There's been lots of discussion around rotations lately, with Lowry's sky-rocketing minutes and the struggles of the starters all season. There have been lots of suggested solutions as well, including from myself. But there have been no detailed suggestions breaking down what a new rotation would look like. Now, yes, a big trade might happen that would change all this (or at least remove a bunch of lineups and pairings as those players may be elsewhere), but until then, we'll continue on assuming nothing changes. And if there is a trade it is still good to know who on the team currently fits together to understand how best to fit any new players into the roster.
I'm going to take a statistical approach (a very simple one) to find the players that play best together, inject some common sense ideas (like DeMar DeRozan and Kyle Lowry both starting, for example), and look at some 5-man lineups that have been very good/bad to re-create or avoid in particular, and hopefully end up with a reasonable rotation idea. There is obviously fluctuation from night to night, but not that much, especially with Casey coaching, as he tends to like to do the same things every night unless a big issue forces his hand.
First, let's state some assumptions. We're looking for a 9 man rotation - so either JJ or Carroll will be in the lineup, but not both. Similarly, for minutes, we will aim to keep averages reasonable - DD around 35-36 minutes, Lowry closer to 32-33 minutes, etc. And of course we will be looking at least a bit at context when creating ideal lineups - if a lineup, such as the Lowry+bench lineup, is very good but has been very good specifically against other benches, then we'll try to put it in mostly as a bench unit, not just make that lineup the starting lineup.
So, step one. Figure out who has worked well with each other on the year.
To do that we take a WOWY (with or without you) approach - find the on-court net rating for a player with a teammate and without that teammate, and use the difference as an "impact" the teammate has on the player's game. Clearly there are lineup considerations, matchups, bench vs starters, and other factors involved here, but this is a good starting point, and we'll make sure to compare full lineups later to ensure we aren't just throwing a hodge podge of players together in ways that don't really work.
Here we go:
That's a lot of data to sort through. Those are just the pairings with at least 100 MP together, too (the list is in order of most minutes played to least minutes played).
Anyway, red is bad, green is good.
Let's start with some obvious ones, the positives or negatives beyond 10 net rating points (looking at the average impact of the two players on each other), and we'll glance at the impacts between 5 and 10 as well.
First, the huge negatives.
DeMar and Lowry (-10.1). Pretty much entirely due to how insanely effective Lowry is with the bench. DeMar is pretty steady with or without Lowry there.
Scola and Lowry (-11.2). Scola absolutely tanks Lowry's effectiveness. And Scola actually plays better away from Lowry too (probably due to him playing better against bench players).
Valanciunas and Lowry (-12.9). Wow, that's a surprise. Again, though, an impact of the terrible starting lineups, and how good Lowry has been with that bench unit. It should be noted that Lowry is only hurt a bit by Valanciunas, and JV is absolutely tanked by Lowry. This makes little sense, and is probably due to the large proportion of their time together spent in the starting lineups, and the small sample of JV's time away from Lowry. We'll file this one under interesting and keep it in mind later.
Valanciunas and Scola (-13.0). Shocker. Scola is about as effective with JV as without him, but JV's production falls through the floor with Scola out there with him. This one has less sample size issues than the Lowry one, as JV has had more significant minutes away from Scola (and Scola has had a lot of minutes away from JV due to his injury).
Biyombo and Carroll (-12.6). This one is from earlier this year - it is basically an effect of that stretch where Casey somehow thought that it was a good idea to leave Carroll out as the only starter with the bench unit. It went poorly, to say the least.
Onto some more reasonable negatives.
DeMar and Scola (-6.2). No surprise here either. Starting lineup strikes again. Both players are equally affected by the other's presence.
DeMar and Bismack (-8.2). Clogging the paint, no surprise DD's game suffers. Big surprise is Biyombo's game suffers more. May be an indication that he struggled when starting while JV was out, may be another side effect of how good that Lowry and bench unit has been.
DeMar and JV (-6.5). Odd that both C's impact DD this way, also odd that DD is less effected by them than they are by him. Lesser negative impact here than with BB.
JJ and Scola (-7.3). Surprise, surprise. Starting lineup strikes again. JJ suffers more than Scola does.
Joseph and Carroll (-8.6). Another indication that Carroll is best not left alone with the bench.
Now, the big positives.
Biyombo and Lowry (+10.7). Yay, bench unit. We'll find a way to get that unit together, for sure.
Lowry and Patterson (+21.7). See above. Although, frankly, we should get these two together as much as possible, look at that impact.
Patterson and Ross (+10.2). Bench unit. No surprises.
Lowry and Joseph (+14.3). Still more bench unit. Will be tricky to maximize point differential without having Lowry out there 48 minutes a game.
Lowry and Ross (+10.0). Getting predictable now. Common theme with all of these is that both players get a lot better (except the Biyombo one, Lowry has a much bigger impact on Biyombo than vice versa).
Joseph and Valanciunas (+14.4). Here's something new. Looks like we should find a way to get both these guys minutes together, which could help offset the minutes Lowry plays with that bench unit.
Patterson and Valanciunas (+20.6). And here's another trend-breaker. Another one of the bench players who has a better net rating away from the bench players (JV is also much better with Patterson than away from him). Let's make sure to find JV some time with those two guys.
And the only moderate positive:
Joseph and Scola (+5.1). Now here's a nice surprise - a player that actually does better with Scola, and vice versa (although Scola is helped more by this pairing than CoJo). This will help finding minutes for the bigger impact guys and our star guards to rest.
Everything else is a wash - guys that can play together, and have been roughly as effective with and without each other on the floor.
So, there we have some pairings to try to recreate as much as possible. Next time we'll look at which full 5-man units have been good and bad to get an idea of the first real building blocks.