Where should Al Horford play: 4 or 5?

[PhoSita]

I'm not impressed by Coach Nick's 'analysis' here.  He keys off the Net Rating difference between the two front-court pairings, but ignores why the sample sizes are so dramatically different.    It is a naive mistake to think that if you simply swapped Olynyk for Amir that the starters would suddenly be posting a +22 Net Rating.

I do think that, as Olynyk's game continues to improve as he gets further from his shoulder surgery, that we may get to a point where it makes sense to start him over Amir, but I don't think we are all that close to that point yet.

Having some fun with nba wowy, look at the Team DRTG:



w/ Horford and w/out IT ---> 103

w/ IT and w/o Horford ---> 112

w/ Horford and w/o Amir ---> 111

w/ Amir and w/o Horford ---> 108

w/ Amir and neither Horford nor Thomas ---> 103

w/ Horford and neither Amir nor Thomas ---> 100

w/ IT and neither Horford nor Amir --> 116

w/ Horford & Amir and w/o IT --> 109


Big man pairings:

Horford and Olynyk ---> 111

Horford and Amir ---> 108

Horford and Jerebko -- 105

Amir and Olynyk ---> 100

Amir and Jerebko ---> 114

Olynyk and Jerebko ---> 107


Backcourt:

IT & Bradley ---> 114

Bradley & Smart ---> 107

Bradley, Crowder & Smart w/out IT or Rozier --> 101



A few observations:


- The defense appears to generally get worse if Isaiah is on the floor, and better when he's off. Not a huge surprise.  Team defense overall is 113 w/ IT and 104 w/out.

- It's not clear that Horford is bad for the defense, but it does appear that Horford isn't good enough defensively to make up for IT.  Horford appears to be at his best when paired with a more mobile big man.

- Lineups with more than two guards are awful defensively.  AB/Crowder/Smart are predictably quite good defensively when they're not part of an ultra-small lineup.

- The numbers suggest Horford & Jerebko are a solid pairing, while Amir & Olynyk are an excellent pairing, at least in terms of defense.

[jpotter33]

I'm not impressed by Coach Nick's 'analysis' here.  He keys off the Net Rating difference between the two front-court pairings, but ignores why the sample sizes are so dramatically different.    It is a naive mistake to think that if you simply swapped Olynyk for Amir that the starters would suddenly be posting a +22 Net Rating.

I do think that, as Olynyk's game continues to improve as he gets further from his shoulder surgery, that we may get to a point where it makes sense to start him over Amir, but I don't think we are all that close to that point yet.

Having some fun with nba wowy, look at the Team DRTG:



w/ Horford and w/out IT ---> 103

w/ IT and w/o Horford ---> 112

w/ Horford and w/o Amir ---> 111

w/ Amir and w/o Horford ---> 108

w/ Amir and neither Horford nor Thomas ---> 103

w/ Horford and neither Amir nor Thomas ---> 100

w/ IT and neither Horford nor Amir --> 116

w/ Horford & Amir and w/o IT --> 109


Big man pairings:

Horford and Olynyk ---> 111

Horford and Amir ---> 108

Horford and Jerebko -- 105

Amir and Olynyk ---> 100

Amir and Jerebko ---> 114

Olynyk and Jerebko ---> 107


Backcourt:

IT & Bradley ---> 114

Bradley & Smart ---> 107

Bradley, Crowder & Smart w/out IT or Rozier --> 101



A few observations:


- The defense appears to generally get worse if Isaiah is on the floor, and better when he's off. Not a huge surprise.  Team defense overall is 113 w/ IT and 104 w/out.

- It's not clear that Horford is bad for the defense, but it does appear that Horford isn't good enough defensively to make up for IT.  Horford appears to be at his best when paired with a more mobile big man.

- Lineups with more than two guards are awful defensively.  AB/Crowder/Smart are predictably quite good defensively when they're not part of an ultra-small lineup.

- The numbers suggest Horford & Jerebko are a solid pairing, while Amir & Olynyk are an excellent pairing, at least in terms of defense.

TP and these statistics seem to line up with the eye test and common sense pretty nicely.

I still think we should be starting JJ and playing him like we play Amir now - as a starting and finishing unit but with limited overall minutes from 15 to 20 minutes a game, of course with potential offense/defense subs and differing matchups if necessary. Amir and KO has always been an ideal matchup, too, so they seem to make much more sense together off the bench.

I just think JJ's extra mobility/length/athleticism and spacing is better for our starting unit as the fifth option offensively and the 4 defensively, along with the added ability of putting Horford closer to the basket both offensively and defensively.

[The Oracle]

I'm not impressed by Coach Nick's 'analysis' here.  He keys off the Net Rating difference between the two front-court pairings, but ignores why the sample sizes are so dramatically different.    It is a naive mistake to think that if you simply swapped Olynyk for Amir that the starters would suddenly be posting a +22 Net Rating.

I do think that, as Olynyk's game continues to improve as he gets further from his shoulder surgery, that we may get to a point where it makes sense to start him over Amir, but I don't think we are all that close to that point yet.

Having some fun with nba wowy, look at the Team DRTG:



w/ Horford and w/out IT ---> 103

w/ IT and w/o Horford ---> 112

w/ Horford and w/o Amir ---> 111

w/ Amir and w/o Horford ---> 108

w/ Amir and neither Horford nor Thomas ---> 103

w/ Horford and neither Amir nor Thomas ---> 100

w/ IT and neither Horford nor Amir --> 116

w/ Horford & Amir and w/o IT --> 109


Big man pairings:

Horford and Olynyk ---> 111

Horford and Amir ---> 108

Horford and Jerebko -- 105

Amir and Olynyk ---> 100

Amir and Jerebko ---> 114

Olynyk and Jerebko ---> 107


Backcourt:

IT & Bradley ---> 114

Bradley & Smart ---> 107

Bradley, Crowder & Smart w/out IT or Rozier --> 101



A few observations:


- The defense appears to generally get worse if Isaiah is on the floor, and better when he's off. Not a huge surprise.  Team defense overall is 113 w/ IT and 104 w/out.

- It's not clear that Horford is bad for the defense, but it does appear that Horford isn't good enough defensively to make up for IT.  Horford appears to be at his best when paired with a more mobile big man.

- Lineups with more than two guards are awful defensively.  AB/Crowder/Smart are predictably quite good defensively when they're not part of an ultra-small lineup.

- The numbers suggest Horford & Jerebko are a solid pairing, while Amir & Olynyk are an excellent pairing, at least in terms of defense.

Coach Nick's analysis is very weak.  You simply cannot use super small sample sizes like that and make any definitive conclusions as mmmmm said.  His conclusion I do believe is correct in spite of the weak evidence as there is plenty of evidence dating back multiple years as to K.O.'s effectiveness defending the perimeter.

As far as those numbers from NBAwowy I don't believe they are very accurate as NBA.com has some of those pairings showing vastly different numbers.  Olynyk/Jerebko for instance is 99.8, Johnson/Olynyk 98.2, Jerebko/Johnson 110.8, Horford/Olynyk 107.6, Horford/Johnson 105.9.

Olynyk/Jerebko have always put up fantastic defensive numbers.  92.8 last year in 522 minutes and 84.7 in 223 minutes post all star beak in 14-15.  Small samples can yield some wildly varying numbers.

[Bobshot]

The poll counter doesn't seem to be working, so I set the poll count to zero--FWIW.

[mmmmm]

I'm not impressed by Coach Nick's 'analysis' here.  He keys off the Net Rating difference between the two front-court pairings, but ignores why the sample sizes are so dramatically different.    It is a naive mistake to think that if you simply swapped Olynyk for Amir that the starters would suddenly be posting a +22 Net Rating.

I do think that, as Olynyk's game continues to improve as he gets further from his shoulder surgery, that we may get to a point where it makes sense to start him over Amir, but I don't think we are all that close to that point yet.

Having some fun with nba wowy, look at the Team DRTG:



w/ Horford and w/out IT ---> 103

w/ IT and w/o Horford ---> 112

w/ Horford and w/o Amir ---> 111

w/ Amir and w/o Horford ---> 108

w/ Amir and neither Horford nor Thomas ---> 103

w/ Horford and neither Amir nor Thomas ---> 100

w/ IT and neither Horford nor Amir --> 116

w/ Horford & Amir and w/o IT --> 109


Big man pairings:

Horford and Olynyk ---> 111

Horford and Amir ---> 108

Horford and Jerebko -- 105

Amir and Olynyk ---> 100

Amir and Jerebko ---> 114

Olynyk and Jerebko ---> 107


Backcourt:

IT & Bradley ---> 114

Bradley & Smart ---> 107

Bradley, Crowder & Smart w/out IT or Rozier --> 101



A few observations:


- The defense appears to generally get worse if Isaiah is on the floor, and better when he's off. Not a huge surprise.  Team defense overall is 113 w/ IT and 104 w/out.

- It's not clear that Horford is bad for the defense, but it does appear that Horford isn't good enough defensively to make up for IT.  Horford appears to be at his best when paired with a more mobile big man.

- Lineups with more than two guards are awful defensively.  AB/Crowder/Smart are predictably quite good defensively when they're not part of an ultra-small lineup.

- The numbers suggest Horford & Jerebko are a solid pairing, while Amir & Olynyk are an excellent pairing, at least in terms of defense.

AAAAAGGGGHHHHHHH!!!!!

Please, Phosita.  You are killing me.

Those numbers mean absolute jack without sample sizes!

And the fact is, almost ALL those sample sizes are going to be too tiny at this point to be useful at all when we have played just 37 games and key guys in the pairings like Al Horford have played in only 27.

I think the stats used here need to be very clearly indicated as being very tiny sample sizes.

For example, keep in mind that both Avery and Isaiah both are carrying a huge -59 & -57, respectively, on their plus/minus data due entirely to just the three games (DEN, WAS & GSW) from way back at the beginning of the season while Al & Jae were out and they were starting with Tyler and Jaylen in the first two and KO & MS in the other.  Those 70 minutes of shared floor time has a huge effect.  Overall, the pairing of AB & IT has a net -1 rating, but in the 704 minutes outside those three games they have a +3 net rating.

So that shows how just a small (less than 10%) portion of that sample skews that pairing significantly.   And AB+IT is the biggest 2-man sample that we have so far.