Quick Stat Hits: Constructing a New Rotation - Part 2

Last week we took a look at a WOWY analysis of the Raptors' player pairings, and ended up with a list of great, good, bad and terrible pairings. Let's summarize those for reference and look closer at a few to make sure they are real.

The Great:

Biyombo and Lowry (+10.7)
Patterson and Ross (+10.2)
Lowry and Joseph (+14.3)
Lowry and Ross (+10.0)
Lowry and Patterson (+21.7)

Bench unit. Enough said.

Joseph and Valanciunas (+14.4)
Patterson and Valanciunas (+20.6)

These two work really well with JV. That's two starters now that Patterson and Joseph show big positive impacts with. Hard to start Joseph, but we can fit him into closing lineups easily enough. But with those massive Patterson numbers we're going to want to find even more minutes for him.

The Good:

Joseph and Scola (+5.1)

The Bad:

DeMar and Scola (-6.2)
DeMar and Bismack (-8.2)
DeMar and JV (-6.5)
JJ and Scola (-7.3)
Joseph and Carroll (-8.6)

One note on the DD and JV pairing - JV has little to no impact on DD, but DD drags JV way down. This appears to be an effect of the very limited minutes JV has played away from DeMar (98 minutes total) skewing the numbers. We're going to disregard this pairing data as it looks to be a trick of small sample sizes.

The Terrible:

DeMar and Lowry (-10.1)
Scola and Lowry (-11.2)
Valanciunas and Lowry (-12.9)
Valanciunas and Scola (-13.0)
Biyombo and Carroll (-12.6)

Again, we get a small sample issue with the JV/Lowry pairing. Even worse this time as JV has played only 90 minutes away from Lowry this season (and done very well, hence the big negative skew with Lowry there). We'll disregard this one as well for the same reason. We'll also mostly disregard the Lowry/DD pairing because, let's face it, they are playing a lot of minutes and a lot of them together, but with the approach being to have one on the court at all times, we'll have them apart from each other as much as possible anyway.

Three Man Lineups

Moving on, we'll take a very quick look at the very best and very worst three man combinations on the team (with over 100 minutes played together) just to get an idea what groups of players are good as a unit that the WOWY data didn't show.

Here are the top 10 and bottom 10 three man units in net rating.

Quickly, to focus on things that were not obvious from the WOWY data:

Patterson, Ross and JV have been incredible together. But I'd hesitate to play them all against starters - that's a lot of rebounding to put on JV alone, and last season the three were very bad together in a bigger sample.

Lowry, Patterson and JV are also amazing together, which is very promising as that would mostly come from closing lineups or early 1st Q subs, which would mean they are playing mostly against starters. This has to happen.

DeMar, Patterson and JV. Here's a nice surprise, as DD has been pretty neutral in the WOWY analysis. Good to find a combination where he appears to really thrive.

DeMar, Joseph and JV. Again, with Lowry expected to play so many minutes with various bench players, it is good that DeMar and JV can be effective beside the team's backup PG.

The rest of the positive lineups are from that Lowry plus bench unit.

The nice thing about the negative lineups is they are all pretty close to neutral. The only real new information is that DeMar appears to fit poorly with bench players (including Scola, which is concerning) for the most part. That said, a good chunk of those three man lineups appear to have come from the unfortunate period where Biyombo was starting beside Scola, a pairing that should have far more luck against bench units than starters.

Lineups

With all that information, we can start building lineups. First, and this is the change I'm sure you all saw coming, we're going to start Patterson. His impact numbers with JV and DeMar are too good not to give this a shot.

Next, we are obviously going to try to keep that Lowry and bench unit together.

Then it gets trickier. We need lineups that will bridge the gap between the bench unit and the starters, and particularly we need a lineup or two that do not have Lowry or Patterson in them as both are in each of the other two main lineups.

Obviously DeRozan and Joseph will be key there, as each are the primary guard missing from the two main lineups. Let's look at the players they play most often with as a pairing.

The good news is every three man lineup has a positive rating with these two, so no matter what we do this lineup isn't going to kill us, and we have an opportunity to roll out a couple versions of this lineup to help spread out minutes. Unfortunately there haven't been many 5-man lineups with both DD and Joseph, but without Patterson or Lowry, playing many minutes, so we'll have to take a common sense approach as well as rely on the WOWY and three man lineup data.

We need a lineup without Patterson, and a lineup without Lowry, and ideally a lineup without both.

For a lineup without Lowry, we have the DeMar plus bench lineup, which has been very good as well. They've played 155 minutes total, and have a +9 net rating. We can write that one in as an option. We can also use a variant of that lineup with JV in for Biyombo.

For lineups without Patterson, we have the closing lineups we used earlier in the year, with Carroll at PF. That lineup has only played 36 minutes across 9 games due to JV and Carroll missing so much time but it was working really well (+20 net rating), so we'll include that. No reason it can't be used earlier in the game, too.

Now, for a lineup without either. Oddly, here is where the JV/Scola matchup works - with lots of big defenders in the backcourt. There are very few lineups that even hit double digits in minutes played without either of Lowry or Patterson, so we don't really have 5-man data. But using the above three man data and some common sense, we'll settle on a lineup of Joseph-DeRozan-Ross-Scola-JV as the ideal, with Biyombo in for JV depending on how heavy JV's minutes are coming in at the end of this.

Okay, so for now we are left with these lineups to build a rotation with.

1 | Kyle Lowry - DeMar DeRozan - DeMarre Carroll - Patrick Patterson - Jonas Valanciunas
2 | Kyle Lowry - Cory Joseph - Terrence Ross - Patrick Patterson - Bismack Biyombo
3 | Kyle Lowry - Cory Joseph - DeMar DeRozan - DeMarre Carroll - Jonas Valanciunas
4 | Kyle Lowry - Terrence Ross - DeMarre Carroll - Luis Scola - Jonas Valanciunas
5 | Cory Joseph - Terrence Ross - DeMar DeRozan - Luis Scola - Bismack Biyombo
6 | Cory Joseph - Terrence Ross - DeMar DeRozan - Patrick Patterson - Jonas Valanciunas

Lineup 2 has a massive sample and a very positive net rating. It's a great lineup.

Lineups 1, 3, 4, 5 and 6 do not have big samples at all. These lineups project to have good performances based on the WOWY and three-man lineup data reviewed above and in the last piece, and most of them show very good net ratings in the limited minutes they've played.

There are also some common sense goals - try to avoid players playing too many minutes without breaks, try to keep players' total minutes reasonable (including scaling back Lowry and DeMar's minutes, and keeping Carroll's minutes low to avoid re-injury and to accommodate him getting back into shape).

In terms of the timing of subs, they will vary, but we'll construct a rotation based on the NBA's TV timeout guidelines, and Casey's typical substitution pattern - 6 minutes and 9 minutes into each quarter, as well as 3 minutes into the second and fourth quarters, though we won't use all of those opportunities to actually make a sub. Here we go.

Start of 1Q: KL-DD-DC-PP-JV (starting lineup)
6 minute mark: KL-TR-DC-LS-JV (Scola in for PP, Ross in for DD, maintains spacing, gives Patterson a breather)
3 minute mark: CJ-DD-TR-LS-BB (Biyombo in for JV, DD and Joseph in for Lowry and Carroll)

Start of 2Q: KL-CJ-TR-PP-BB (Lowry "plus bench" unit)
6 minute mark: CJ-DD-TR-PP-JV (JV and DD back in for Lowry and Biyombo)
3 minute mark: KL-CJ-DD-DC-JV (closing lineup to close out the half)

Start of 3Q: KL-DD-DC-PP-JV (starting lineup)
6 minute mark: CJ-DD-TR-PP-JV (Joseph and Ross in for Lowry and Carroll)
3 minute mark: CJ-DD-TR-LS-BB (bench lineup with DD)

Start of 4Q: CJ-DD-TR-LS-BB (bench lineup with DD)
9 minute mark: KL-CJ-TR-PP-BB (Lowry "plus bench" unit)
6 minute mark: KL-DD-DC-PP-JV (starting lineup)
3 minute mark: KL-CJ-DD-DC-JV (closing lineup)

Final minutes count:

Lowry: 33
DeRozan: 36
Carroll: 24
Patterson: 30
Valanciunas: 30
Joseph: 30
Ross: 27
Scola: 12
Biyombo: 18

Carroll can also take a few of Ross' minutes, as he fits wherever Ross does, if he proves able to handle more minutes (or is needed against a playoff opponent like LeBron James). We get 9 minutes of that Lowry-with-the-bench lineup, up from 6.6 per game now. Scola only plays 3 minutes beside JV, 3 minutes beside Lowry, and only 9 minutes beside DeRozan.

Only instances of a player playing an entire Q are DeMar in the 3rd Q and Joseph in the 2nd Q.

There are lots of variations you can make on a game-to-game basis with that rotation as a baseline, and even if there are nights where you need a few extra minutes from Lowry, he's getting bumped up to 36-37 minutes rather than into the 40's.

What do you guys think? Is there a weakness in that rotation? Is there a lineup you don't like, or one you do that you don't see here? Does this fix the current problems or does it create new ones?

All stats per NBA.com.