538: Will Cavs' Defense Persist?

[Guy]

Interesting piece by Neil, suggesting Cavs can expect only about a quarter of their Games 1-3 defensive success to persist going forward: http://fivethirtyeight.com/datalab/the- ... eep-it-up/.

However, I worry that looking only at 7-game series may have biased the results in a way that understates the carryover. A dominant defensive team in games 1-3 will only play a 7-game series if their defense fades, and vice-versa for a defensive underperformer. So it would be important to look also at 5- and 6-game series, and perhaps focus on games 4 and 5 (across all series) to see how well the defense performed in those next two games. Perhaps if Neil swings by here, he can let us know if that changes the conclusion at all.

[Mike G]

Neil Paine wrote:

... since 1985, only the 1998 Utah Jazz in their matchup with the Chicago Bulls have done a better job holding an opponent below offensive expectations (2) through the opening three games of an NBA Finals.

And the footnote reads:

In this case, the efficiency we’d expect based on a team’s regular-season ratings and whether the games were played at home or on the road.

The Cavs' season DRtg was 106.3, which was 0.7 weaker than NBA avg.
The Warriors' RS ORtg was 111.6, which was 6.0 better than the norm. So they should be humming along at 112.3 . Then if HCA is worth 3.2, half on O and half on D, add about (1.6/3) 0.5 more -- 112.8

GSW ORtg in 3 games is 101.5, which is short of expected by 11.3

The '98 Bulls had ORtg of 107.7, which was 2.7 > avg. The Jazz DRtg was 105.4, worse than avg by 0.4
When they meet, Chi should have ORtg = 108.1
In games 1-3, Bulls had ORtg of 92.2, 115.9, and 104.5 -- avg 104.2 -- just 3.9 short of expected. ???

EDIT: I believe Neil stated it backwards, re the Bulls/Jazz of '98. It was Utah, with the league-best ORtg of 112.7, bottoming out in game 3 with 58.8.
Chi DRtg was about 5 better than avg, not 55.

[bchaikin]

the cavs defense these playoffs has been much better with mozgov on the floor than without him...

so not sure why he's playing just 31 min/g in the finals. it's certainly not due to foul trouble...

[xkonk]

Maybe related to Mike's calculation difficulties, but a question I have anyway... why is Neil doing so much weighting by leverage in his 538 articles? I understand in some circumstances it might make sense, like if you want to say something about clutch/playoff performances. But why weight games by leverage for this? Are later rounds and elimination games more predictive of defensive ability?

[DSMok1]

Maybe related to Mike's calculation difficulties, but a question I have anyway... why is Neil doing so much weighting by leverage in his 538 articles? I understand in some circumstances it might make sense, like if you want to say something about clutch/playoff performances. But why weight games by leverage for this? Are later rounds and elimination games more predictive of defensive ability?

I totally agree with this criticism. Only weight by leverage for very specific objectives, that are narrative based. Like Championships Added or something. Otherwise, you're just cutting your sample size down and biasing it.

[Mike G]

Neil fixed the Jazz/Bulls thing.

[Mike G]

the cavs defense these playoffs has been much better with mozgov on the floor than without him...

so not sure why he's playing just 31 min/g in the finals. it's certainly not due to foul trouble...

I'd guess they're also much better on offense when he's in. He's the only Cle player remotely shooting with volume and efficiency:

Code: Select all

Cavs            TSA/36   TS%
LeBron James     21.0   .474
Tristan Thompson  4.0   .456
Iman Shumpert     4.1   .353
J.R. Smith        6.2   .376
Timofey Mozgov   14.0   .619
Matt Dellavedova  7.9   .434
James Jones       3.6   .558
(Kyrie Irving     8.9   .512)
Mike Miller       1.1   .000

LeBron is actually hitting a better % than anyone else with even a sixth as many TSA/36.
With Mozgov off the floor, LeBron is taking more than half the Cavs' shots.

Mozgov's on/off numbers in 4 games:

Code: Select all

gm     on   off
1      +3   -11
2     +11    -9
3      +1    +4
4      -5   -16

In game 4, the Cavs refused to follow GS' lead in going small. While Mozgov flourished, the team didn't; and they might have done worse.

Here's the progression of Warriors' lineup shrinkage, by minutes weighted height, weight, and nominal position:

Code: Select all

gm  inches  pounds  pos.
1    79.1    212    2.59
2    78.9    211    2.50
3    78.8    212    2.51
4    78.4    206    2.18

A traditional lineup (C, PF, SF, SG, PG) would have a position avg of 3.0
NBA avg size this year was about 79" (6-7) and 218 lb.
Dellavedova (200) is the only Cavs player listed at under 206; then Shumpert at 212.