25 replies to this topic

[Red94]

New post: Lineup data to this point
By: Rahat Huq

After seeing Houston pull away last night to start the fourth, my hypothesis was that the Rockets' offense is at its best with Corey Brewer in the lineup.  I set NBA.com's parameters to a minimum of 10 games played together, and found that the top Rockets quintet is Ariza/Harden/Motiejunas/Smith/Terry, with an offensive rating of 143 in 23 minutes together.  For lineups that have played at least 35 minutes together, the top quintet is Brewer/Harden/Motiejunas/Smith/Terry with an offensive rating of 127 in 36 minutes.  The common theme there is that the team seems to be at its best offensively with Motiejunas/Smith/Terry/Harden and a wing.

 

Strangely, defensively, for lineups that had played at least 35 minutes together, the stingiest unit was Ariza/Brewer/Jones/Smith/Terry with a defensive rating of 72.9 in 40 minutes of shared court time.  For lineups that have played 10 games together, the stingiest unit is Ariza/Brewer/Motiejunas/Smith/Terry with a defensive rating of 74.8 in 37 minutes player together.

 

Interesting to note above the glaring omissions of Patrick Beverley and Dwight Howard.  And in Howard's case, its not that he hasn't played enough to qualify for the parameters - he can be found in some of the later units.

 

Obviously, small sample size alert.  And a lot is lost in the numbers.  But some of the combinations seem to meet the eye test.  Once a sufficiently sizable sample accumulates, I'm very curious to see how lineups with Terrence Jones fare in general to the other combinations.

[Sir Thursday]

The small sample size essentially renders a lot of this irrelevant, I'm afraid. Groups who haven't played together much will naturally produce more extreme numbers both good and bad, so you're drastically biasing towards lineups containing players who are new to the team or haven't played many games. Lineups containing Jones, Smith and Brewer were always going to appear in the top section - 40 minutes is nowhere near long enough to smooth things out - that's not even a full game! I don't know how much data you would actually need to get things to settle, but I think you'd have to put a limit of at least 100 minutes for it to be reasonable.

 

By this token, the reason why Howard and Beverley are missing is because they will only have played in lineups that have accumulated a lot of minutes together. These lineups will naturally be in the middle as the good and bad performances even out. Which is not to say that there aren't interesting things to be found in the lineup data, it's just that there's not enough of it yet. There's a reason why RPM and similar stats need about three seasons worth of data to be useful - when parsing things out of on/off court data it's really difficult to account for the noise.

 

ST

[King's Gambit Accepted]

my hypothesis is that the Rockets' offense is at its best with Corey Brewer in the lineup when hes driving a lot to the rim and only occasionaly firing up a 3pt

[Alituro]

The small sample size essentially renders a lot of this irrelevant, I'm afraid. Groups who haven't played together much will naturally produce more extreme numbers both good and bad, so you're drastically biasing towards lineups containing players who are new to the team or haven't played many games. Lineups containing Jones, Smith and Brewer were always going to appear in the top section - 40 minutes is nowhere near long enough to smooth things out - that's not even a full game! I don't know how much data you would actually need to get things to settle, but I think you'd have to put a limit of at least 100 minutes for it to be reasonable.

 

By this token, the reason why Howard and Beverley are missing is because they will only have played in lineups that have accumulated a lot of minutes together. These lineups will naturally be in the middle as the good and bad performances even out. Which is not to say that there aren't interesting things to be found in the lineup data, it's just that there's not enough of it yet. There's a reason why RPM and similar stats need about three seasons worth of data to be useful - when parsing things out of on/off court data it's really difficult to account for the noise.

 

ST

^This

On another note, you also have to look at the different dynamics of the lineups that were posted and the effects they have on opposing teams playing style. Even if the sample sizes were larger.

 

When Howard and Beverley are on the floor, teams are wise to push the pace. When those two guys are set on defense, they are much deadlier than in transition. So teams will try and score early and naturally increase their Per 36. Doesn't necessarily mean they're more efficient at the faster pace either. Inversely, with those two on offense, particularly Howard, we slow our game down to give him his touches and naturally the Per 36 pace falls off some. I'm not saying this is good for us, because it's probably not, but S O C K S.

 

The high-scoring lineups mentioned are designed to pick up the pace. The tenacious defensive units mentioned are designed to generate transition opportunities, so naturally opposing teams are going to try and slow the game down on both ends with these squads on the floor.

 

To me, more than anything this post illustrates the extreme diversity we have in our rotations and the ability to control the game and create mismatches at will. It's not much of a statement on actually comparing our lineups against each other. Different opponents require different implementations to beat them and we have an awesome tool box.

[thejohnnygold]

S O C K S = eso si que es

 

I had to google that--very clever. 

 

Also, good post (as always), Alituro.

 

Regarding the topic, I think all of our eye tests are telling us the same thing--and that thing is good.  Very, very good.  There will be times when it bites us in the behind (too much risk-taking on defense, too many turnovers/bad shots taken on offense), but I think if Prigioni gets more playing time with that second group it will help limit the offensive part and our bench is going to propel us to greatness.  Onward!

 

[thejohnnygold]

For everyone's enjoyment (and relative to the topic) here is a video from Clutchfans of that 4th quarter run.

 

[YaoMan]

For everyone's enjoyment (and relative to the topic) here is a video from Clutchfans of that 4th quarter run.

 

That was one of the better runs I have seen this season especially against a team that the Rox usually falter against...

[thejohnnygold]

It should be noted that they did this against Matt Barnes, Jamal Crawford, Big Baby Davis, Hedo Turkoglu, and Austin Rivers.  Still awesome to watch.

[YaoMan]

Which showcases the exceptional bench that the Rockets have!

[NorEastern]

Data derived from line ups is fraught with difficulties. The depth of the data may at times be deep, but the breadth is as shallow as can be imagined. There is one metric based on line ups that I do pay attention to. xRAPM. But there are numerous proprietary adjustments made to achieve its excellent results. Perhaps line up data could be blended with SportVU input to achieve excellent analyses. I have a shortcut to the R statistical suite on my desktop and cannot grasp how to do that analysis.

 

On another topic - the Rockets bench. The bench is not yet excellent. Since acquiring Smith and Brewer they are 10th in the NBA in points scored. Once Howard comes back the bench will have three NBA starters sitting on it. That is when it may shine brightly.

 

The Rockets and injuries this season. What does not kill you makes you stronger, as Nietzsche said. The Rockets are a much much stronger team because of their development of personnel. When the line up shortens to eight players in the play offs the development of D-Mo will pay huge dividends.

 

Is it time to do a post on the contract quandaries faced by Morey this summer? Who do the Rockets seek to retain, and what will be the cost? If it comes down to one of Brewer, Bev and Smith which one should be retained?

[Losthief]

I vote brewer, though if smith is willing to stay for a pay cut (possible he's still gonna be getting his DET money) his value might be higher. I like bev but I think he's replacable in the long run, and for the amount we're going to pay him maybe try to bring Llull over?

[thejohnnygold]

I think we prioritize Smith, Brewer, and then Bev.  From what I can tell, Bev has a qualifying offer for $1.18M so that does not seem to be an issue.  Brewer's player option is for $4.9M--I think he takes that.  He's been a journeyman much of his career and only tasted the playoffs a bit.  He did get a ring with Dallas, but only played in 6 games about 4 minutes per game.

 

He has found a solid role, solid minutes, on a solid team.  I think he stays at least for the remainder of the deal and then might cash in after that.

 

Morey will find a way to retain Smith.  He has coveted him for a while and he's not going to let a few dollars or salary cap restrictions get in the way of that.

 

So, my vote is we keep all three, but if anyone goes it will be Bev.  Nick Johnson can do what Bev does more or less.

 

The bigger question is where do we get money for KJ McDaniels?  Do we just sign and trade him this Summer?  Package him with the New Orleans pick, Papa, and Beverley for a PG upgrade?  I kind of like that idea.  That's all very far off.  How we fare in the playoffs will determine the much of the Summer.

[Losthief]

hadn't even thought about him (KJ) and the pellies pick. Is KJ seen as more valuable around the league tradewise than canaan? explains a bit about including the 2nd round pick in that trade if so.

[NorEastern]

I vote brewer, though if smith is willing to stay for a pay cut (possible he's still gonna be getting his DET money) his value might be higher. I like bev but I think he's replacable in the long run, and for the amount we're going to pay him maybe try to bring Llull over?

Smith presents a very interesting contract situation. Because Smith's contract was "stretched" he will be earning ~$5.4 million a season for quite a while. Every penny Smith is paid by an NBA franchise will be deducted from that $5.4M. It is a quandary.

[NorEastern]

hadn't even thought about him (KJ) and the pellies pick. Is KJ seen as more valuable around the league tradewise than canaan? explains a bit about including the 2nd round pick in that trade if so.

KJ's upside, because of his size and athleticism, is well above Canaan's. However the Rockets are "hiding" him so to speak. No minutes means that NBA teams cannot get a read on him. No read, no big contract offers.

[NorEastern]

I think we prioritize Smith, Brewer, and then Bev.  From what I can tell, Bev has a qualifying offer for $1.18M so that does not seem to be an issue.  Brewer's player option is for $4.9M--I think he takes that.  He's been a journeyman much of his career and only tasted the playoffs a bit.  He did get a ring with Dallas, but only played in 6 games about 4 minutes per game.

Brewer declined his player option for the upcoming season. Basically he is a blank contract slate. While he has produced well for the Rockets, his career averages do not justify a $4.6M contract.

[Losthief]

doesn't the fact he clear waivers mean DET is on hook for the whole thing though? Or is a season by season basis? (Josh Smith)

[NorEastern]

doesn't the fact he clear waivers mean DET is on hook for the whole thing though? Or is a season by season basis? (Josh Smith)

If a player is waived his salary is automatically "stretched" to twice the contract length plus one year. Therefore Smith is actually earning a little less than $6M a season. Any NBA salary he earns is deducted from that amount.

[thejohnnygold]

Guys like KJ McDaniels come and go every year it seems.  Most of them end up going, but GM's can't help themselves.  The potential is too tempting.  My theory on these guys is most spend so much of their lives being so much more athletic than everyone else they don't develop solid skills--like shooting the ball--and once they get to the NBA (where everyone is hyper-athletic) they hit a wall they can't get over.

 

Here is KJ's shooting chart.  It's not good, but it's not hopeless either--it really could go either way.

 

 

Playing for Philly doesn't help.  I agree that we are not likely to see much of him this season.  The coaches are probably working to see if he can improve his 3 pt. shooting.  If so, we keep him.  If not, he's gone.

[NorEastern]

Guys like KJ McDaniels come and go every year it seems.  Most of them end up going, but GM's can't help themselves.  The potential is too tempting.  My theory on these guys is most spend so much of their lives being so much more athletic than everyone else they don't develop solid skills--like shooting the ball--and once they get to the NBA (where everyone is hyper-athletic) they hit a wall they can't get over.

 

Here is KJ's shooting chart.  It's not good, but it's not hopeless either--it really could go either way.

 

 

Playing for Philly doesn't help.  I agree that we are not likely to see much of him this season.  The coaches are probably working to see if he can improve his 3 pt. shooting.  If so, we keep him.  If not, he's gone.

KJ has a reputation for being an excellent defender. And he is taller than most point guards and can jump out of the gym. If he can hit the corner three at 35%+ he will be a valuable NBA player. And yes players can improve their three point shot. Just look at D-Mo.

 

Houston gets a couple of dozen wide open looks at the three every game. KJ could be a valuable addition down the line.