Building the 2019 Raptors playoff rotation

We tackle this idea every season. Which combinations of players have worked well for the Raptors this year, which have not, and what lineups (or players) should we see (and not see) in the playoff rotation? Heck, I even copied this opening paragraph. We’ve done this before. So let’s get to it.

We’ll start at the beginning — of the game that is.

The Starters

We’ve covered this question a few times this season. The Raptors should certainly start Marc Gasol. All his strengths are enhanced playing with the good shooters and scorers the Raptors start. But let’s check back in with the lineup data and see how it holds up.

Lineup | MP | ORTG | DRTG | RTG
KL-DG-KL-PS-SI: 547 minutes, 113.2 ORTG, 104.9 DRTG, +8.3
KL-DG-KL-PS-MG: 104 minutes, 121.1 ORTG, 110.3 DRTG, +10.7

MP: Minutes Played
ORTG: Offensive Rating, points scored per 100 offensive possessions
DRTG: Defensive Rating, points scored per 100 offensive possession
RTG: Net Rating, ORTG minus DRTG, essential point differential per 100 possessions

Both lineups, with either Serge Ibaka or Gasol in the middle, track as good, which is great news for Toronto. If need be for match-up purposes, the Raptors can comfortably switch between the two. However, the deal for Gasol at the deadline was an attempt to raise the ceiling, and bringing Gasol off the bench probably doesn’t do that as much as starting him. We’ll assume Gasol starts for the purposes of this exercise.

The Bench

This is the section where in the past I explored lineups with one or zero starters. Anyone who has followed the team closely this year knows that’s not a particularly likely scenario this year. The Raptors traded their depth a couple times to consolidate and get Kawhi Leonard and Gasol, and the free agent signings this spring have hardly filled that gap as you’d hope.

For example, although each individual bench lineup has had a minuscule sample (due to the team usually losing those minutes fast), they’ve all been pretty bad, in aggregate. In the 585 minutes the Raptors have played without any of Kyle Lowry, Leonard, or Pascal Siakam on the floor this year, the Raptors have posted a -8.6 net rating (mostly on the, er, weakness of their offense, with a 96.2 ORTG).

So, instead of looking for lineups where all of those guys rest, we’ll look for lineups to buy one or two of them rest. First thing to do, though, is decide how many players we’re looking at. With the troubles the Raptors have had with their bench this year, it doesn’t make sense to go 10 or 11 deep in the playoffs. So, who gets cut from the playoff rotation?

Roster Cuts

Let’s start with the easiest filter: who has been on the court for the worst minutes for the Raptors? The guys whose minutes Toronto always loses are the obvious candidates for not getting playing time. These are filtered for results just since the trade deadline so they aren’t coloured by player combinations that aren’t available anymore.

Player | MP | Net Rating
VanVleet: 278 MP, +17.0
Green: 538 MP, +14.9
Gasol: 512 MP, +12.8
Siakam: 674 MP, +12.2
Leonard: 457 MP, +11.8
Lowry: 561 MP, +9.8
Ibaka: 428 MP, +2.0
Powell: 406 MP, -3.6
Anunoby: 395 MP, -3.9
Lin: 375 MP, -7.2
McCaw: 265 MP, -7.7

Everyone else has played too few minutes for a decent sample and is pretty clearly not in the running for a spot, no matter how awesome Malcolm Miller is or how fun Chris Boucher is.

The top seven players there are pretty obvious fixtures in the rotation, so it’s no surprise they have won their minutes. The Raptors will definitely lean heavy on the starters group with Gasol, plus Ibaka and Fred VanVleet.

The bottom four players there are pretty clearly trying to win the final one or two spots in the rotation. It’s basically impossible to run a seven man rotation, that requires 34 minutes per game from every player, on average, which just doesn’t work for the role players and gets completely derailed with even the slightest foul trouble.

But an 8-man rotation is manageable. And a 9-man is typical. So, let’s look for at least one guy, and hope we can find a second to give the team some more flexibility, even if the ninth guy doesn’t necessarily play every night.

Now, these guys have basement net ratings because of how much they’ve played in those all-bench units, so we’ll turn back to the question we started with. What lineups can Toronto use to get rest for a couple stars at a time? Obviously any variation of these lineups where you roll out two stars instead of one should work even better, so we’ll try to find single-star lineups that work well.

Buying Stars Rest

For each star, we’ll look for a lineup or two that can buy them some rest. Five-man lineups are going to be small samples because we are stuck with the post-trade sample with all that lineup shuffling. We’ll start with looking at who has succeeded with only one of Lowry, Leonard or Siakam on the floor.

Player | Net Rating with only Lowry | Only Kawhi | Only Siakam
VanVleet: +24.0; +23.1; -8.7
Green: +17.1; +26.8; N/A
Gasol: +12.6; +3.3; +9.3
Ibaka: -20.7; +4.7; +4.6
Powell: -10.6; -10.5; -3.0
Anunoby: -12.2; +4.1; -2.3
Lin: -23.1; +3.9; -7.3
McCaw: -2.4; -19.1; -17.7

Some of even these samples are inadequate so I’ve left them out. Pat McCaw and Norman Powell seem to be write-offs here, and Jeremy Lin has only managed to find success in Kawhi lineups, and I don’t know how much I trust that considering the small sample size. That leaves only OG Anunoby, who has only been moderately better than the other guys, but we need at least one extra guy, so he wins by default.

Some obvious conclusions here: Lowry and FVV together is automatically great, it seems. And really, Lowry in any starter-esque lineups, judging by Danny Green and Gasol’s success is usually a good thing. Lowry doesn’t seem capable of carrying a bench unit by himself though, which is a dramatic change from years past. This could perhaps be because of a change in approach, with Lowry playing more of the facilitator than the scorer.

Meanwhile, none of the Siakam-only looks are all that promising. Kawhi at least posts some decent numbers beside Ibaka and VanVleet. Perhaps Lowry and Siakam together could more successfully carry a bench unit? In general, the one star approach seems problematic, with most of the successful looks having a lot of starters involved. Still, Toronto will have to use the bench players at some point. Let’s look at one star resting, rather than pairs of stars resting.

Player | Net Rating without Lowry | Without Kawhi | Without Siakam
VanVleet: +14.6; +20.2; +11.5
Green: +13.9; +15.3; +26.5
Gasol: +21.7; +7.7; +26.0
Ibaka: +9.5; +18.4; +13.9
Powell: +35.1; +3.1; +0.9
Anunoby: +11.5; +17.9; +6.1
Lin: +5.6; -3.4; N/A
McCaw: +11.4; +19.5; -15.0

Ah, much better. The pairing approach looks effective.

Now let’s make a real simplistic rotation. We open with 18 minutes of the starting lineup. And then 10 minutes of rest for each star. That may be a little heavy on minutes for Siakam, but that could be edged down with a few minutes of one-star lineups here and there to fill gaps. So, let’s identify some two-star lineups we trust.

In any of these lineups, we ideally want to use Ibaka at C, and he shows out fairly well across the board above. VanVleet is an obvious entry for any bench lineup. Green would be too, except he needs to sit at some point. So, the top lineups with Ibaka and FVV, and two of the stars. Again, sadly the sample sizes on some of these these are very small due to the constantly changing rotation with all the injuries, rest, and roster turnover.

Lineup | MP | RTG
FVV-DG-KL-PS-SI: 187 MP, +15.3 (Lowry rest lineup)
KL-FVV-DG-PS-SI: 187 MP, +13.8 (Leonard rest lineup)
KL-FVV-DG-KL-SI: 37 MP, +52.9 (Siakam rest lineup)

Now, that’s good. We have some lineups that have looked successful with two of the bench pieces. But to make the minutes load manageable, we really do need to find some looks where we can fit Anunoby into the rotation. It just hasn’t happened consistently with OG playing PF beside Ibaka, leaving the Raptors generally too small in those looks. Those smaller looks tend to be abandoned quickly, barring the all-bench units that tend to fail anyway but serve a regular season purpose in resting the starters. All of which means, the samples are very small.

Still, let’s look at some 4-man groupings to increase that sample size, again holding VanVleet and Ibaka as fixtures, and checking Anunoby’s lineup success beside each star.

Lineup | MP | RTG
FVV-KL-OGA-SI: 59 MP, -12.1 (Leonard lineup)
KL-FVV-OGA-SI: 54 MP, -3.4 (Lowry lineup)
FVV-OGA-PS-SI: 27 MP, +33.6 (Siakam lineup)

OG doesn’t seem to perform well at PF in those looks. But we have seen him have success in more starter-heavy lineups, in much the same role he played last season as the fifth starter. Take a look what happens when you only rest one starter.

Lineup | MP | RTG
KL-DG-OGA-PS-SI: 32 MP, +43.8 (Leonard rest lineup)
KL-DG-KL-OGA-SI: 37 MP, +15.5 (Siakam rest lineup)

Okay, so pieces are starting to come together. We have versions of the starters where OG and Serge can plug in, giving two members of the starting frontcourt rest, Gasol and one of Kawhi or Pascal. We have versions of the starters where FVV and Serge can plug in, giving any one of the stars and Gasol some rest.

Might need one more look to give Lowry some rest — but that one’s easy.

Lineup | MP | RTG
FVV-DG-KL-PS-MG: 69 MP, +22.5

A straight up FVV sub works wonders.

Now we come to the big problem. I’m going to list all the lineups we’ve settled on as effective, and see if you can see a common theme.

Lineup | MP | ORTG | DRTG | RTG
KL-DG-KL-PS-MG: 104 MP, +10.7 (Starters)
KL-DG-KL-PS-SI: 547 MP, +8.3 (Ibaka starters)
FVV-DG-KL-PS-SI: 187 MP, +15.3 (Lowry rest lineup)
KL-FVV-DG-PS-SI: 187 MP, +13.8 (Leonard rest lineup)
KL-FVV-DG-KL-SI: 37 MP, +52.9 (Siakam rest lineup)
KL-DG-OGA-PS-SI: 32 MP, +43.8 (Leonard rest lineup)
KL-DG-KL-OGA-SI: 37 MP, +15.5 (Siakam rest lineup)
FVV-DG-KL-PS-MG: 69 MP, +22.5 (Lowry rest lineup)

See a pattern?

The Green Problem

Yep, Danny Green is in every single one. Now, Green can do a lot of things (as evidenced above), but playing 48 minutes probably isn’t on that list.

Here’s a not-so-fun exercise. I’m going to repeat the list above, but in every line I’ll put the aggregate performance of the lineup with anyone else in Green’s place.

Lineup | MP | ORTG | DRTG | RTG
KL-notDG-KL-PS-MG: 18 MP, +2.3 (Starters)
KL-notDG-KL-PS-SI: 106 MP, -14.6 (Ibaka starters)
FVV-notDG-KL-PS-SI: 96 MP, -3.1 (Lowry rest lineup)
KL-FVV-notDG-PS-SI: 187 MP, +9.0 (Leonard rest lineup)
KL-FVV-notDG-KL-SI: 96 MP, -6.1 (Siakam rest lineup)
KL-notDG-OGA-PS-SI: 21 MP, +32.5 (Leonard rest lineup)
KL-notDG-KL-OGA-SI: 29 MP, -16.8 (Siakam rest lineup)
FVV-notDG-KL-PS-MG: 7 MP, +15.0 (Lowry rest lineup)

Did you notice the same thing I did? Excluding the super tiny sample lineup at the end there, the only two lineups whose performance doesn’t go through the floor with Green sitting are the lineups where Leonard is resting too. The two have carried over a good deal of chemistry from their time as teammates on the Spurs, and keeping them joined at the hip as much as possible seems like a good idea. So, those Leonard rest lineups, let’s take Danny out of those, and see who we can plug in instead.

The first lineup, with FVV in Leonard’s place, has been used with a wide variety of players in Green’s spot, with practically universal success (again, in tiny samples). But the success in aggregate suggests that lineup should work no matter who you plug in.

The second lineup, with OG in Leonard’s place, same idea. Universal success. Tiny samples.

So, the easiest way to make the rotation work is to plug both guys in. Sadly, there is no reliable sample of that lineup. But we’re out of time to build a sample, so lineups we think should work is what we are stuck with.

Therefore, our new set of lineups for the Raptors to use are:

Lineup | MP | ORTG | DRTG | RTG
KL-DG-KL-PS-MG: 104 MP, +10.7 (Starters)
KL-DG-KL-PS-SI: 547 MP, +8.3 (Ibaka starters)
FVV-DG-KL-PS-SI: 187 MP, +15.3 (Lowry rest lineup)
KL-FVV-OGA-PS-SI: No sample (Leonard rest lineup)
KL-FVV-DG-KL-SI: 37 MP, +52.9 (Siakam rest lineup)
KL-DG-KL-OGA-SI: 37 MP, +15.5 (Siakam rest lineup)
FVV-DG-KL-PS-MG: 69 MP, +22.5 (Lowry rest lineup)

That doesn’t really solve the issue. Danny Green has averaged 28 MPG this year. Even if we assume he gets scaled up to, say, 32 MPG — as most of the key contributors will get a minutes increase — we still need 16 minutes of non-Green time. Also, it would be a real bad idea to enter the playoffs with a plan to run a third of every game with a lineup that has basically never played together before.

Let’s ignore the work above for a moment and just take a look at the most used lineups without Green this season to see if any have worked. There are seven lineups without Green (and without any players who aren’t on the team anymore) that played at least 15 minutes.

Lineup | MP | ORTG | DRTG | RTG
KL-FVV-KL-PS-MG: 62 MP, -1.6
KL-JL-NP-OGA-MG: 23 MP, -13.6
KL-FVV-KL-OGA-SI: 21 MP, -5.7
KL-JL-NP-OGA-SI: 20 MP, -19.0
JL-NP-KL-OGA-SI: 17 MP, -16.2
JL-NP-PM-OGA-SI: 15 MP, -34.7
JL-NP-OGA-PS-SI: 15 MP, +7.6

Boy, not exactly a murderer’s row there. Only that last teeny sample looks half decent, and with the general struggles that sort of lineup has shown, I wouldn’t bet on that holding up. The first lineup, with FVV in Green’s spot with the starters, looks like it can tread water, at least, and has a decent sample to suggest it can.

OK, last desperate attempt here. With so little sample from Gasol, let’s expand our look to before the trade deadline, and see if any lineups worked in a large sample with Valanciunas or Monroe, and hope that plugging Gasol into those spots would work out.

There are five lineups with no Green in them that played at least 15 minutes with JV. They all had negative net ratings. There were actually six lineups for Monroe, and five of them had very nice positive net ratings — but every one of them also had Delon Wright in them. Nothing to be had there.

So, what are we left with? We have the Leonard rest minutes, which will buy Leonard his 10 minutes of rest each game with a unit we’ve never seen before but which theoretically should work. We have a quasi-starting group that gives Green his extra six minutes of rest by plugging FVV in and hopefully keeping afloat. That might be enough. But it’s tight. Let’s grab one single-star lineup to try to make everything fit. We’ll make it a Lowry rest lineup.

The Rotation

Okay, let’s wrap up then. We’ll use the following lineups to try to build a rotation.

Lineup | MP | ORTG | DRTG | RTG
KL-DG-KL-PS-MG: 104 MP, +10.7 (Starters)
KL-DG-KL-PS-SI: 547 MP, +8.3 (Ibaka starters)
FVV-DG-KL-PS-SI: 187 MP, +15.3 (Lowry rest lineup)
KL-FVV-OGA-PS-SI: No sample (Leonard rest lineup)
KL-FVV-DG-KL-SI: 37 MP, +52.9 (Siakam rest lineup)
KL-DG-KL-OGA-SI: 37 MP, +15.5 (Siakam rest lineup)
FVV-DG-KL-PS-MG: 69 MP, +22.5 (Lowry rest lineup)
KL-FVV-KL-PS-MG: 62 MP, -1.6 (Green rest lineup)

Let’s walk through an example substitution pattern that can get us close to an ideal rotation. Nurse will likely wait longer for the first substitution, and end up with more bench-heavy units on the floor. Personally I prefer the following.

1st Q: KL-DG-KL-PS-MG
4-minute mark: KL-FVV-KL-PS-MG
8-minute mark: KL-DG-OGA-PS-SI
10-minute mark: FVV-DG-KL-OGA-SI
2nd Q: FVV-DG-KL-OGA-SI
2-minute mark: KL-FVV-OGA-PS-SI
6-minute mark: KL-DG-KL-OGA-SI
8-minute mark: KL-DG-KL-PS-MG

That makes for the following minutes breakdown, assuming the second half is the same as the first.

Lowry: 40
Green: 32
Leonard: 36
Siakam: 36
Gasol: 24
FVV: 24
Anunoby: 24
Ibaka: 24

A more complex breakdown with more staggered subs would get a few of those a little lower, a few a little higher, e.g. Gasol can certainly slot into a few of those Ibaka lineups to skew the minutes towards Gasol depending on match-ups. But that’s basically the outline of what I’d be thinking.

********

What do you think? Should the Raptors chase a staggered short rotation like this in the playoffs? Or do you have faith they can make one star or all bench units work?