og15 wrote:esqtvd wrote:og15 wrote:Spoiler:
Interesting points about the Lou/Shamet/Harrell combination and defense. Making up for three players on defense is too much though. Ideally you only want to be making up for one guy, and at most two. If there are more players on the floor to make up for than there are guys anchoring the defense, then it's very unlikely those lineups will consistently perform well on defense.
Or if you do have to make up for more than two guys on defense, one of those guys probably shouldn't be your big.
All that said, I don't ever feel it is fair to evaluate with perception and no data. I've been eyeing Harrell's on/off Drtg all season to see how the team performs on defense with him on the court (while of course understanding that he's not the only factor).
As of this moment:
Harrell on: 108.2 Drtg
Harrell off: 107.2 Drtg
Defense is a little bit better when he's off, but not significant, and who else he's playing with factors in.
What was evident before and is no longer evident is the offensive boost with him vs without.
At this moment:
Harrell on: 113.5 Ortg
Harrell off: 113.0 Ortg
This means that as of now, Harrell net on/off is -0.5, this does not mean he is bad or negative or anything like that. What it does mean is that the team so far is performing just as well (0.5 is such a small difference) with him on the court than without.
Zubac is obviously the big question when it comes to who should be playing instead.
Zubac on: 107.1 Drtg
Zubac off: 108.2 Drtg
Zubac on: 113.4 Ortg
Zubac off: 113.2 Drtg
His NetRtg is, +1.3. Small differences, there's some noise there, but at the least we can theorize that it's pretty much a wash as to which on is on the court at this moment. This gives great support to those who are calling for a greater use of Zubac, especially in moments where Harrell is playing long extended minutes and maybe starts to look winded or unable to give second and third efforts.
There's nothing in his play or the numbers suggesting that the team would start bleeding points if Zubac played more and more consistently, especially to give Harrell a breather. Now of course on/off data is just one piece, it's not taking into account teammates. Zubac plays more (most) of his minutes with better defenders than Harrell does, but then Harrell plays more minutes against bench lineups than Zubac does and those lineups are more often inferior on offense though maybe not on defense to their starter counterparts.
Good analysis, rather than opinions and rancid emotions, LOL. But put Zubac out there with the second team instead of the first and you'll see everything including his numbers go to hell.
His numbers are good with the first team because most of those minutes are with Kawhi, who makes everybody look good. And as you saw in a typical case like in Boston when he's having a bad game, we started getting killed at the beginning of the 3rd quarter and Doc gave him the hook. This not only preserved Zu's stats [although he was still a team-worst minus-13 for the game] but also it stopped the bleeding on the court.
Trezz was the only reason we were still in that game, that and Lou's 35 points. And yet this game thread turns into bashing them. Unreal.
That’s of course the second question, and maybe a harder one to figure out, or maybe not. One way we can attempt to hypothesize is look at two and three man lineups such as Kawhi/Zubac vs Kawhi/Harrell and Lou/Zubac vs Lou/Harrell and Kawhi/George/Zubac vs Kawhi/George/Harrell and try and get a feel for what’s happening there. Of course sample sizes might not be sufficient to make conclusions. Here’s some of that.
With Leonard:
Zubac: 698 mins, 112.5 Ortg, 105.3 Drtg, +7.2 NetRtg
Harrell: 635 mins, 116.8 Ortg, 100.4 Drtg, +16.5 NetRtg
With Williams:
Zubac: 299 mins, 103.1 Ortg, 107.2 Drtg, -4.2 NetRtg
Harrell: 1222 mins, 110.9 Ortg, 106.5 Drtg, +4.4 NetRtg
With Leonard and George:
Zubac: 355 mins, 114.7 Ortg, 104.9 Drtg, +9.8 NetRtg
Harrell: 199 mins, 109.7 Ortg, 97.4 Drtg, +12.3 NetRtg
So the first thing to notice is that when they are on the floor with Kawhi, a combination that has good and similar sample sizes for both, the units with Harrell is really moping the floor with the units that have Zubac. This is certainly a good argument for the coaching staffs use of Harrell. That unit is also 5 points/100 better on defense.
Second is that when you isolate each with Williams, we see that Harrell is actually still winning on both ends, but also that Zubac does not provide enough offensive help and there’s a drastic drop off offensively. Overall there’s a 8.2 point/100 difference in team production, and again, defensively, you’re still getting a better result with Harrell. Large discrepancy in sample size here though.
Now we look at with both George and Leonard, and we see something different offensively, we get a drop there, and maybe surprising to some, but defensively again, the lineups with Harrell win by a large margin, this time 7.5 pts/100 better on defense, and still having a better NetRtg despite being inferior on offense.
That’s a lot of data pointing in one direction and should certainly be analyzed and taken into account.
The Bench Lineups
In terms of the bench struggling defensively, SlayrofDynasties mentioning of the 3 man combination of Williams/Shamet/Harrell is telling, and obviously it makes sense. Two smaller guards like that can be trouble on defense depending on who the players are. With Williams, Shamet will get forced to guard less favourable matchups for him, reducing his defensive effectiveness, and Williams is just not a good defender either way.
Williams/Harrell/Shamet: 460 mins, 110.0 Ortg, 110.9 Drtg, -1.0 NetRtg
BUT, If we add Leonard to the Williams/Shamet combo:
Williams/Leonard/Shamet: 249 mins, 112.9 Ortg, 107.5 Drtg, +5.3 NetRtg
So Kawhi is able to balance out that combination when he plays with them.
Okay, what about having Leonard on with Williams and Harrell?
Leonard/Williams/Harrell: 464 mins, 114.8 Ortg, 98.8 Drtg, +16.0 NetRtg
The lineups with Leonard/Williams/Harrell is actually one of the strongest 3 man combinations on the team. So Kawhi has been a balancing force for the Williams/Harrell combo in a whole season sense, they are elite both on offense and defense for their 464 minutes. If we look at 3 man lineups with 150+ mins, out of the top 15 in Drtg, Harrell is part of 7 of them. In the top 20, he’s part of 10 of the 20. Zubac on the other hand is part of just 1 of the top 15, and that is 12th, and just two in the top 20. Harrell is part of the best 3 man unit with 150+ minutes, Beverley/Leonard/Harrell (246 mins, 122.5 Ortg, 91.1 Drtg, +31.4 NetRtg).
Since January
Piggy backing off the other thread about Williams and Harrell in the New Year, since January:
With Leonard:
Zubac: 293 mins, 115.9 Ortg, 110.5 Drtg, +5.5 NetRtg
Harrell: 225 mins, 120.5 Ortg, 108.8 Drtg, +11.8 NetRtg
With Williams:
Zubac: 121 mins, 103.0 Ortg, 109.3 Drtg, -6.3 NetRtg
Harrell: 456 mins, 110.2 Ortg, 111.7 Drtg, -1.5 NetRtg
With George and Leonard:
Zubac: 83 mins, 114.0 Ortg, 110.7 Drtg, +3.3 NetRtg
Harrell: 32 mins, 113.6 Ortg, 102.9 Drtg, +10.7 NetRtg
Williams/Leonard/Harrell: 171 mins, 118.4 Ortg, 107.6 Drtg, +10.8 NetRtg
I’ll have to say that I did not necessarily expect the numbers to look like this, but if we’re using lineup data and normalizing to an extent for who is on the court with the bigs (Zubac vs Harrell), lineups with Harrell are winning by a landslide when it comes to defense and overall net per possession production. This is both for the season as a whole and since the start of the new year.
So even in that sample size since January, you’re still getting a better result both offensive and defensively when you pair Kawhi and Harrell vs pairing Kawhi and Zubac. Playing either with Williams suggest that you are still getting a better overall result with Harrell, and then combined with the w/Leonard and w/George+Leonard data suggests that Harrell on his own is not giving an inferior defensive result to Zubac, but it is the pieces around them making the difference. The data is actually leaning towards Harrell giving a superior defensive contribution altogether, though it is not comprehensive and complete data to make a definitive statement like that. We also see that the Williams/Leonard/Harrell lineup even in January has been performing very well.
Great work, G, and it seems to have hushed up Planet Zubac. For how long, I don't know, LOL.
Facts are stubborn things. My first assumption is always that Doc Rivers knows more about basketball than I do, so I look for stats that explain his reasoning. The stats uncannily always seem to back Doc up when you dig below the surface.
He looks at the stats too, you know. First and foremost, he wants to win. The personal attacks are silly.
I was confident that Trezz would come through clean but even I am surprised that Lou--who is provably a minus defender and lately has been less than his best offensively--would still come out so positively. But you just proved that any apparent suckage can be attributed the crud they've been obliged to be out on the floor with--McGruder, Green, Robinson, and assorted G Leaguers. We should be toasting Trezz and Lou, not trashing them. We'd be dead meat without them.
I see OG spending an hour rounding up the facts and you sitting on your couch d-bagging Doc
