clock menu more-arrow no yes

Filed under:

Numbers Game: What’s going on with the Rotation?

New, comments

We’ve seen some pretty nonsensical Raptors lineups over the recent slump. Let’s try to figure out what we should be seeing instead.

NBA: Orlando Magic at Toronto Raptors Nick Turchiaro-USA TODAY Sports

I write a piece like this one every year, usually in the run up to the playoffs, when the Raptors are meant to be fine tuning their rotation for prime time. But with the team scuttling of late, and, in my opinion, much of that coming from poor rotation choices, I thought I’d roll this out to see if we can find some reasons to be more optimistic about this team as constructed.

Over the most recent stretch, each game has, for the most part, had at least one big glaring rotation error that cost the team.

vs ORL - 1 point loss. A disastrous lineup of Valanciunas, Siakam, Carroll, DeRozan and Joseph (possibly the result of an experiment intended to find the worst fitting lineup possible) played 5 minutes and was -17.
vs MIL - big win, no glaring issues.
@ MEM - 2 point loss. JV and Sullinger played 10 minutes together, were -5. Lowry, Joseph and Van Vleet were all on the floor together for about 2 minutes and were -7 in that time.
vs SAS - 2 point loss. The starting lineup with Sullinger at PF was -6 on the night.

Small mistakes, for mostly short durations. But when you lose three games by a total of 5 points, the small mistakes add up quickly, and these small mistakes are pretty avoidable.

But the mistakes are obvious in hindsight. Going forward, are there combinations that should be sought after, and some that should be avoided? Let's take a look.

With or Without You

A common approach in stats for several sports, most predominantly hockey but also catching on with basketball of late, is to judge player's impacts on their teammates. This is done by looking at each player's performance with and without each of their teammates on the floor (known as WOWY’s, or With or Without You stats). What we'll do is look at the team's net rating (point differential per 100 possessions) while these players are on the floor together (and apart) to judge that impact.

There is lots of context here — some players play a lot together, so one's impact on another teammate will look similar to the other's, for example. For now, let's not worry about that and just look at the numbers.

With You

First, the "with" data set — how the team performs with certain pairings of players on the floor. The following chart shows each player with significant playing time so far on the left column, and the players they have been paired with along the top row. The numbers in the chart are the team's on-court net rating with that pairing on the floor — colour coded from very good (green) through average for the team (yellow) to terrible (red). The values for players “with” themselves is simply the player’s overall on-court net rating.

I’ve left Jared Sullinger off, as he’s got tiny sample sizes, and has been awful pretty much with everyone, so his results made the colour chart mostly useless. I’ve also left off guys like Jakob Poeltl and Fred Van Vleet because of sample sizes.

You’ll notice a few patterns right away. A couple of nice green columns for Patrick Patterson and Lucas Nogueira, who have both been tremendous with everyone they’ve played with. Nogueira’s samples with the starters (Jonas Valanciunas and DeMarre Carroll in particular) are mostly driven by his minutes at PF. There’s a pretty consistent orange-red column for Pascal Siakam, and another one for Cory Joseph (though less consistent).

Without You

To really get an idea of who is actually impacting each others’ play though, we also need to look at how each player does away from each teammate. The following chart shows the same thing as the above chart, but without each teammate instead of with them.

Again, some obvious patterns. Everyone plays pretty terribly when Kyle Lowry sits, and the same goes for Patterson to a lesser degree. Several players do well without Joseph (which makes sense since they do so poorly without Lowry). Most others are pretty neutral, with everyone doing at least OK without DeMar DeRozan, Valanciunas or Siakam on the floor (driven mostly by those effective Lowry plus bench units, no doubt, and the ineffectiveness of that primary starting lineup).


But what we really want is the net impact each player has on each teammate. So we find the difference between the player’s performance with each teammate and his performance without that teammate, and we get the following results.

Now we get a pretty clean picture. Lowry is fantastic across the board. Patterson is solid for some, great for others. Same for Nogueira. As expected, Siakam is a disaster for many, but surprisingly not so much for some. Joseph is predictably rough for everyone what with him being Lowry’s direct substitution. Players like Valanciunas, Carroll, Terrence Ross and DeRozan are roughly average (with DeRozan skewing particularly negative due to the relative ineffectiveness of his bench unit compared to Lowry’s), but have occasional very good fits and occasional very bad ones.

Some stand out conclusions based on the above, before we move on to some lineup construction.

There is one player on the roster who does better away from Lowry than with him, Pascal Siakam, the guy the Raptors shoehorned into the starting power forward role beside Lowry for most of the season so far. It’s hard to find guys who can perform without Lowry on the court, so Siakam should be used almost exclusively without Lowry. Siakam’s impact on Lowry is both a) the worst impact of any player on Lowry and b) the worst impact of Siakam on any player, so yeah, let’s not play Siakam with Lowry ever again. His impact is so terrible it can be seen bleeding into the impacts the other starters have on each other.

But his impact is not terrible for everyone. It’s just poor for Joseph and Norman Powell, and is actually positive for Ross and significantly so for Nogueira. Ross, Nogueira and Powell all have positive impacts on Siakam’s play, the former two to a pretty significant degree. All signs point to playing Siakam with the bench units being a good idea.

There are all the obvious ones we’ve seen all year, like how significant Patterson’s impact is on the starters compared to the bench.

One concerning thing that will be hard to navigate around is just how much Joseph has struggled with basically everyone this season. He’s been OK with Lowry, but by all accounts he absolutely cannot play with DeRozan or Valanciunas (with his impacts on those players, and theirs on him, all worse than -12). This defeats the purpose of a backup point guard, especially if you want to stagger rotations to keep an all-star on the court. But perhaps you don’t stagger them, and run some bench lineups with all the players Joseph has done the best with — like Powell, Ross, Siakam and Nogueira. The offence might come in spurts, but the defence should be pretty good with a lineup like that. It’s possible Delon Wright’s return could unlock some possibilities but we have no data on his impact, of course.

We’ll get deeper into full five man unit construction next week, using some three-man lineup samples and trying to find enough units to build a rough rotation to get an idea of how many minutes each player would have to play. But for now, let’s settle on some simple ground rules based on the above.

Three Rotation Conclusions

  1. Play Siakam next to JV, Carroll, DD or Lowry as little as possible, but definitely never with more than one at the same time, let alone with all of them (and ideally literally zero minutes ever with Carroll or Valanciunas). Find him time with the bench units if Sullinger is still unable to play basketball (although I should note: the one player Sullinger has a positive “with” value beside is Nogueira, so if you have to use him, use him exclusively as the bench PF in Siakam’s place).
  2. Start Patterson, and don’t stop. He really unlocks the other starters’ games, and the bench unit’s productivity seems far less reliant on his presence.
  3. Play Joseph as little as possible with DeRozan and Valanciunas. DeRozan is hard to avoid (maybe try some no-star bench units for short stretches to see if they survive — it’s not like the DeRozan plus bench unit is holding its own, anyway), but never put JV out there with Joseph, and certainly not with both Joseph and DeRozan. Save his time for playing with the starters, who should also be the closers, by the way, since Joseph has struggled so much all year.

I think that’s enough for this week. Do you disagree with any of the above? Have any commentary on any surprising or particularly egregiously bad (or good and underused) pairings I missed? Let me know what you think.

All stats per