Fun with Lineups: The Cleveland Cavaliers

The 82-game Cavaliers preseason is finally complete. With the postseason on the horizon, what will Cleveland coach Ty Lue roll out in the way of a rotation. Better question: What should he do?

The playoffs are here at last. While the NBA regular season is all “fun” (read: “boredom,” “drudgery,” “tedium”) and “games” (82, to be exact), the playoffs are all grit, grime, glory and games (anywhere from four to 28, to be inexact). While many teams in the league view survival of the regular season as an accomplishment in itself, others (like the Cleveland Cavaliers) view the regular season as merely a preamble to the playoffs, the real season; or, at least the only part that matters, anyway.

For the better teams assured of a spot in the championship tournament, the regular season is a rehearsal — a series of test takes to try some things out. “You talkin’ to me? … You talkin’ to me? … You tal-kin’ to me?” (Warning: Foul language in the clip below. Though, to be fair, it does have a 70s Robert DeNiro, so 3.2 F-Bombs/min. were to be expected.)

What’s “fun” about the NBA regular season is messing around, experimenting, and tinkering with a team to see what “works.” There’s five people on the court at all times in basketball (well, four on defense if you have Mo Williams out there), and there’s a certain scientific process to finding the lineups that work best. It’s one part chemistry, one part engineering, one part artistry, and several parts trial-and-error. Every game is a series of exercises involving LEGOs, Lincoln Logs, and Tinkertoys to discover effective combinations. Find the right block at the right time and you can take out the other team with the orgasmic satisfaction of eliminating four lines at once in Tetris. Pull the wrong block at the wrong time, and whole thing tumbles over in a wooden avalanche of Jenga blocks.

Unlike baseball, in which swapping your cleanup hitter with a ham sandwich won’t affect how the team’s sixth batter performs, or football, in which the combination of skill position players are rotated for each eight-second burst of action, based on the individual play call and down and distance, the choice of what lineup at what time in basketball can swing your fortunes or spell your doom. Find the right spot for a tetrimino,¹ and the other team’s battleship sinks. Pull the wrong Jenga block, and risk losing the Ukraine to the Pink Army. The consequences are so drastic that my metaphors can’t even make sense of them. But my point is, finding the right combination of lineups is important in basketball.

On any given night, there are 13 or so players on an NBA roster, with five playing at once. This means that, in your average NBA game, a coach has 1287 lineup combinations at his disposal.² Granted, a handful of players will never enter the game, the starting lineup largely determines who plays when and how rotations are managed, and another bunch of lineup combinations would be so outlandish that they should hardly count (I’m still waiting for that Timofey Mozgov-Kevin Love-Tristan Thompson-Sasha Kaun-Channing Frye town-pillaging, grind-your-bones-into-bread lineup). In any event, there are a lot of possible lineups from which a basketball coach has to choose.

Eighty-two games of basketball provides a decent sample size to examine not only the frequency of certain types of lineups, but the efficacy of them. If you’re a Cleveland Cavaliers fan, which lineups performed the best? What lineups played the most? What lineups should Coach Tyronn Lue play more often? Let’s take a peek.

By dicking around on NBA.com or another website,³ one can find some good lineup data. For instance, it doesn’t take much effort to ascertain that the Cavs best lineup in terms of rating (net points per 100 possessions)⁴ is a lineup of James-Love-J.R.-Delly-Thompson. Good to know. Or, you can find that LeBron James and Kevin Love have the highest plus/minus among two-man groupings throughout the season, meaning the Cavs have outscored opponents by 485 points when James and Love play together. However, James and Dellavedova, not Love, have the highest plus/minus per 48 minutes, with a +16.0 per 48 minutes.⁵

However, once one ventures beyond the top lineup in any statistical category, it becomes a little overwhelming. It becomes an interstate pileup of numbers, commas, and permutations. It sort of looks like LeBron James’ and Tristan Thompson’s names appear in a lot of lineups with favorable statistics, but it’s difficult to extract anything worthwhile out of it, or to generalize the results in any meaningful way. Maybe I should just go back to looking at pictures of Corgis on Instagram.

The main reason rifling through this number-rubble of lineup data is daunting is because the raw data from NBA.com doesn’t do any of the thinking for you — none of the thinking that you or I would do during a game. For instance, anyone who’s followed the NBA casually or listened during broadcasts has heard lots of talk over recent years about “big” and “small” lineups … but there’s no box to click on NBA.com that will show you the performance of Cavs small vs. big lineups. The NBA hasn’t made information on big and small lineups public yet, and Elias Sports Bureau and us sports-consumer peasantry can’t afford to pay Elias Sports Bureau to do that work for us.

So, despite my better judgment, I used by brain to think about common basketball lineup motifs, as well as Cavs lineup quirks I’ve noticed throughout the season. I took all the available lineup data from NBA.com, and hand-picked (well, mad my computer pick) a few lineup varieties worth looking at, then put it in a way that made sense to me — a way that didn’t make me cross my eyes and drool on my keyboard. I looked at small lineups (lineups with one or zero “bigs”) and “biggish” lineups (think medium or large combo meals).⁶ I also looked at three-guard lineups, wing-heavy lineups (with three or more shooting guards or small forwards), variations of the big three (all three, Kyrie Irving and LeBron James but no Kevin Love, etc.), lineups with Kevin Love at the five, lineups with potent two-man combos (such as Dellavedova and James), etc.

You get the point: I tried to pick out all the compelling lineup varieties the Cavs use. (Note: the lineups aren’t mutually exclusive. For example, a lineup with Delly and James could also be a three-guard lineup, which could also be a small-ball lineup, which could also be a lineup with Love at the five, which could also be a lineup with James and four shooters.) Adequately confused, yet? Anyway, let’s take a look at the chart.

The chart above shows the playing time and performance of several different Cavs lineup varieties. It’s a busy chart — there’s a lot going on, especially in the middle. But here’s the gist: the lineups on the right get more playing time than those on the left, and those higher up have outperformed their opponent better than those below. The dotted yellow line shows the Cavs average for qualifying lineups.⁷

What can we learn from the chart? If we look at it strictly from the big vs. small dichotomy, the Cavs have played big over three times as much as they’ve played small — and they’ve played better, too. The Cavs are below their season average per 48 minutes playing small, but over when playing a biggish, traditional lineup (think of those with a power forward and center, such as Thompson and Mozgov, or even Love and Thompson). I think this would be counter to what most people would expect, as most expect the Cavs to perform better playing small ball.

Another thing: Kyrie Irving hasn’t had a great year, in terms of team performance anyway. That might be expected, given that his net rating has decreased from sixth to ninth on the team this season. But this fact is  more obvious once you see that most of the Cavs’ higher-performing lineups don’t include Irving — and Irving’s on the floor 31.5 minutes per game.

The chart also shows that not all small-ball lineups are created equal. While the Cavs have been below their average with a generic small lineup, their best lineups are variations of small lineups. A lineup with Love at the five (where he’d be the only big on the floor) is the Cavs’ best performing odd lineup of the season, on pace to outscore opponents by 34.7 points per game in 149.4 minutes of playing time (and scoring 126.6 points per 48 minutes). That 34.7 point mark is nearly 10 points better over the course of a game than the Cavs’ best five-man lineup.

Finally, it might be noticeable that I kept tinkering with lineups that involved Dellavedova and Tristan Thompson, basically trying to see how well variations of those two had performed, especially when combined with non-bench counterparts like James and Love. Nearly all of the Cavs’ most powerful lineups involve some combo of Delly, Thompson, James, and Love, as evidenced by the upper-left corner of the chart.

How to use this information to complain about coach Tyronn Lue’s moves during the game? Well, first, I think it’s fair to say that any NBA coach has a difficult responsibility juggling lineups — especially in today’s matchup-oriented league. How to find what matchups work when, in an 82-game season, only seven lineups have shared more than one game’s worth of time on the court, against 30 different teams, no less? But I think the chart is helpful in making a go-to plan for the Cavs for most playoffs game, to deviate from as needed.

Now that coach Lue has given up on Timofey Mozgov, the starters (Irving-J.R.-James-Love-Thompson) will receive a substantial amount of playing time as a unit.⁸ Coach Lue should go to the bench 6-7 minutes into the first quarter, giving Irving an early rest to afford Dellavedova time with James, Love, and Thompson. Rest Smith, Love, and Thompson as needed.⁹

Rely on Irving to stabilize the bench unit at the start of the second quarter, because while the Cavs aren’t great with only Kyrie Irving of the Big Three, they’re not an outright disaster like they are are without any of the Big Three or with only Love. Let Irving — fresh off a lengthy rest — terrorize the team’s second unit. Then bring James off the bench to play with a shooters’ unit, then swap Irving for Love. Consider playing Love at center some, depending on how much of a defensive liability it is, before bringing Irving back for the last two minutes of the second quarter.

Repeat the process in the second half — sit Irving early, bring in Delly, and let Delly and James run the offense together. Let Irving do his dribbling/offensive exhibition against the team’s second unit at the start of the third, lean hard on lineups involving Delly-Thompson-James-Love, and play Irving-Thompson-James-Love and either Delly or a wing (J.R. Smith or Iman Shumpert) in crunch time.

I don’t know if this game plan is a good way to maximize the Cavs’ opportunity for success — and none of it may ever approach the Warriors’ Lineup of Death (which is an astounding +46.3 per 48 minutes). But simple tweaks like sitting Irving early in quarters, staggering the Big Three so that Love never plays without James and all three are never sitting at once, and playing small lineups with James-plus-shooters or Love at the five could help the Cavs score points in bunches without suffering too greatly for long stretches.

1. Yes, this is what the individual game pieces are called in Tetris. I did not know this off the top of my head, and I definitely did not spend 300 hours of high school study hall playing Tetris on my calculator. [↩]
2. The result of “13 choose 5.” I think that’s the correct way to calculate the number of lineup combinations. I assume I’m wrong. [↩]
3. I trust NBA.com’s data the most. [↩]
4. Among lineups with over 50 minutes. [↩]
5. For lineups with over 200 minutes. [↩]
6. The Cavs never really played with any “big” lineups. They only played with small lineups (those with one or zero bigs) and two-big lineups — no lineups with three bigs. For example, a lineup with Irving-Smith-James (smalls) and Love-Thompson (bigs) has two bigs. However, a two-big lineup is how traditional positional basketball is played, so it doesn’t make much sense to call a lineup with a power forward and center “big.” Hence, “biggish” means, in other words, “not small.” [↩]
7. I only took data on lineups that had played more than five minutes. That included the top 153 lineups, in terms of playing time, the Cavs played this year. [↩]
8. Notably, the Cavs’ starters have only played together for 465.5 minutes. Over 82 games, that doesn’t sound like much, does it? [↩]
9. Coach Lue has actually already been dabbling with this idea, but only recently. [↩]