2016 RAPM/RPM/etc. *Full 2016 RS + PS RPM & RAPM Updated 6/24*

[MyUniBroDavis]

"A lot" so how much exactly in comparison to his numbers with Green? And bTW, such roster means he had very good spacing. And 88 threes isn't small sample. His drop off here - almost 8 percentage points - is significant.

It's not significant. Using a Binomial distribution, you would expect a 45% 3P shooter to make 34/88 or less about 14% of the time.

Looking at it in spots makes it seem like sample size

The areas effected by playmaking, the "regular 3 point shots" (not 30+ feet) still is above 40%

Add in that he is shooting 88 threes in 241 minutes, 1.3 more per 36 than this year

Which is larger than the jump between 2013 and 2014

And I think it's pretty normal if he was making his usual percentage from 30 feet out his 3p% is above 40%

[lorak]

Green without Curry is better in every box score category except of FG%, that includes better 3p% with increased volume as well as tov are lower with more asisst.

On the other hand Curry is way worse shooter without Green, his 3p% is down by almost 8 percentge points (so maybe we should start talking about how having elite playmaker like Green helps Curry?), tov are also higher, with the same amount of ast, but bigger volume.

And Green's +/- w/o Curry is better than Steph's w/o Draymond.

So maybe instead of using small sample excuse (while BTW these samples are not so small, more than 10 full games) try to find resonable explanation why results over and over again favour Green. And by resonable I mean something better than "GSW have more plays w/o Green" - as there is no way we can check if that's true or not.

Since this came right after mine I'm gonna assume you are referring to me.

1.

Again, this was me using the-powers third point in his long post a few pages ago. I can't remember what it said but it said something about how using it in spots vs using it for a whole game and how sustainable it is.


2.

Curry in this sample size attempted 88 threes in 241 minutes.
That's 13.1 threes per 36 minutes, he is averaging 11.8 per 36 this year
As for sample size

Corner threes. 7 attempts, 2/7 (keep in mind, he Has a career average of 52.5% from there)

24-29 feet
68 attempts, 41.2% 28/68

30+ feet
18 attempts, 30%
6/18

For the season

48.2% from the corner
44.8% from 24-29 feet
46.7% from 30+ feet

So the corner, he shot 7. That's sample size (and no, he isn't shooting less of them than usual either)
From normal three point range it's a 3.5% drop
From 30+ feet it's a 16.7% drop
So unless green is getting him open looks from 30 feet I don't think it's as big of a drop as you are saying.

Curry's turnover percentage is 11%, lower than his season average
Assist rate is 35.3%, slightly higher than his season average
As for sample size,

Curry with Bogut on the floor and draymond off 1.18 defensive rtg
Curry with Bogut off the floor and draymond off 1.063 defensive rtg

So either Bogut is the worst defender in the NBA
Or it's sample size

Or you can't draw such conclusion based on one data point...
In Curry and Green case there are multiple data points showing more or less the same. That's BTW also one of the reasons why trying to "explan" it by small sample is unresonable. You need more than that, for example if "different plays" are explantion, then you have to show how GSW playbook changes with and without both players or how opponents defenses focus differently. BUt no one of you did that, you just trying to provide ad hoc "explanations" without backing it up with facts.

[MyUniBroDavis]

Green without Curry is better in every box score category except of FG%, that includes better 3p% with increased volume as well as tov are lower with more asisst.

On the other hand Curry is way worse shooter without Green, his 3p% is down by almost 8 percentge points (so maybe we should start talking about how having elite playmaker like Green helps Curry?), tov are also higher, with the same amount of ast, but bigger volume.

And Green's +/- w/o Curry is better than Steph's w/o Draymond.

So maybe instead of using small sample excuse (while BTW these samples are not so small, more than 10 full games) try to find resonable explanation why results over and over again favour Green. And by resonable I mean something better than "GSW have more plays w/o Green" - as there is no way we can check if that's true or not.

Since this came right after mine I'm gonna assume you are referring to me.

1.

Again, this was me using the-powers third point in his long post a few pages ago. I can't remember what it said but it said something about how using it in spots vs using it for a whole game and how sustainable it is.


2.

Curry in this sample size attempted 88 threes in 241 minutes.
That's 13.1 threes per 36 minutes, he is averaging 11.8 per 36 this year
As for sample size

Corner threes. 7 attempts, 2/7 (keep in mind, he Has a career average of 52.5% from there)

24-29 feet
68 attempts, 41.2% 28/68

30+ feet
18 attempts, 30%
6/18

For the season

48.2% from the corner
44.8% from 24-29 feet
46.7% from 30+ feet

So the corner, he shot 7. That's sample size (and no, he isn't shooting less of them than usual either)
From normal three point range it's a 3.5% drop
From 30+ feet it's a 16.7% drop
So unless green is getting him open looks from 30 feet I don't think it's as big of a drop as you are saying.

Curry's turnover percentage is 11%, lower than his season average
Assist rate is 35.3%, slightly higher than his season average
As for sample size,

Curry with Bogut on the floor and draymond off 1.18 defensive rtg
Curry with Bogut off the floor and draymond off 1.063 defensive rtg

So either Bogut is the worst defender in the NBA
Or it's sample size

Or you can't draw such conclusion based on one data point...
In Curry and Green case there are multiple data points showing more or less the same. That's BTW also one of the reasons why trying to "explan" it by small sample is unresonable. You need more than that, for example if "different plays" are explantion, then you have to show how GSW playbook changes with and without both players or how opponents defenses focus differently. BUt no one of you did that, you just trying to provide ad hoc "explanations" without backing it up with facts.

1. You mean the data point that is essentially the reason why curry with green off the court is worse than green with curry off the court team wise? (Btw, 61 minutes of ezili point to this as well, 117 offensive rtg and 95 defensive rtg, with Thompson it's 117 offensive rtg and 110 defensive rtg, and yes, many minutes with Bogut)

2. That was hardly a main point I was making, merely restating way has already been said

3. So the fact that the Warriors defense turns from league average to worst defense in NBA history adding Bogut, who is a solid defender, doesent show that noise is a problem?

4. Are you saying that if Bogut have regular impact on defense, rapm wouldn't be effected?

5. viewtopic.php?p=46576671#p46576671

6. As for plays, I can't prove it, but in 5 full games without curry, the Warriors offense in 2 of those 5 games was pretty horrible, in 2 more of those 5 games benefitted due to offensive rebounding and opponent turnovers, and only had 1 real game where they were very good offensively.

7. The difference in their rapm is mostly defense, and to separate them from each other the only sample size they can use, since they are on the same team, is 436 minutes from green without curry and 241 minutes for the opposite.

The offensive numbers out curry on top. Larger advantage in wowy than in rapm.

The defensive numbers show a huge gap. But let's say Bogut has his impact and the defense with curry and Bogut is 103, which is fair.
So the defensive rtg with curry is sub 105.

Right now it's above 110.

And let's say that boosts his rapm (and yes, I'm pretty sure it would have) defense quite a bit, up to a 2 or a 2.5.

Would we then say there is a problem with noise? Yeah, yeah we would.

I mean, right now his defense puts him at 2nd of pg if I'm not mistaken. I don't see that as wrong but everyone laughs at the notion of him being a top 5-10 pg defender in the league.

So let's say his defense is magnified by being next to green. Considering that curry is clearly more impactful on offense alone than green is on defense alone, could we say that, since the sample size is so small with them both seperate, that this effect could be magnified?

I don't know much about rapm, so I apologize if I'm wrong on this.

[therealbig3]

Green without Curry is better in every box score category except of FG%, that includes better 3p% with increased volume as well as tov are lower with more asisst.

On the other hand Curry is way worse shooter without Green, his 3p% is down by almost 8 percentge points (so maybe we should start talking about how having elite playmaker like Green helps Curry?), tov are also higher, with the same amount of ast, but bigger volume.

And Green's +/- w/o Curry is better than Steph's w/o Draymond.

So maybe instead of using small sample excuse (while BTW these samples are not so small, more than 10 full games) try to find resonable explanation why results over and over again favour Green. And by resonable I mean something better than "GSW have more plays w/o Green" - as there is no way we can check if that's true or not.

I simply disagree that spot minutes during the RS against random teams when Curry goes to the bench proves anything, when the point is, against a focused defense (aka in the playoffs) that doesn't have to prepare for Curry, Green is most likely going to struggle...which we've actually seen happen against the Rockets so far. And it's the Rockets, not exactly the Spurs.

And instead of about 4.3 mpg without Curry (which is what 346 minutes over 81 games averages out to), he's played about 30.3 mpg without Curry during the 3 games, with the Rockets actually being able to prepare for him rather than Curry. It's easier to adjust your defense over the course of 30 minutes than it is over the course of 4 minutes.

And we've also seen that Green himself has not been very efficient in the 3 games that Curry missed during the RS. I'm interested to see how he looks in games that Curry straight up misses (including the 2 playoffs games, giving us 5 games total) vs how he plays in the minutes that Curry is simply on the bench. I'll check on that later.

[lorak]

I simply disagree that spot minutes during the RS against random teams when Curry goes to the bench proves anything, when the point is, against a focused defense (aka in the playoffs) that doesn't have to prepare for Curry, Green is most likely going to struggle...which we've actually seen happen against the Rockets so far. And it's the Rockets, not exactly the Spurs.

So what exactly Rockets did differently on defense (vs Green) during games 2 and 3 than in G1? Really, if your hypothesis about different defensive preparation is true, then you should be able to point out differences.

And instead of about 4.3 mpg without Curry (which is what 346 minutes over 81 games averages out to),

Without "each other sample" is more than 500 minutes, that means more than 10 full games. That's hundreds of possessions and that's valuable information. Some of guys here use less games to judge players. And it's not like that's the only data point showing how valuable is Green. That information is consistent with others and thus explaining it by "small sample" doesn't seem convincing.

[lorak]

...

if your hypothesis (teams prepare differently to games when Curry is out, and thus off the floor samples don't have value) is true, then we should see some changes in Green's shooting splits. Especially in amount of his shots that are contested, because if without Curry teams prepare better (and Steph creates so much for Drymond), then we should see more contested shots for Green. But that's not the case. In 5 games w/o Curry 55,2% of Green's shots are tight/very tight deeded, while with him 56,7% So in fact without Curry Green has slightly more open shots.

So once again - what exactly do you have to back up your hypothesis?

[bballexpert]

...

if your hypothesis (teams prepare differently to games when Curry is out, and thus off the floor samples don't have value) is true, then we should see some changes in Green's shooting splits. Especially in amount of his shots that are contested, because if without Curry teams prepare better (and Steph creates so much for Drymond), then we should see more contested shots for Green. But that's not the case. In 5 games w/o Curry 55,2% of Green's shots are tight/very tight deeded, while with him 56,7% So in fact without Curry Green has slightly more open shots.

So once again - what exactly do you have to back up your hypothesis?

All i do know is that green has fallen big time in ps against the bottom rockets i mean he is putting up 11 points on 417 ts 13 per. I am pretty sure that him playing with curry makes him leaps better he been look just ok with out him in the ps.

[therealbig3]

...

if your hypothesis (teams prepare differently to games when Curry is out, and thus off the floor samples don't have value) is true, then we should see some changes in Green's shooting splits. Especially in amount of his shots that are contested, because if without Curry teams prepare better (and Steph creates so much for Drymond), then we should see more contested shots for Green. But that's not the case. In 5 games w/o Curry 55,2% of Green's shots are tight/very tight deeded, while with him 56,7% So in fact without Curry Green has slightly more open shots.

So once again - what exactly do you have to back up your hypothesis?

Well, I have some questions then:

In that sample, how many shots is Green taking without Curry vs with? For example, 5 uncontested shots out of 10 is 50% uncontested, but 4 uncontested shots out of 5 is 80% uncontested...but Green is getting more opportunities to shoot the other way (and I don't know what's being called "uncontested" or not...is an open 3 from Green with a defender closing out late considered "contested"? Or if he's attacking the rim with minimal resistance, is that "contested"?)

Also, what kind of shots is he getting? The defense might be more willing to give up open perimeter shots, but defend his opportunities at the basket much better.

As for your questions with regards to proof, I have to watch the Rockets games more closely and get back to you. I'm busy with other stuff right now that I haven't been able to watch as many games as I would like, nor have I been able to focus as much as I would like on the playoff games I have been able to catch.

[MyUniBroDavis]

...

if your hypothesis (teams prepare differently to games when Curry is out, and thus off the floor samples don't have value) is true, then we should see some changes in Green's shooting splits. Especially in amount of his shots that are contested, because if without Curry teams prepare better (and Steph creates so much for Drymond), then we should see more contested shots for Green. But that's not the case. In 5 games w/o Curry 55,2% of Green's shots are tight/very tight deeded, while with him 56,7% So in fact without Curry Green has slightly more open shots.

So once again - what exactly do you have to back up your hypothesis?

If you are using NBA.stats, I think blackmill posted an article showing that it's kinda bs.

[lorak]

^
Sure, article made by company which fights with sportVU for NBA contracts and so far is loosing... seems really convincing.
Besides, you are free to present any data based argument to back up discussed hypothesis.

therealbig,
Green is shooting more (per game) without Curry - both contested and uncontested shots.

[MyUniBroDavis]

^ I'm preeeeetty sure a lot of people have said why sportsvu tracking isn't exactly perfect.

If we wanted to see, we could just see all of greens shots without curry on the court in those 5 full games in videos, but those aren't exactly here anymore iirc.

And looking at how "open" a shot is is misleading. A runaway fading corner 3 would be open, while a jumper contested from behind would be tight.

Besides, going back to rapm data, the sample size is still too much of a problem. Heck, JE himself said it. The Bogut example still stands, I don't know how to calculate rapm, but I'm sure that had Bogut had typical defensive impact, that the rapm data for curry and green would be much, much closer

Looking at the lineup data really makes a lot of it seem like sample size. 10 full games is not a big sample at all.

Right now, draymond is shooting 44% from 2 with curry off the court, and 38% from 3. Overall his TS is 53.2%, much lower than the 59 TS with curry on the court. Compare this to curry's drop from 67.7 to 60.2, which we know for a fact is noise because of three point percentage.

(His percentage of field goal percentages from 30+ feet drop. This has absolutely nothing to do with green for obvious reasons. Plugging in his normal percentage from there puts him at 42% from 3, and then we have the fact that the sample size meant he only has 7 corner threes, which is his "best area" he is 2/7. Under normal conditions with green he would be closer to 3/7. Again, green or no green he isn't sub 30% from the corner. From regular range his percentages are down somewhat, from 45 to 41. Then again, increased attempts mean he is attempting nearly 1.3 more per 36)

Btw, obviously shots in the paint will rarely be open

With draymond, 30% of curry's shots are within 9 feet

Without, that number is 32.4

Otoh

74.8 of draymond shots are within 9'feet with curry in

49.6 with curry out

Now would draymond shoot 44% from 2 with curry out for a season? Obviously not. Same reason why curry wouldn't shoot 38

[MyUniBroDavis]

^
Sure, article made by company which fights with sportVU for NBA contracts and so far is loosing... seems really convincing.
Besides, you are free to present any data based argument to back up discussed hypothesis.

therealbig,
Green is shooting more (per game) without Curry - both contested and uncontested shots.

More than 70% of draymonds shots are inside the paint with curry on the floor.

Barely 50% with him off the floor


To put that into perspective. Draymond attempts a higher percentage of his shots within 3 feet with curry on the floor than within the 10 feet with curry off the color.

Making curry shoot 42% from 3 in this sample, which would be his percentage if his 30 footers were of normal accuracy, makes the offensive rating nearly 115.

Add in sample size/noise issues in the stats of other players

(Speights and Barbosa are combined 11/32 from 0-3 ft)

Or how outside of curry the team is shooting 65% on free throws (vs 71% outside of green)

Heck, in 124 minutes with Barbosa, speights, and green off the court and curry on it, the Warriors offense is above 120

Again, sample size.

But a 241 minute sample of curry on and green off doesent show enough. The other way around has less noise.

[lorak]

^ I'm preeeeetty sure a lot of people have said why sportsvu tracking isn't exactly perfect.

Nobody says it's perfect, but criticism you are refering to is wrong. (and bTW I'm not using 3,5 feet as definition of close shots - what they criticized in that artice you are refering to.)

http://nyloncalculus.com/2015/10/01/mountains-out-of-molehills-and-a-known-sportvu-issue/

[MyUniBroDavis]

^ I'm preeeeetty sure a lot of people have said why sportsvu tracking isn't exactly perfect.

Nobody says it's perfect, but criticism you are refering to is wrong. (and bTW I'm not using 3,5 feet as definition of close shots - what they criticized in that artice you are refering to.)

http://nyloncalculus.com/2015/10/01/mountains-out-of-molehills-and-a-known-sportvu-issue/

I mean, we have to look at where they took their shots,

Green took ALOT more of his shots out of the paint than in the paint with curry. and I Doubt this can be explained just by 3 point volume, since he shot 44% in the arc without curry vs 55% with curry.

I don't think he will shoot below 50% without curry inside the arc to be honest

And I don't believe curry will shoot under 43% outside it without draymond

[lorak]

Shot locations might be telling a lot, but we also have to keep in mind that Livingston is Curry's replacement, and while he is ok offensive player, he also doesn't have 3p shot - what matters a lot for spacing.

[lorak]

per 100 poss, Green's FGA from different shot locations with Curry on and off (FG% of shots from given location):

Code: Select all

ON   OFF   TYPE
24,1 (60,8%)   23,9 (52,1%)   RA
18,2  (34,9%)   15,9 (46,2%)   PAINT NON-RA
4,8 (34,5%)   6,5 (10,0%)   MIDRANGE
31,3 (38%)   43,8 (42,2%)   3P


Without Curry he takes more threes on better %, but is much more worse from midrange (however that's not too important because he doesn't take a lot of shots from that location) and in RA. That might be "Curry effect", but also "Livingston effect" as he doesn't provide spacing, so with him on the floor teams can focus on defending the rim.

[MyUniBroDavis]

per 100 poss, Green's FGA from different shot locations with Curry on and off (FG% of shots from given location):

Code: Select all

ON   OFF   TYPE
24,1 (60,8%)   23,9 (52,1%)   RA
18,2  (34,9%)   15,9 (46,2%)   PAINT NON-RA
4,8 (34,5%)   6,5 (10,0%)   MIDRANGE
31,3 (38%)   43,8 (42,2%)   3P


Without Curry he takes more threes on better %, but is much more worse from midrange (however that's not too important because he doesn't take a lot of shots from that location) and in RA. That might be "Curry effect", but also "Livingston effect" as he doesn't provide spacing, so with him on the floor teams can focus on defending the rim.

Are you sure those numbers are right?

Curry off the court

57 shots 0-3 feet

13 shots 3-9 feet

10 shots 10-15 feet

16 shots 16-23 feet

50 shots outside the arc

903 possessions

141 shots

Curry on the court

710 total

368 shots 0-3 feet

93 shots 3-9 feet

16 shots 10-15 feet

18 shots 16-23 feet

202 shots 3 point

5195 possessions.

He takes a higher percentage of his shots inside 3 feet both ways.

Also, while there might be a Livingston effect, the difference is way too big.

[lorak]

I'm sure. They are from nba.com without playoffs, you are probably using from nbawowy with playoffs.

And BTW, that effect is seen in whole minutes without Curry (not only games he missed), so that's not backing up hypothesis about bigger importance of games in comparison to on/off.

[MyUniBroDavis]

I'm sure. They are from nba.com without playoffs, you are probably using from nbawowy with playoffs.

And BTW, that effect is seen in whole minutes without Curry (not only games he missed), so that's not backing up hypothesis about bigger importance of games in comparison to on/off.

I'm not referring to that point

I don't see why we wouldn't count playoffs, and even then I doubt the difference in stats is because of a 2 and half game sample

[MyUniBroDavis]

Putting wwy data excluding playoffs still gives 45 inside 3 feet and 45 threes