The Offense and Shot Distribution

[dysgenic]

Discussions like this make me think of Terry Francona's philosophy which is to combine what you see with your eyes, with the available stats in order to come up with a plan.  In this case, for me there is not too much of a difference.  What I see:

Green should be getting more shots.
I don't think Ray necessarily needs more shots.
KG should be getting a couple more shots.
PP should be getting more shots.  When he is going well especially, they don't seem to look for him enough.
BBD should be getting less shots.
Rondo should be getting a couple less shots.
Delonte should be getting a couple more shots.
Kristich should be thrown off the team (jk).  Seriously, I think he should get a couple less shots.

TEAM: should be getting a couple more shots per game.  Not fast breaking/early office as much as they should.
Not taking advantage of mismatches as much as they should.  Tommy brought up that when he was coaching, he liked to exploit speed mismatches and the Celts don't do it enough.  Great point.

Do the stats posted in the OP bear most of this out?  I think they do, with a couple of tweaks.

 

[kozlodoev]

Did you even bother to read the paper I linked?

It's funny that the paper is authored by a Brian Skinner who is apparently not Brian Skinner, the basketball player.

Also, it's a little annoying that he already published a paper I had an idea to consider

[Fafnir]

OT aside: The whole Einstein/Newton comparison isn't a good one in my view. Physicists didn’t accept Newtonian physics as Gospel. Instead they kept working at the remaining problems that were unexplained or unsolved. They started to find holes in the  Newtonian framework, their experimentation eventually paved the way to Einstein’s discoveries and modern physics.

I don't think anyone other than Dave Berri with wins produced thinks they have an advanced basketball stat that answers everything. Its all a work in progress in trying to find better tools to evaluate the game.

No i did not click through to your link. I do not have time at the moment.

You really should read what someone posts before you respond to it.

The entire second half of your post, your "statistical sense" about how having an efficient shooter "Ray Alen" take more shots at diminished efficiency is better for the team than having BBD take his current share, is what a large section of that paper is about.  It is similar to Braees’ paradox from network theory in that common sense of optimization (or individual optimization) for a team doesn't necessarily give you the optimal result.

And I gave you an explanation for the free throw modifier, I’ll say it again.  It is based on studies of the average number of free throws per possession using years of aggregate data from the NBA. I believe Hollinger came up with .44 from his own study of it, the first estimate used was typically .4 which was posited by Dean Oliver (some sites still use .4 others use .44) The number of technical fouls, “and-1s”, three shot fouls, and one shot violations are all baked into that estimate.

It is not perfect, but it’s a reasonable standard backed up by evidence.

[Fafnir]

Did you even bother to read the paper I linked?

It's funny that the paper is authored by a Brian Skinner who is apparently not Brian Skinner, the basketball player.

Also, it's a little annoying that he already published a paper I had an idea to consider

I know it took me an extra five minutes to find that paper because I had to sort through all sorts of Brian Skinner news clips!

[Carhole]

OT aside: The whole Einstein/Newton comparison isn't a good one in my view. Physicists didn’t accept Newtonian physics as Gospel. Instead they kept working at the remaining problems that were unexplained or unsolved. They started to find holes in the  Newtonian framework, their experimentation eventually paved the way to Einstein’s discoveries and modern physics.

I don't think anyone other than Dave Berri with wins produced thinks they have an advanced basketball stat that answers everything. Its all a work in progress in trying to find better tools to evaluate the game.

No i did not click through to your link. I do not have time at the moment.

You really should read what someone posts before you respond to it.

The entire second half of your post, your "statistical sense" about how having an efficient shooter "Ray Alen" take more shots at diminished efficiency is better for the team than having BBD take his current share, is what a large section of that paper is about.  It is similar to Braees’ paradox from network theory in that common sense of optimization (or individual optimization) for a team doesn't necessarily give you the optimal result.

And I gave you an explanation for the free throw modifier, I’ll say it again.  It is based on studies of the average number of free throws per possession using years of aggregate data from the NBA. I believe Hollinger came up with .44 from his own study of it, the first estimate used was typically .4 which was posited by Dean Oliver (some sites still use .4 others use .44) The number of technical fouls, “and-1s”, three shot fouls, and one shot violations are all baked into that estimate.

It is not perfect, but it’s a reasonable standard backed up by evidence.

But Braess's paradox seems to be focused on individual bodies choosing "selfishly"

As I said in an earlier post if PP wants 10 more shots a game that is bad if he wants to take 2 or 3 of the good shots he passes up a game that is good for the overall efficiency.

I also never said that "Ray Allen" should take more shots. I said the opposite that of the big 3 he takes the highest percentage of good shots he is presented with.

"you really should" check what someone posted above before youo get all "" in your response

[Fafnir]

I also never said that "Ray Allen" should take more shots. I said the opposite that of the big 3 he takes the highest percentage of good shots he is presented with.

"you really should" check what someone posted above before youo get all "" in your response

You misunderstand what I'm referring to when I put "Ray Allen" in quotes. Ray Allen is the name given to the hypothetical shooter in the paper I referenced. I did substitute BBD in for the 4 nameless teammates he's paired with in the hypothetical offensive network that is analyzed.

I wasn't refering to your post at all.

Braee's paradox deals with individual actors. But what the paper does is model each player as a potential path to a basket. Thus each possesion becomes a individual actor in modeling.

I did read your post, I was directly responding to the second portion where you posit that its going to be better for the team if BBD shot attempts are reduced and others pick up a few extra shots. That's a very common sense viewpoint, one that the paper I linked to examines and challenges.

[Carhole]

I also never said that "Ray Allen" should take more shots. I said the opposite that of the big 3 he takes the highest percentage of good shots he is presented with.

"you really should" check what someone posted above before youo get all "" in your response

You misunderstand what I'm referring to when I put "Ray Allen" in quotes. Ray Allen is the name given to the hypothetical shooter in the paper I referenced. I did substitute BBD in for the 4 nameless teammates he's paired with in the hypothetical offensive network that is analyzed.

I wasn't refering to your post at all.

Braee's paradox deals with individual actors. But what the paper does is model each player as a potential path to a basket. Thus each possesion becomes a individual actor in modeling.

I did read your post, I was directly responding to the second portion where you posit that its going to be better for the team if BBD shot attempts are reduced and others pick up a few extra shots. That's a very common sense viewpoint, one that the paper I linked to examines and challenges.

I read the paper. So do you believe this to be true of this celtics team or of all teams in general? You do not see flaws with how this unfolds in a real game?

It is an interesting 15 pages of information but I do not agree with its application to this team because many of the assumptions needed do not add up for me.

[Fafnir]

I also never said that "Ray Allen" should take more shots. I said the opposite that of the big 3 he takes the highest percentage of good shots he is presented with.

"you really should" check what someone posted above before youo get all "" in your response

You misunderstand what I'm referring to when I put "Ray Allen" in quotes. Ray Allen is the name given to the hypothetical shooter in the paper I referenced. I did substitute BBD in for the 4 nameless teammates he's paired with in the hypothetical offensive network that is analyzed.

I wasn't refering to your post at all.

Braee's paradox deals with individual actors. But what the paper does is model each player as a potential path to a basket. Thus each possesion becomes a individual actor in modeling.

I did read your post, I was directly responding to the second portion where you posit that its going to be better for the team if BBD shot attempts are reduced and others pick up a few extra shots. That's a very common sense viewpoint, one that the paper I linked to examines and challenges.

I read the paper. So do you believe this to be true of this celtics team or of all teams in general? You do not see flaws with how this unfolds in a real game?

It is an interesting 15 pages of information but I do not agree with its application to this team because many of the assumptions needed do not add up for me.

If KG/Pierce/Allen shot more they'd likely see their efficiency drop, and that would hurt the team's offense. Its a somewhat counter-inituitive result of analyzing networks.

I don't think this is always going to hold, but this sort of analysis is definitely worth thinking about with our current team and its very even usage pattern.

I think the KG/Pierce/Allen's offensive skills have diminished to the point that they are no longer best utlitzed as higher usage players. I also think that our offense is at its best when shot attempts are relatively even and the ball is moving. Our team has perhaps the most even shot distribution in the league, I think this is largely responsible for our current level of offensive success.

The paper itself doesn't claim to answer questions in real life basketball, but it does point to how higher efficiency options shouldn't always be used more in a team setting.

[Fafnir]

Discussions like this make me think of Terry Francona's philosophy which is to combine what you see with your eyes, with the available stats in order to come up with a plan.  In this case, for me there is not too much of a difference.  What I see:

Green should be getting more shots.
I don't think Ray necessarily needs more shots.
KG should be getting a couple more shots.
PP should be getting more shots.  When he is going well especially, they don't seem to look for him enough.
BBD should be getting less shots.
Rondo should be getting a couple less shots.
Delonte should be getting a couple more shots.
Kristich should be thrown off the team (jk).  Seriously, I think he should get a couple less shots.

TEAM: should be getting a couple more shots per game.  Not fast breaking/early office as much as they should.
Not taking advantage of mismatches as much as they should.  Tommy brought up that when he was coaching, he liked to exploit speed mismatches and the Celts don't do it enough.  Great point.

Do the stats posted in the OP bear most of this out?  I think they do, with a couple of tweaks.

I'd love for Rondo to get more shot attempts at the rim.

[Chris]

I think the KG/Pierce/Allen's offensive skills have diminished to the point that they are no longer best utlitzed as higher usage players. I also think that our offense is at its best when shot attempts are relatively even and the ball is moving. Our team has perhaps the most even shot distribution in the league, I think this is largely responsible for our current level of offensive success.

The paper itself doesn't claim to answer questions in real life basketball, but it does point to how higher efficiency options shouldn't always be used more in a team setting.

I just think the two biggest killers of offense in the NBA (or at least on the C's) is stagnancy, and predictability.

If an offense is not getting movement, both of the players and the ball, then the defense has the advantage, and they will generally end up with a bad shooter taking a bad shot. 

When the offense is predictable, its the same thing.  NBA defenses are in general, pretty smart.  If they know one guy is going to shoot it, one guy is going to drive, and one guy is going to pass, then it means they can basically take away half the floor, which again, leads to a bad shooter taking a bad shot.

So, the solution is you have to mix it up.  You have to move the ball, and the players, and you cannot fall into too much of a repetitive rhythm.

This means you need the bad shooters to shoot, when they have a good shot.  It also means, sometimes you need the shooters to drive, drivers pass, etc.  This opens up the floor, and does not let defenses to take away the shots for the shooters, drives for the drivers, and passes for the passers.  Even though they are doing something that the defense knows is not their strength, the defenders still will react to it if it is done in rhythm. 

They may lay off Rondo shooting, but if he takes a jumpshot in rhythm, they are going to run at him, because they won't have a chance to think about it.  That also may move them up an inch or two closer, so the next time down, the passing lane to Ray in the corner will suddenly open up, or the driving lane will be there with the use of a pump fake. 

However, if he catches the ball at the elbow, and does not even look at the shot, the defense will sag, and clog the passing lane, forcing them to break the offense, or the shotclock will run down forcing him to take a flat footed jumper.

Unfortunately, stats can't show these things.

Sure, sometimes they overpass, or sometimes guys will take an ill advised shot, but that is part of basketball.  As soon as they get it in their head that they should not be shooting, is when the offense will completely stall.

[BballTim]

I think the advanced stats are much better than the basic stats and provide some useful insight but I am still very much a believer in "watching and understanding the game" tells you everything you REALLY need to know.

  Agree, with the caveat that stats can tell you a lot if you know when to use them. If I see a stat that disagrees with what I see I consider whether my opinion is wrong or if the stat doesn't show what I think it does (or what it claims to).

I find it very interesting how people refer to the new stats as some sort of gospel. In this thread it has been the statistical assumption that FTA are basically 44% of a possession. And when people dont fall right in line with that their is a strong implication that they are somehow "uninformed" and less accurate with their assessments.

  I don't think this is really the case. It's not important to the discussion that FTA are basically 44% of a possession. What is important is that free throws usually use a possession. If PP makes a layup we get 2 points from that possession. If he gets fouled going up for that layup, misses the shot and makes both free throws we still get the  same two points from the possession. Neither possession was more efficient (scoring-wise) than the other.

  From a points per shot perspective, one possession netted 2 points on one shot and the other netted 2 points from 0 shots. This shows one possession to be much more efficient (scoring-wise) than the other. That's not really the case. What's important is allowing for this discrepency, not believing explicitly that a FTA automatically equals .44 possessions.

Someone somewhere did their own individual analysis to come up with these now accepted statistical "facts" but what is the proof that a fta is 44% of a possession in how it effects a game? It is just a guys/girls determination.

  I think that many discussions about statistics have a "forest vs trees" issue, where people lose perspective and can't see the forest for the trees. Assuming you believe that significantly more free throws are the result of 2 shot fouls than and-1s and technicals, FTA tends to make up a small enough percentage of a player's total possessions that the .44 being 10% off would only lead to a player's total possessions being 1% or so off. If two points per possession numbers being discussed are with a few percent of each other (again, forest vs trees) people should understand that the margin of error too large to reach any conclusions. For the current discussion, the ppp for Baby and Rondo are 15% or more off from the big three, so that a 1-2% difference from that statistical .44 is unimportant.

[ballin]

Discussions like this make me think of Terry Francona's philosophy which is to combine what you see with your eyes, with the available stats in order to come up with a plan.  In this case, for me there is not too much of a difference.  What I see:

Green should be getting more shots.
I don't think Ray necessarily needs more shots.
KG should be getting a couple more shots.
PP should be getting more shots.  When he is going well especially, they don't seem to look for him enough.
BBD should be getting less shots.
Rondo should be getting a couple less shots.
Delonte should be getting a couple more shots.
Kristich should be thrown off the team (jk).  Seriously, I think he should get a couple less shots.

TEAM: should be getting a couple more shots per game.  Not fast breaking/early office as much as they should.
Not taking advantage of mismatches as much as they should.  Tommy brought up that when he was coaching, he liked to exploit speed mismatches and the Celts don't do it enough.  Great point.

Do the stats posted in the OP bear most of this out?  I think they do, with a couple of tweaks.

 

OP here, and yes, I agree with your conclusions. Like you, I try to mesh what I'm seeing with what the stats say. Problems arise when you rest your opinion too heavily on one side or the other.

For instance, as much as I LOVE wins produced (best advanced stat, in my opinion) David Berri admittedly has his head up is butt. He literally made a post about Kyle Lowry for MVP the other day based on wins produced. Pretty hilarious actually.

On the other hand, people that like what they're seeing with BBD and Rondo taking long jumpers are equally insane. "Taking the open shot" somehow fits meshes with their mental aesthetic as to what an efficient offense is supposed to look like. Same thing with (over)passing the ball. And yes, while it looks like the right thing to do, and in fact many of us have been taught growing up that it's the right thing to do, in reality, it clearly ISN'T. The stats clearly show that it is better for the Big 3 to attempt just about any type of shot than for Rondo or BBD to take a long jumper. Classic basketball be [dang]ed, but it's true.

[Fafnir]

The stats clearly show that it is better for the Big 3 to attempt just about any type of shot than for Rondo or BBD to take a long jumper. Classic basketball be [dang]ed, but it's true.

I'm not sure about this. A bad contested shot by one of them is likely worse than an open two point look for Rondo/Davis.

This is one of those cases where I really wish I had access to a database that tracked open shooting percentages.

[BballTim]

For instance, as much as I LOVE wins produced (best advanced stat, in my opinion) David Berri admittedly has his head up is butt. He literally made a post about Kyle Lowry for MVP the other day based on wins produced. Pretty hilarious actually.

  Did you read the post? He wasn't really saying that Lowry should be an MVP candidate. His point was that Rose shouldn't because his "wins production" (or however he refers to it) is close to Lowry's, not close to the league leaders.

[ballin]

I also want to address a few points that have been made:

As I expected, a lot of people have come in this thread reiterating the same old usage argument. These posts all follow the same formula where the poster says that if the Big 3 take more shots they'll be less efficient for reasons x, y, and z without even attempting to address the stats I have brought up in the original post which show that the usage argument is bogus.

To use a slightly different example than the Big 3, look at BBD:

Glen Davis (last 3 years):
FGA               TS%
5.9              .502
5.2              .500
10.2             .504

Glen Davis essentially doubled his number of shots over the course of one year and his TS% went UP. Further, when he decreased shots the previous 2 years his TS% went DOWN. This is more evidence that the number of shots one takes isn't strongly correlated efficiency. Partially or slightly correlated? Sure, maybe. Strongly? Not at all.

The bottom line is this: the Big 3 could definitely take a lot more shots without their efficiency dropping significantly, and they should do so.

The usage argument is a load of junk. It really, really is. If the correlation exists, it's very weak. And nobody has given me stats that prove otherwise, and until they do, I'm not inclined to agree with their position. Yet it's so very hard for people to let go of the usage argument because it's so-called "common sense". It sounds logical. GM's, coaches, players, the media, and most fans all believe it.

But there is just mounting evidence showing it's just. not. true. What it boils down to is this: good players can almost always get "their" shots. They can get to their spots on the floor and get what they want offensively due to the rules the NBA has on defense. Efficient shooters tend to be efficient shooters regardless of the number of shots they take. And if you really don't want to believe this, don't take my word for it. Go scour the TS% history of dozens of players like I have and see what you find. You might be surprised at what you see.