NBA Games Discussion Thread - Part 6

[TheBlackCzar]

Co-worker thinks someone like Zion makes a lot of sense for them. We were saying Wemby and 4 guards is probably not a recipe for success. He needs a bruising 4 that can also be a threat from 3.

Honestly the best game I've ever watched.

Zion has stopped shooting 3s but maybe the Spurs coaching staff could fix that.

This obsession with shooting 3's is why SAS blew a 30 point lead.....
Like how many times do people have to see it before they believe it....

Shooting ALL 3's and no going to the hole is just moronic basketball, that the entire league has adopted......
Like nobody on the damn team thought let me get fouled.....
Jumper, jumper, jumper.....

That was some stupid situational basketball on display last night by SAS......

[Rafael122]

Co-worker thinks someone like Zion makes a lot of sense for them. We were saying Wemby and 4 guards is probably not a recipe for success. He needs a bruising 4 that can also be a threat from 3.

Honestly the best game I've ever watched.

Zion has stopped shooting 3s but maybe the Spurs coaching staff could fix that.

This obsession with shooting 3's is why SAS blew a 30 point lead.....
Like how many times do people have to see it before they believe it....

Shooting ALL 3's and no going to the hole is just moronic basketball, that the entire league has adopted......
Like nobody on the damn team thought let me get fouled.....
Jumper, jumper, jumper.....

That was some stupid situational basketball on display last night by SAS......

I said a threat from 3 though. Zion's got the bully ball that the Spurs are sorely lacking.

[Paris41]

Sigh.

C'mon NatP4. You are better than this. The available sample sizes for 3-man combinations are obviously much smaller than the sample sizes for 2-man combinations. That's how statistics work.

If you want to believe larger 2-man sample sizes that ignore the warping presence of Wemby more than smaller sample sizes that factor the existence of Wemby, that's your prerogative. I think independent-minded observers here are going to see the data I provided and acknowledge that the Harper/Castle combination is exceptionally good and Castle is not some kind of chemistry problem here.

And your 2-man data shows nothing definitive. The difference between +1.6, +2.4 and +3.6 is meaningless given the sample sizes. At least with my 3-man data, even assuming broad margins for error, the massive difference between a +38.3 and a +18.8 is noteworthy.

I still don’t see why 2 man playoff data of a 300-400 minute sample is meaningless, but 3 man regular season data of less than 200 overall minutes is definitive. Explain this:

Castle has played 728 minutes in the playoffs and has posted a -8.4 on/off differential, worst on the team of players who have played substantial minutes.

Harper has posted a +9.6 in 584 minutes

Fox has posted a +0.3 in 667 minutes

Again, I’m not arguing with you regarding Castle’s fit next to Wemby. I’m saying Fox and Harper are both significantly better players and that Castle has been a boat anchor in the playoffs and completely misused.

[Paris41]

For reference, here’s the playoffs numbers:

Fox/Castle/Wemby: 314 minutes +3.4
Castle/Harper/Wemby: 202 minutes +2.5
Fox/Harper/Wemby: 160 minutes +2.5

Looks to me like Wemby makes every lineup good.

There’s not much to conclude from this regular season data set, only the Castle/Fox/Wemby combination played substantial minutes:

Castle/Fox/Wemby: +12.2 NET (621 minutes)
Castle/Harper/Wemby: +38.3 NET (173 minutes)
Fox/Harper/Wemby: +18.8 NET (193 minutes)

[Kanyewest]

Zion has stopped shooting 3s but maybe the Spurs coaching staff could fix that.

This obsession with shooting 3's is why SAS blew a 30 point lead.....
Like how many times do people have to see it before they believe it....

Shooting ALL 3's and no going to the hole is just moronic basketball, that the entire league has adopted......
Like nobody on the damn team thought let me get fouled.....
Jumper, jumper, jumper.....

That was some stupid situational basketball on display last night by SAS......

I said a threat from 3 though. Zion's got the bully ball that the Spurs are sorely lacking.

Yeah Zion could stick to his post game if he joined the Spurs, just a couple of 3s would do. Just like all 3s, all 2s isn't a good strategy either.

[closg00]

Co-worker thinks someone like Zion makes a lot of sense for them. We were saying Wemby and 4 guards is probably not a recipe for success. He needs a bruising 4 that can also be a threat from 3.

Honestly the best game I've ever watched.

Zion has stopped shooting 3s but maybe the Spurs coaching staff could fix that.

This obsession with shooting 3's is why SAS blew a 30 point lead.....
Like how many times do people have to see it before they believe it....

Shooting ALL 3's and no going to the hole is just moronic basketball, that the entire league has adopted......
Like nobody on the damn team thought let me get fouled.....
Jumper, jumper, jumper.....

That was some stupid situational basketball on display last night by SAS......

Tre and more-so Bub are this type of player, we don’t need them both:

[Kanyewest]

I am trying to coming to terms with Fox's line of reasoning for the play even though it was still ultimately a bad play that seemed foolish at the time. It actually looks like he had an open lane to the hoop and it looks like he may have gotten fouled, certainly looks like a foul when comparing it to the foul that KAT got called earlier in the game, which could have been let go but maybe in the back of his mind he thought he was going to get the call with an opportunity for an and 1. That being said, he has to realize that not only is the foul unlikely to get called because it is a close game but because it is a playoff game.

Wemby only shot 3 of 14 from the field. 2 of his baskets were jumpers, and 1 came off an offensive rebound. Wemby did a poor job attacking the paint to close out the game and maybe he was afraid of getting fouled after missing 2 free throws.

[nate33]

I still don’t see why 2 man playoff data of a 300-400 minute sample is meaningless, but 3 man regular season data of less than 200 overall minutes is definitive.

There is a big margin for error with a sample size so small. In your 2-man data the difference between the worst combo with an +/- of +1.6 and the best combo with a +/- of +3.2 is extremely small, almost surely smaller than the margin of error, giving the data no meaning.

In my 3-man data, the margin for error is even bigger, but the difference between the Fox 3-man lineup and the Castle 3-man lineup is absolutely massive, a 19.5 point difference. That is likely larger than the margin for error so therefore it does have meaning.

Explain this:

Castle has played 728 minutes in the playoffs and has posted a -8.4 on/off differential, worst on the team of players who have played substantial minutes.

Harper has posted a +9.6 in 584 minutes

Fox has posted a +0.3 in 667 minutes

Again, I’m not arguing with you regarding Castle’s fit next to Wemby. I’m saying Fox and Harper are both significantly better players and that Castle has been a boat anchor in the playoffs and completely misused.

I think it's random perturbations due to a small sample size. In the regular season, with a much larger sample size, Harper, Fox and Castle had an on/off differential of +2.7, +3.2 and +2.8 respectively. That's with sample sizes over 1500 minutes. Why would you ignore that in favor of playoff data with much smaller sample sizes?

Another possibility is that Castle is deleterious to the team unless he is alongside Wemby, which makes some sense because the no-Wemby lineup has trouble scoring and Castle is the least offensively capable scorer among the guards. But when good scorers like Harper, Vassell and Wemby are on the floor, Castle's elite defense helps while his mediocre offense is mitigated because he becomes a 4th option.

Mostly, I think it's simply that you tend to put to much stock in on/off data over small sample sizes. From what I understand from the eggheads, on/off data isn't even all that accurate with an entire season of data. You need 2 or 3 years to really be able to tease out the effects of teammates and opponent differences.

[FAH1223]

I still can't believe Mitch Johnson didn't let Wemby sit for a few minutes either.

[FAH1223]

Read on Twitter

[montestewart]

There must be some good "I'm in your head" memes floating around today

[doclinkin]

I will say this. As a reformed Nix fan from way back. If they win the series, their fans gonna be utterly insufferable next year. Like everybody else’s arena sold out with imported douchery.

[montestewart]

I will say this. As a reformed Nix fan from way back. If they win the series, their fans gonna be utterly insufferable next year. Like everybody else’s arena sold out with imported douchery.

As a counterpoint, and I know they are the minority, but there are a handful of New Yorkers (granted, mostly transplants like my brother-in-law) that are a little more humble about it. Knicks were the first team I rooted for because my grandfather was a fan, and it initially paid off with titles in 1970 and 1973. When the Bullets moved to DC, it was not a hard transition, since they were winning more than the Knicks almost immediately, but I've always followed the Knicks and they sure have had disappointments and bad ownership through the years.

Years ago I had court side seats (only time in my life) to a Wizards-Knicks game at Capital Centre. There were some obnoxious Knicks fans court side as well. At one point Ewing and Larry Johnson were walking by, and they looked almost embarrassed by the fanboy drivel tossed their way.

[Paris41]

There is a big margin for error with a sample size so small. In your 2-man data the difference between the worst combo with an +/- of +1.6 and the best combo with a +/- of +3.2 is extremely small, almost surely smaller than the margin of error, giving the data no meaning.

In my 3-man data, the margin for error is even bigger, but the difference between the Fox 3-man lineup and the Castle 3-man lineup is absolutely massive, a 19.5 point difference. That is likely larger than the margin for error so therefore it does have meaning.

Of course not. You’re comparing a 3-man combo that played 600+ minutes to another one that played 173 minutes.

Your entire argument rests on data from a 3-man lineup that played 173 minutes together in one regular season.

Earlier, you argued that 300-400 minute playoff 2-man samples were too small to draw conclusions from. You also described 500-700 minute playoff on/off samples as “random perturbations” and argued that even a full season of on/off data is heavily affected by teammate and opponent context.

If that’s true, why does that standard suddenly disappear for a 173-minute 3-man lineup?

A large observed difference does not make sample size irrelevant. If you want to argue that the 19.5-point gap is meaningful, you need to demonstrate that it is statistically distinguishable from the noise inherent in a 173-minute lineup sample.

What’s even stranger is that the playoff 3-man data doesn’t show anything close to the massive separation you’re inferring from the regular season lineup data:

Fox/Castle/Wemby: +3.4 in 314 minutes
Castle/Harper/Wemby: +2.5 in 202 minutes
Fox/Harper/Wemby: +2.5 in 160 minutes

Those samples are comparable to or larger than the 173-minute regular season lineup you’re relying on, yet the results are remarkably similar. If the 173-minute regular season sample is meaningful because of its large effect size, why doesn’t the playoff 3-man data receive the same consideration?

[montestewart]

Nate dominated in the first half but Paris coming on strong in the second half.

[Paris41]

I actually agree with part of your argument.

It’s probably better not to draw strong conclusions from a single regular season or playoff run of raw on/off data. Raw on/off is heavily influenced by teammate combinations, opponent quality, rotations, and deployment. Those are all legitimate limitations of the metric.

If we’re going to take that position seriously, though, then I don’t think the answer is to rely on a 173-minute 3-man lineup sample. The answer is to move toward metrics that are specifically designed to account for teammate and opponent context like RAPM.

If we’re throwing out single-season on/off and lineup data because of context concerns, then RAPM seems like a much more appropriate place to look.

Harper: +5.0 RAPM (8th in the NBA)

Fox: +2.0 RAPM (72nd in the NBA)

Castle: -0.8 RAPM (315th in the NBA)

At a minimum, RAPM provides much stronger evidence for Harper and Fox than a single 173-minute lineup sample does.

[nate33]

I actually agree with part of your argument.

It’s probably better not to draw strong conclusions from a single regular season or playoff run of raw on/off data. Raw on/off is heavily influenced by teammate combinations, opponent quality, rotations, and deployment. Those are all legitimate limitations of the metric.

If we’re going to take that position seriously, though, then I don’t think the answer is to rely on a 173-minute 3-man lineup sample. The answer is to move toward metrics that are specifically designed to account for teammate and opponent context like RAPM.

If we’re throwing out single-season on/off and lineup data because of context concerns, then RAPM seems like a much more appropriate place to look.

Harper: +5.0 RAPM (8th in the NBA)

Fox: +2.0 RAPM (72nd in the NBA)

Castle: -0.8 RAPM (315th in the NBA)

At a minimum, RAPM provides much stronger evidence for Harper and Fox than a single 173-minute lineup sample does.

I'll agree with this. RAPM is going to be better than the raw data. But the RAPM also develops more validity over time.

Harper ranking 9th is mind-boggling

And you make a good point that Fox is outperforming Castle by this metric. It may be fair to say that, overall across all lineup situations, Fox has been more helpful than Castle. Nevertheless, the 3-man lineups involving Castle and Fox when alongside Wemby and Harper show little distinction between Fox and Castle in the playoffs, and a significant advantage for Castle in the regular season. So I think it's fair to say that Castle has the edge as the complimentary player paired with the two stars. That's another argument to keep Castle as a starter alongside Harper. And have Fox be the 6th man because he apparently helps more when not alongside Wemby.

We shouldn't overlook that Castle is 21 years old and makes 20% of what Fox makes. Going forward, I maintain that Castle is going to be guy who is kept, and Fox will be the one moved.

[tontoz]

Harper and Wemby are the only guys they really have to keep.

[Paris41]

The bigger point here is that Dylan Harper should be a Wizard.

[TheBlackCzar]

Ya'll can argue analytics, bottom line, Harper is better than Fox or Castle....
He needs to be used as #2.....

Castle and Fox are neither PG's, and are shaky under pressure......
Castle needs to give the ball to Harper...
Fox needs to give the ball to Harper.....
They would be much better off by making this one change......