clock menu more-arrow no yes

Filed under:

Quick Stat Hits: The starting power forward question that just won't quit

New, comments

The Raptors have started off pretty well, the game against the Heat notwithstanding. So why change anything? Let's take a look.

Robert Mayer-USA TODAY Sports

This is an issue I raised last week as a note, but I think it deserves a more in-depth look. Luis Scola has simply not been working with the starters. I've got all sorts of theories about why (his need for the ball, his footspeed combined with Jonas', his less-than-prolific shooting) but that's just theorizing. Let's look at some facts.

Here are the on-court net ratings for the usual starters:

Jonas Valanciunas: +11.7 (he's been amazing and we'll get to him another week)
DeMar DeRozan: +8.0
Kyle Lowry: +7.2
Luis Scola: +3.8
DeMarre Carroll: -1.7

That's five guys close to breaking even or killing it. And here is what it looks like when they play together:

KL-DD-DC-LS-JV: -1.2 (in 82 minutes played)

See, that makes no sense. These guys are mostly performing very well and yet together they don't break even.

Let's look at the pairings of the players involved (presented in order of most minutes played to least).

KL-DD: +8.5
KL-DC: +5.6
KL-JV: +6.5
DD-JV: +13.7
DD-DC: +2.7
DC-JV: +6.5
KL-LS: +0.6
DD-LS: +5.7
LS-JV: +5.4
DC-LS: -0.2

So, what does that mean? To start, this all filters out to Lowry being good no matter what, unless he plays with Scola; JV being good no matter what; Carroll being at least decent no matter what, unless he plays with Scola; and DeMar being good no matter what.

There's a couple "unless he plays with Scola" bits in there that should be concerning.

Now, let's take a quick left turn, and glance at a few of our bench units (again, in order of most minutes played to least).

CJ-TR-DD-PP-BB: -14.1
KL-CJ-DD-DC-JV: +41.1
CJ-TR-DC-PP-BB: -56.8
KL-DD-DC-PP-BB: +20.4

Just a few quick observations here. There are some lineups that are insanely good, and some that are insanely bad. These will normalize over time to a more reasonable number, but we can say that the bad lineups aren't working. But why?

Well, first glance tells us that Patrick Patterson plays a lot better with Lowry than with Cory Joseph. And that Carroll does not fit as the starter to stay on the floor with the bench. No surprise there - look at the bench: Terrence Ross and Patterson are well entrenched as off-ball, shoot-'em-up gunners. Adding another one doesn't help, especially with Bismack Biyombo being completely ineffective on that end in any capacity. That leaves CoJo on an island, or worse, leaves Patterson or Ross or Carroll to try to be playmakers, and that tends to end very badly.

So, let's take a look at player pairings for Scola and Patterson and see if we can find a solution here.

First, Scola's pairings:

KL: +0.6
DD: +5.7
JV: +5.4
DC: -0.2
CJ: +26.9
TR: +13.2
BB: +19.5

Scola has played about four times as many minutes with the starters as with the bench, but even in a small sample those bench numbers look really good.

Now, Patterson's pairings:

CJ: -7.5
BB: -6.5
TR: +1.1
DD: +16.1
KL: +16.4
DC: -9.4
JV: +28.7

Now, Carroll's stats are coloured by that insane and insanely overused lineup with all three of Carroll, Ross and Patterson on the floor together. 33 of the 72 minutes DC and PP have played together had Ross sitting beside them, and all three combine for a -25.5 net rating, meaning in minutes where Ross isn't there as well, DC and PP have a net rating of +4.2. What these numbers suggest is that Patterson plays well-to-extremely-well with all of the starters and much worse with the bench.

For every 5-man lineup, there are 10 two-man combinations (the bench is essentially four players, so there will be six combinations). If you average the two-man combinations' net ratings, you should get an idea how effective that lineup will be together. So let's take a look at the lineups - Casey's version, and what I would argue is infinitely more logical.

KL-DD-DC-LS-JV: average net rating of all player pairings = +5.5
CJ-TR-PP-BB: -10.6

KL-DD-DC-PP-JV: +9.5
CJ-TR-LS-BB: +1.5

And that doesn't even include the benefits of the other bench pairings and starter pairings probably improving since they'd be playing in better lineups (for example, the Ross-Joseph pairing may be positive if Ross isn't forced to play on-ball as much).

Now, this is not foolproof, there will be some adverse effects on Patterson if he plays against starters (though you'd expect a similar boost to Scola against bench players). But I think this shows that it is definitely worth trying, right?  Seems like almost overwhelming evidence to me.

As an added note, with Carroll resting, it just makes all that much more sense with James Johnson playing. You need that spacing Patterson provides if JJ is in the game. And the early returns for Johnson (small sample warning, but still: +36 net rating with PP, -14.5 with Scola) support that idea as well. It also provides a nice excuse to make the change - two losses and a different small forward gives you all the collateral you need for a change at power forward (that you can stick with long term when Carroll comes back).

What do you guys think? Do we need a change? If you disagree, I'd love to see the reasoning, and am happy to continue the conversation in the comments.

(I can also provide more detailed numbers on what was summarized towards the end of the article if anyone is interested.)

All statistics from NBA.com.