Building the best Raptors rotation for 2018-19

Every year we do three things in the off-season in this space. We look at the salary situation of the team, and update that accordingly with each move the team makes. We project win totals for the Raptors and where they will land in the standings.

And finally, we try to construct a reasonable rotation based on some lineups we think will be successful, with as much evidence behind our choices as possible.

Now, with Nick Nurse in charge, and his reported desire to experiment with different looks, including different starting lineups, this will be a little different than usual. We’ll start out by just coming up with some promising lineups to be tried over time. And we’ll examine a few potential starting lineups based on matchups.

At the end, though, we’ll still take our favourite lineups and try to construct a rotation out of them, just to give an idea how the minutes distribution might go — even if there will inevitably be more fluctuation in the rotation this year than in the past.

To do this, we’ll look at what “type” of lineups worked well last year, and we’ll also try out a fun new tool from Jacob Goldstein, the same source as for PIPM that we used in our win projection piece, which attempts to use that data (along with position identifiers, usage rates, assisted field goals, and a bunch of other things) to project how effective a given set of players would be playing together. The average ORTG (and DRTG) the tool assumes seems to be more in line with basketball-reference numbers, while we usually use NBA.com numbers (which are about 3 points per 100 possessions lower on average) so I’ll be adjusting the numbers down manually.

Now, let’s get to it.

What Worked Last Year

Let’s take a quick look at the most used Raptors lineups last season to get a refresher in what worked really well over large samples. These are the sort of lineups that would be considered safe bets to continue to have success this season.

Here’s the list of the most heavily used lineups, along with their total minutes played, offensive rating, defensive rating, and resulting net rating. Offensive rating is points scored per 100 possessions, defensive rating is points allowed per 100 possessions, and net rating is the difference between the two, and is essentially a pace-adjusted point differential.

Lineup | MP | ORTG | DRTG | RTG
Lowry-DeRozan-Anunoby-Ibaka-Valanciunas: 801 MP 112.9 101.7 +11.2
VanVleet-Wright-Miles-Siakam-Poeltl: 340 MP 116.2 99.1 +17.1
Lowry-DeRozan-Powell-Ibaka-Valanciunas: 271 MP 107.2 111.4 -4.2
Lowry-VanVleet-DeRozan-Ibaka-Valanciunas: 101 MP 128.4 103.6 +24.9
VanVleet-DeRozan-Miles-Siakam-Poeltl: 101 MP 121.4 110.0 +11.5
VanVleet-Wright-Powell-Siakam-Poeltl: 100 MP 104.7 97.2 +7.5

Those are the only six lineups with 100 minutes played (by a decent margin, the next closest lineup played only 65 minutes together).

The starters were a rock solid group considering how many minutes they logged and how indiscriminately they were used (they started against every team when they were available, never being subbed in and rarely subbed out for a specific desired matchup). Only three lineups league-wide played as many minutes as the Raptors starters, and the Raptors group had the best net rating of those three. Of the top 25 most used lineups league wide, the Raptors starters had a worse net rating than only four other lineups — and one was the Raptors bench.

The other lineups tell us some interesting information — the Raptors bench was still very good (though not quite as good as the all-bench look) with a star playing in Delon Wright’s place. Norman Powell earned his way out of the rotation, with the bench and starters both posting significantly worse numbers with him present. He’ll really need to earn his way back into the rotation, which will be difficult with so many wings in front of him. And Fred VanVleet was awesome in the closing lineup (in OG Anunoby’s spot with the other starters). These are things we can consider replicating in this season’s rotation.

The Starting Lineup

There are a few options I’ve seen bandied about in terms of who should start for Toronto. Quite a few, in fact, and I don’t doubt we’ll see a few of these as coach Nurse experiments with different looks. But Nurse reportedly described his starting lineup conundrum as follows:

He spoke of a roster that now includes “six starters,” or six players who are accustomed to starting — including Lowry, Leonard, Green, Jonas Valanciunas, OG Anunoby and Serge Ibaka — and talked of the importance of some of those players being open to coming off the bench. — from the The Toronto Star

So to keep things simple, we’ll explore only variations with those six players. Starting with the obvious upgrade to last year’s starting lineup — simply replacing DeMar DeRozan with the superstar player he was traded for: Kawhi Leonard.

Lineup | Projected ORTG | Projected DRTG | Projected RTG
Lowry-Leonard-Anunoby-Ibaka-Valanciunas 121.2 102.4 +18.8

This lineup would functionally work much as last year’s starting lineup did, with Kyle Lowry and Leonard initiating the offense. It has the upside of building off an established successful lineup, and keeps OG in place, where he made a massive difference last year, and where he can continue to play his low usage role without needing to do too much.

That’s already a promising start. The next obvious option is simply swapping in Leonard’s wing-mate from San Antonio in Anunoby’s place, to fill the same low usage, 3-and-D role, but with some added familiarity for Kawhi and more of a true SG for the lineup (which I would argue is not very important).

Lineup | Projected ORTG | Projected DRTG | Projected RTG
Lowry-Green-Leonard-Ibaka-Valanciunas 118.9 102.3 +16.6

This lineup works just the same way as the last. The difference being there is more of an unknown factor replacing two spots from last year’s lineup.

Kind of looking hard to go wrong here, but the first lineup edges this one out if you take the projection seriously. Another idea is to move Serge Ibaka to the bench, bringing OG back into the lineup, but at the PF spot this time — an idea that has already been attached to Nurse.

Lineup | Projected ORTG | Projected DRTG | Projected RTG
Lowry-Green-Leonard-Anunoby-Valanciunas 120.0 101.8 +18.2

Here, again, the idea is to run Lowry and Leonard as the offense creators with Jonas Valanciunas as a screener, while the other two defend and spot up. The difference being this version has a little less size, but a little more flexibility defending the perimeter.

For what it’s worth, that lineup has the best defensive projection so far, though not dramatically different from the other lineups. In the interest of chasing down the top defensive lineup, let’s see what happens if we remove the nominal worst defensive player in that lineup in Valanciunas and go small, risking some rebounding issues but spacing the heck out of the floor.

Lineup | Projected ORTG | Projected DRTG | Projected RTG
Lowry-Green-Leonard-Anunoby-Ibaka 115.9 101.8 +14.1

Well, that’s a move in the wrong direction, but still a very good look. The presumed defensive improvement looks to be off-set by the lack of rebounding, while the lack of a screen setter and inside scorer hits the offense a bit. But these are first world problems — this still projects as a very good lineup.

One other suggestion I’ve seen a lot is to put Pascal Siakam into the starting lineup. Here are some options for that and the projected performance.

Lineup | Projected ORTG | Projected DRTG | Projected RTG
Lowry-Anunoby-Leonard-Siakam-Valanciunas 121.4 102.1 +19.3
Lowry-Green-Leonard-Siakam-Valanciunas 119.1 102.0 +17.1
Lowry-Anunoby-Leonard-Siakam-Ibaka 117.5 102.1 +15.3
Lowry-Green-Leonard-Siakam-Ibaka 114.9 102.0 +13.0

That first one looks great, with Siakam replacing Ibaka. Personally I worry about the spacing in that lineup, and about the removal of some ball handling from the bench unit, but worth keeping in mind.

So we have a lot of very good looks there. Basically no way to go wrong. Personally I lean to the safest option (simply replacing DeMar with Kawhi), and it doesn’t hurt that the projection tool likes it a lot as well.

In any case, let’s move on to some other lineups that need looking at.

Bench Lineups

Here we have some complications, as we don’t know exactly who will be coming off the bench, because of that starting lineup uncertainty. But when looking to build all-bench units (as adding a star to a bench unit has typically been shown to be foolproof, for the most part), we’ll focus on the four obvious bench pieces (VanVleet, Wright, C.J. Miles and Siakam), as well as the presumed backup centres (Greg Monroe and Ibaka), plus Danny Green.

So let’s start from last year’s bench unit and try to re-create it as much as possible. That means we need to test out the two backup centres. Personally, I don’t like the idea of putting the relatively defensively deficient Monroe and Miles on the court at the same time too much, so I’d lean to Ibaka anchoring any unit with Miles in it, while Monroe gets his time with a more stable defender like Green playing the wing, putting four rock solid perimeter defenders in front of him, not too dissimilar to the starting lineup in front of Valanciunas.

Here are the four variations based on changing the wing and centred around VanVleet, Wright, and Siakam.

Lineup | Projected ORTG | Projected DRTG | Projected RTG
VanVleet-Wright-Miles-Siakam-Ibaka: 108.5 104.7 +3.8
VanVleet-Wright-Miles-Siakam-Monroe: 110.8 105.9 +4.8
VanVleet-Wright-Green-Siakam-Ibaka: 106.5 102.4 +4.0
VanVleet-Wright-Green-Siakam-Monroe: 109.0 103.6 +5.4

You’ll notice the defense in all of these is worse than with the starters, which won’t make sense based on how good defensively the bench was last season. Keep in mind, these lineup projections don’t take into account the opposition the lineup will face. So although the defensive grade is a little worse than the starters project to be, the results may be much better, as bench units generally struggle to score relative to starters.

In any case, all four lineups seem fine. Not world-beaters, but good enough that most nights they will be building leads, not losing them. Both versions show Monroe as the superior option, though the difference is smaller in the Miles scenario. And as it seems likely we will see bench lineups with each centre in there, let’s establish the best two-lineup option as Miles-Ibaka and Green-Monroe. Though really there is little difference even if you believe the projections, so again there is no real wrong answer. Except using Powell, who cuts the ratings above in half as soon as he replaces one of the other wings.

Now, we touched on the possibility of using Siakam in the starting lineup, which would imply either Ibaka or Anunoby play PF in the bench units. Here’s a quick list of the variants that could be used.

Lineup | Projected ORTG | Projected DRTG | Projected RTG
VanVleet-Wright-Miles-Anunoby-Ibaka: 109.3 104.5 +4.9
VanVleet-Wright-Miles-Anunoby-Monroe: 111.6 105.7 +5.9
VanVleet-Wright-Miles-Ibaka-Monroe: 110.8 106.2 +4.6
VanVleet-Wright-Green-Anunoby-Ibaka: 107.3 102.2 +5.1
VanVleet-Wright-Green-Anunoby-Monroe: 109.8 103.4 +6.4
VanVleet-Wright-Green-Ibaka-Monroe: 109.1 103.9 +5.2

All good options as well, though the bench units with Siakam are safer bets to continue the bench chemistry from last season. Only the Anunoby versions are noticeably better than the Siakam versions above, and he seems a key to basically any lineup, which means it’s probably best to keep OG with the starters.

Closing/Transitional Lineups

One final thing to look at is how to incorporate some extra ball handling as the Raptors have been wont to do late in games, and generate the best possible lineups with which to close games. It wouldn’t hurt to try to find the best defensive lineups they can use either.

So, let’s start with the same structure we used for the starting lineup — last year’s closers (which were dominant) with Kawhi subbed in for DeMar. And also look at some other variants with VanVleet and Lowry helming the lineup.

Lineup | Projected ORTG | Projected DRTG | Projected RTG
Lowry-VanVleet-Leonard-Ibaka-Valanciunas 119.5 102.7 +16.8
Lowry-VanVleet-Leonard-Siakam-Valanciunas 119.7 102.4 +17.3
Lowry-VanVleet-Leonard-Anunoby-Valanciunas 120.6 102.2 +18.4
Lowry-VanVleet-Green-Leonard-Valanciunas 118.4 102.0 +16.3

And some small ball options as well:

Lineup | Projected ORTG | Projected DRTG | Projected RTG
Lowry-VanVleet-Leonard-Siakam-Ibaka 115.8 102.3 +13.6
Lowry-VanVleet-Leonard-Anunoby-Ibaka 116.7 102.1 +14.6
Lowry-VanVleet-Leonard-Anunoby-Siakam 116.8 101.8 +15.1
Lowry-VanVleet-Green-Leonard-Ibaka 114.3 102.0 +12.4
Lowry-VanVleet-Green-Leonard-Siakam 114.4 101.6 +12.8

And if we want to exclude VanVleet to be able to get more switchability to chase the idea defensive lineup:

Lineup | Projected ORTG | Projected DRTG | Projected RTG
Lowry-Green-Leonard-Siakam-Ibaka 114.9 101.9 +13.0
Lowry-Green-Leonard-Anunoby-Ibaka 115.8 101.7 +14.1
Lowry-Green-Leonard-Anunoby-Siakam 115.9 101.4 +14.5
Lowry-Wright-Leonard-Siakam-Ibaka 118.2 102.3 +15.9
Lowry-Wright-Leonard-Anunoby-Ibaka 117.2 102.5 +14.7
Lowry-Wright-Leonard-Anunoby-Siakam 118.4 102.0 +16.4
Wright-Green-Leonard-Siakam-Ibaka 114.3 102.0 +12.2
Wright-Green-Leonard-Anunoby-Ibaka 115.2 101.8 +13.4
Wright-Green-Leonard-Anunoby-Siakam 115.3 101.5 +13.8

Loads of options, but look out for that Lowry-Green-Leonard-Anunoby-Siakam lineup when they need a stop in a short clock situation where rebounding is less important.

Rotation

Now, let’s see if we can build a functional rotation out of those lineups. The difficult part here will be getting a decent number of minutes for all the quality rotation pieces.

Right away, to make that even remotely achievable, I’m leaving Powell out of the rotation for now. Maybe he earns his way into a spot in training camp, or frequent rest days up and down the roster opens up a role for him, but until then he’s the obvious guy on the outside looking in. I am going to try to get both Ibaka and Monroe minutes as backup centres, to keep Monroe active while also opening up minutes for all of Toronto’s wings.

Let’s start off with the skeleton — the starting lineup, bench lineup to open the second and fourth quarters, and the closing lineup. We’ll just look at the first and second quarters here, and assume they replicate those two quarters in the second half.

Start of 1st Q: Lowry-Leonard-Anunoby-Ibaka-Valanciunas
End of 1st Q: ?
Start of 2nd Q: VanVleet-Wright-Miles-Siakam-Ibaka
End of 2nd Q: Lowry-VanVleet-Leonard-Anunoby-Valanciunas

So, right away, big concern: Green and Monroe aren’t in those lineups. So let’s slot their bench lineup into the end of the 1st quarter.

Start of 1st Q: Lowry-Leonard-Anunoby-Ibaka-Valanciunas
End of 1st Q: VanVleet-Wright-Green-Siakam-Monroe
Start of 2nd Q: VanVleet-Wright-Miles-Siakam-Ibaka
End of 2nd Q: Lowry-VanVleet-Leonard-Anunoby-Valanciunas

Usually the starters run about the first nine minutes of the first quarter, while the bench unit runs about the first six minutes of the second quarter. So already, Ibaka is lined up for 30 minutes per game — that’s a little too much. We can also find some more Green minutes by transitioning to another candidate for the starting lineup and subbing Green in for Ibaka partway through the first.

Start of 1st Q: Lowry-Leonard-Anunoby-Ibaka-Valanciunas
Middle of 1st Q: Lowry-Green-Leonard-Anunoby-Valanciunas
End of 1st Q: VanVleet-Wright-Green-Siakam-Monroe
Start of 2nd Q: VanVleet-Wright-Miles-Siakam-Ibaka
End of 2nd Q: Lowry-VanVleet-Leonard-Anunoby-Valanciunas

That first quarter is pretty solid. Call it five minutes for the first lineup, four for the second, and three minutes for that bench unit at the end of the quarter. The second quarter needs work though. VanVleet is lined up to play the entire quarter, which is probably a bad idea if you want him to have legs under his shot in the final possessions, and Green isn’t playing in those quarters at all and is still only lined up for 14 minutes a night. Monroe is also slotted for only six minutes a night, which is too few.

As a solution, let’s give Monroe a couple minutes right at the start of the second, and transition from the Ibaka bench to the closing lineup by shifting Ibaka to PF and giving Green some extra run (using another of the starting lineup candidates above). I’ve included the projected net ratings for each lineup for reference.

1Q:
Starters: Lowry-Leonard-Anunoby-Ibaka-Valanciunas +18.8
5 minute mark: Lowry-Green-Leonard-Anunoby-Valanciunas +18.2
9 minute mark: VanVleet-Wright-Green-Siakam-Monroe +5.4

2Q:
Start: VanVleet-Wright-Miles-Siakam-Monroe +4.8
2 minute mark: VanVleet-Wright-Miles-Siakam-Ibaka +3.8
6 minute mark: Lowry-Green-Leonard-Ibaka-Valanciunas +16.6
9 minute mark: Lowry-VanVleet-Leonard-Anunoby-Valanciunas +18.4

Valanciunas is penciled into the closing lineup, but those closers will sometimes be anchored by Ibaka or Siakam instead, so this rotation is a sort of “maximum JV minutes” version. Keep that in mind as we look at the minutes totals, which are below.

Full Game:
Lowry: 30
Leonard: 30
Anunoby: 24
Ibaka: 24
Valanciunas: 30
VanVleet: 24
Wright: 18
Green: 20
Miles: 12
Siakam: 18
Monroe: 10

If we assume Ibaka takes half of those six “closing lineup” minutes available, he gets bumped up to 27 minutes per game, while Valanciunas gets bumped down to 27, or even lower on nights where the small looks are needed.

You’ll notice almost every player is seeing fewer minutes than you’d expect. There are just so many rotation quality players. You can see how it would be difficult to jam Norman Powell in there as well, with all the wings and guards already suppressed down to 24 MPG or less (some far less, like Miles who only plays the all-bench minutes at the start of the second and fourth quarters).

This seems like a reasonable starting point assumption on what a functional rotation could look like. You can pretty easily swap most of Siakam and Anunoby’s minutes if that’s what you wanted to do, or Siakam and Ibaka’s PF minutes, and the picture doesn’t change much.

And that rotation projects to an overall net rating of +13.0, which would have ranked first last season by far, well ahead of the Rockets (+8.5). That suggests there might be some optimistic assumptions going into that lineup projection tool.

Looking at the starting lineups Jacob Goldstein posted league wide here, the average starting lineup is projecting at a +7.5 net rating. That suggests that the tool is overrating starters. Last season, looking at the most used lineup for each team, as a proxy for starting lineups (by the way, congratulations to the Phoenix Suns for having their most used lineup only total 155 minutes all season long), suggests that the average starting lineup would be around a +4.2 net rating.

This is probably a competition effect — starters have to play against other starters a lot, so their raw results should be lower than their talent would indicate. The opposite is true of bench lineups — which is supported by the projection tool pegging last year’s bench lineup at only +7.4 net rating, versus their actual +17.1 net rating last season. So if we reduce the effectiveness of the Raptors’ starting and closing lineups by 3.3 points, and increase the effectiveness of the bench lineups by the same 3.3 points, that reduces the overall team net rating to +12.2. Still a ridiculous number.

If the tool is overrating everyone for some reason, and not just starters, then removing 3.3 points from every lineup yields a +9.7 net rating — which, again, would have ranked first in the league by over a point. This does line up nicely with the value from the wins projection piece we did last time out, as a +9.7 rating projects to 62 wins, just above the 59 wins predicted in that piece.

No matter how you slice it, and no matter what specific rotation decisions are made, the Raptors seem built to be dominant for large stretches of every game. I like the rotation set out in this piece, but the biggest takeaway I got from this exercise is fans should not be worried about Nurse experimenting too much with the rotation — almost everything he is likely to try is also likely to work very, very well.

We’ll re-visit this question once we have real data in-season, but going into the year knowing there are probably a lot of right answers (and therefore a lot of flexibility for matchups) is a great place to be.

For now, what lineups do you think will work best? Be sure to check out the lineup projection tool at Goldstein’s Twitter account to try out any combinations not captured here. Are there any lineups posted here you think won’t work as well as projected, or much better? And why?

Lineup stats from 2017-18 per NBA.com.