clock menu more-arrow no yes mobile

Filed under:

Numbers Game: Gauging Potential Raptors Lineups

Which lineups will the Raptors find the most success with this season? Welcome to Numbers Game.

NBA: Toronto Raptors-Media Day Peter Llewellyn-USA TODAY Sports

CBS recently posted an article discussing what each team’s most potent lineups could be around the league. Here’s a look at the Raptors’ entry.

Kyle Lowry - Cory Joseph - DeMar DeRozan - DeMarre Carroll - Jonas Valanciunas

This lineup only played 36 minutes last season because Carroll only played in 26 regular-season games. If he's able to stay on the floor, Toronto might end up relying on this group to close games. Coach Dwane Casey loves having Joseph's extra playmaking on the court in those situations, and Carroll gives the Raptors much-needed versatility and floor spacing at the 4. -- James Herbert

Herbert is right — that lineup was potent last season in the small glimpse we got of it. But let’s go deeper than one lineup, let’s try to find a few. Once you have a few effective lineups (that are different enough from each other), you can start to build a rotation around them. So, let’s play the numbers game.

The Closers

We’ll start here, with the lineup suggested by Herbert above. The lineup only played 36 minutes together, as noted, but posted a tremendous 124.9 ORTG (points scored per 100 possessions) and decent 104.5 DRTG (points allowed per 100 possessions), good for a very impressive +20.4 Net Rating (ORTG minus DRTG). But that’s a very small sample (I usually use 50 MP as the bare minimum to judge lineup data once the season is well under way). It’s not very reliable.

But look at what makes that lineup what it is. You have the three stars of the team, a super-sub ball handler off the bench in Joseph, and a quick floor spreader (who is undersized and a poor rebounder for the PF spot) in Carroll. There aren’t any players who replicate what the stars bring, and Joseph is unique in his effectiveness on the team. But what lineup did the Raptors often close with when Carroll was hurt?

They found another poor rebounding, quick defending floor spreader to play the 4 in that lineup. Yes, insert Patrick Patterson in Carroll’s place and you get a much bigger sample and another solid result. With Patterson, the closing lineup posted an even more potent 125.6 ORTG and a somewhat less impressive 111.8 DRTG, good for a +13.8 Net Rating in 97 minutes played. That’s still a very impressive lineup. Combining the two into a full sample for these sorts of closing lineups gives you 133 minutes, a 125.4 ORTG, 109.8 DRTG and a net rating of +15.6.

Last season, the Raptors only had two lineups that posted a better net rating in over 100 minutes played, and both were bench units playing against inferior opponents. This lineup, whether you have Carroll or Patterson in there, looks like a keeper.

Speaking of bench units...

The Bench

The top three lineups on the team last season (with over 100 minutes played together) were all bench lineups that featured one or both of the star guards. All of them featured Patrick Patterson, Cory Joseph and Bismack Biyombo. And then some combination of Kyle Lowry, Demar DeRozan and Terrence Ross.

The resultant ratings were as follow:

KL-CJ-DD-PP-BB: 142 MP, 115.8 ORTG, 97.5 DRTG, +18.3 net rating
KL-CJ-TR-PP-BB: 297 MP, 111.1 ORTG, 94.7 DRTG, +16.4 net rating
CJ-DD-TR-PP-BB: 218 MP, 104.7 ORTG, 91.0 DRTG, +13.7 net rating

Last season, those three lineups were the only heavily used lineups (>100 MP) to post a positive net rating. The other six heavily used lineups (mostly variations on the starting lineup) got outscored by the opposition.

Those bench lineups all achieve the same thing — outscoring the opposing bench unit by a significant margin — in different ways. The top lineup, with two star guards, has solid defence and blistering offence. Lowry with the bench posts an impressive offence and defence, and DeRozan with the bench posts an incredible defence and solid offence.

In any case, the concern is that Biyombo is no longer with the team, so there’s no cut and paste option for this season. The team will need to find lineups that replicate this success without Biyombo.

There are two obvious places to look: the new backup centres. For now we will leave Poeltl aside, as we have no real data on him, and focus on Jared Sullinger and Lucas (Bebe) Nogueira.

Nogueira is easy: let’s see what those lineups looked like in the small samples where he played in Biyombo’s place. It will be in limited minutes, so not entirely reliable, but should give us an idea if the lineup is performing similarly.

KL-CJ-DD-PP-LN: 9 MP, 154.0 ORTG, 99.6 DRTG, +54.4 net rating
KL-CJ-TR-PP-LN: 26 MP, 104.9 ORTG, 96.2 DRTG, +8.7 net rating
CJ-DD-TR-PP-LN: 19 MP, 124.2 ORTG, 118.8 DRTG, +5.4 net rating

Again we have a bunch of small samples (with some silly looking numbers; obviously you can’t expect a +50 net rating from that top lineup). So let’s quickly combine the samples into a catch-all “Bebe with the bench” lineup data set.

CJ-KL/DD/TR-PP-LN: 54 MP, 119.9 ORTG, 104.7 DRTG, +15.2 net rating

Now that’s a more reasonable sample (barely), but the results look very similar to the bench lineups above, with a heavier skew to offence than defence, but not with terrible defence, just average. So even if Nogueira-led lineups see a fall off defensively from the Biyombo-led lineups of last season, the overall trend of outscoring opposing benches seems like it could continue.

But Nogueira, in all likelihood, will be playing only a fraction of the backup C minutes. What about Sullinger?

We won’t be able to go quite as in-depth here, as he played in an entirely different system with different teammates, but let’s take a quick look at his top lineups and see if there is a pattern for when he was most effective.

Among lineups with over 50 MP that included Sullinger, the most effective ones were:

Bradley, Crowder, Olynyk, Sully, Thomas: +20.8 net rating
Bradley, Crowder, Turner, Sully, Thomas: +19.4 net rating
Bradley, Johnson, Turner, Sully, Thomas: +11.2 net rating

See any archetypes that fit? Ball dominant scoring guard? Lowry and Thomas. Defensive guard? Joseph and Bradley. Stretch big or small ball 4? Patterson and Amir/Crowder. Sometimes another ball dominant scoring guard? DeRozan and Turner. 3+D wing? Ross and Crowder. In those lineups, Sullinger is often the only real inside presence on the boards and in the post. He is playing with several bench players in each lineup, but also at least one other starter, similar to how he’ll be used with the bench here.

Although we can’t state concretely that he will fit, all signs point to him being a positive contributor to the bench, and hopefully able to replicate the success the Raptors’ bench units had last year, and the success Sullinger’s most used lineups had last year as well.

The Starters

The starters are easy, Sullinger will slot into Scola’s place in all likelihood. But with Sullinger playing large minutes with the bench, we’ll probably see a lot of Patterson in Scola’s spot as well.

For Sullinger, once again we lack direct lineup comparisons. But from a simple player quality standpoint, he is head and shoulders above Scola, by basically any metric (superior PER, WS/48, BPM). And the Raptors struggled big time defensively with Scola on the floor, so even if Sullinger is not an ideal fit on that end, he’s destined to be a better fit than Scola was.

Some will say that the starting lineup’s defensive struggles might have had to do with Valanciunas rather than Scola, and to that I present the following 4-man lineup stats (as the starting SF varied so much through the year).


Yeah, it was Scola. For a quick estimate of impact, Scola’s defensive box plus minus (BPM, an estimate of a player’s impact on the floor) was -0.4, while Sullinger’s was +2.3, which suggests plugging Sullinger into those lineups will immediately improve the defensive rating by almost 3 points. It doesn’t work out quite so cleanly, but there’s every reason to expect improvement here.

As for the minutes where Patterson plays with the starters, that’s another opportunity besides the bench units where the team can make up some real ground on their opponents. Every one of Patterson’s lineups from last season with all three of DeRozan, Lowry and Valanciunas are listed below.

KL-CJ-DD-PP-JV: 97 MP, +13.8 net rating
KL-DD-TR-PP-JV: 85 MP, +30.2 net rating
KL-DD-JJ-PP-JV: 33 MP, +6.3 net rating
KL-DD-NP-PP-JV: 20 MP, +9.7 net rating
KL-DD-DC-PP-JV: 16 MP, +41.4 net rating

That last line item looks very promising, but as ever with Carroll we are left with a small sample. But every single one of those lineups rated out as a significant to incredible positive impact lineup, and with all three of DeRozan, Lowry and Valanciunas playing, the opposition probably had their best players lined up against them. And removing the small sample by looking at the 4-man lineup of Lowry, DeRozan, Patterson and Valanciunas shows us a nice set of numbers that promise big things from that combination of players this year.

KL-DD-PP-JV: 251 MP, 127.4 ORTG, 107.9 DRTG, +19.5 net rating

That would rank nearly as good as the best bench lineup from last season (the one where both DeRozan and Lowry were on the court) and better than both bench lineups that played major minutes (Lowry-plus-bench and DeRozan-plus-bench). And we do have small sample lineup data that says that whether Ross, Carroll or Powell fill that 5th spot, the lineup has had success, ranging from moderate to amazing, in the past.

That’s where we’ll leave it for now. We can say with some confidence that the Raptors have a few lineups worth trying to build a rotation around. We’ll re-visit lineup data and other numbers as the season starts up in later editions of Numbers Game.

It will be interesting to see what lineups are rolled out early in the season — remember, last season started with Scola starting (which persisted all year, unfortunately) and bench units with neither of Lowry or DeRozan in them (which were disasters and thankfully were corrected by December).

All lineup stats per and all individual player stats per