre Mann
Noah Locke
Scottie Lewis
Anthony Duruji
Colin Castleton
For the record, this is Florida’s third-most used lineup on the season.
The net rating of this lineup is an astonishing +15.
Okay, I know I’m throwing a lot of numbers and a lot of lineups at you, so let’s recap. The lineup of:
Tre Mann
Tyree Appleby
Noah Locke
Anthony Duruji
Colin Castleton
…is GOOD. +16.
The lineup of:
Tre Mann
Tyree Appleby
Scottie Lewis
Anthony Duruji
Colin Castleton
…is BAD. -10
The lineup of:
Tre Mann
Noah Locke
Scottie Lewis
Anthony Duruji
Colin Castleton
…is GOOD. +15
Are you starting to see why lineup data is so interesting and why rotations in basketball matter so greatly? These are three incredibly similar lineups with only one change between each of them, but the results are wildly different. This is also why coaches can easily miss what lineups are working and which aren’t if they aren’t using these numbers. They look so similar to each other that it’s easy to forget what works and what doesn’t, and can get you into situations like against South Carolina where a poor grouping is on the floor and plays as such.
Last season the Gators had a lineup of two bigs at once that played very well together. It was:
Andrew Nembhard
Noah Locke
Keyontae Johnson
Omar Payne
Kerry Blackshear Jr.
In only a few games, they put up gaudy numbers–a +91 net rating. Yes, it was only over 35ish minutes over a couple of games, but in those games they were tremendous, and that lineup was a huge reason they were able to beat #4 ranked Auburn.
Despite playing their best basketball of the season with that lineup starting and playing key minutes, they inexplicably stopped playing them altogether. They subsequently lost 4 of their next 6, and the two-big lineup was reduced to nothing more than legend.
Recently Florida has started to play with Omar Payne next to Colin Castleton and some of those lineups have gone really well.
Tre Mann
Tyree Appleby
Scottie Lewis
Omar Payne
Colin Castleton
This lineup has a +8 net rating and has been solid. But, as is the theme of this article, a little change has broken up chemistry and made the two center lineup less effective.
Tre Mann
Tyree Appleby
Noah Locke
Omar Payne
Colin Castleton
This lineup has a -9 net rating, and hasn’t looked nearly as sharp. Again and again, we can see little changes making big differences. It’s also worth noting that the two-big lineup with a -9 net rating has played more than the two-big lineup with the +8 net rating, though some of that is due to the fact the Gators went to it some in the games Scottie Lewis was missing due to health and safety protocol.
You have probably noticed that Tre Mann has been involved in a lot of these lineups and that’s because he gets a lot of minutes, and for good reason. The Gators have a +10 net rating on the floor, and quite frankly when he has been off the floor things haven’t always gone particularly well. When Tyree Appleby has played the point with Noah Locke next to him and Mann has been on the bench, the Gators have a -2 net rating. Even worse has been with sophomore Ques Glover running the point. When he’s out as a lead guard with Noah Locke next to him, the Gators are -25. Those minutes have gone about as poorly as possible, and that’s why the Gators should make sure they have Mann or Appleby on at all times to make sure one of them is the lead ball handler.
Another thing that’s great about lineup data is that you can also look at how certain players affect various statistical categories. For example, Colin Castleton is a player who really changes the statistical profile of the game when he’s on the floor versus off of it.
For example, the Gators are a 3% better rebounding team with him on the floor.
His shot blocking also changes the way teams want to attack the Gators. Opponents take 4% more of their shots from the midrange when Castleton is on the floor.
He also completely changes the way fouls are committed, as the Gators have a 27% free throw rate with him on the floor and a 22% free throw rate when he’s on the floor. The Gators also commit fouls on 21% of possessions when Castleton is on the floor, and 31% when he’s off the floor, a huge change that is worth noting for a team that is fouling way too much.
Lineup data is a nearly endless supply of interesting numbers, an eternal fountain of information that can be harnessed to give a team an edge. There are plenty of programs who slave over the lineup data and use it to get every incremental advantage possible, something that would hardly be possible even a decade ago.
There are always going to be internal and external factors that go into lineup decisions by a coaching staff, whether it be fatigue, focus on the bench, a player’s quality of practice the day before, or any number of things you can imagine. The numbers shouldn’t be the sole driver of what combinations of players get on the floor together, but they should give some indication. These numbers can be accurate predictions of what could happen, as shown when the Gators let a game slip against a lowly South Carolina squad.
How much have the Gators used this data in the past and how much will they look at in the future? Only they know. Now you have a bit of an indication of what lineups have worked for the Gators and which haven’t.
Thank you to my friends at Pivot Analysis, Evan Miya, and Hoop Lens for the lineup data. All numbers are via them.
Fantastic article and data. Much appreciated