Good Teammates Make Efficient Shooters?

One issue that comes up now and again is whether a player is likely to shoot more efficiently if he plays alongside other good players. Intuitively, the answer is "yes." If you need to carry the offense single-handedly, the defense can collapse around you and it's hard to score. If you play alongside other stars, by contrast, the defense is spread out and you can get easy shots. But just because it sounds reasonable doesn't make it true. Some evidence is suggested by the above chart of the Celtics new "Big Three," Kevin Garnett, Paul Pierce, and Ray Allen, all of whom shifted from being the clear focal point of an offense to forming one leg of a more evenly-balanced offensive stool.

As we can see, all three saw their shooting efficiency take a tumble last season and now they're doing better. Even more strikingly, all three are putting up career-high numbers in True Shooting Percentage thus far this season. That's not definitive proof that they're scoring more efficiently because they're scoring together but it's suggestive. Certainly it'll be interesting to see if that trend holds up over the course of the season. Another good case to look at would be Carmelo Anthony and Allen Iverson.

Comments

It looks to me like you might be able to make a case for an effect on Garnett, but not the others.

Posted by SomeCallMeTim | November 30, 2007 2:45 PM

When Wages of Wins came out, my main argument against it was that it overvalued Garnett's shooting efficiency. The chart represents whether it's true that efficiency goes down. A different (perhaps more relevant) question is whether it should go down.

Garnett would make the "right" play and pass out of the double team to his inept teammates, keeping his shooting effficiency relatively high. Might it have been more in his team's interest for Garnett to take on the double team, and let his efficiency suffer?

Someone's shooting efficiency will take a dive if a star player is forced to carry the load, but even moreso if he accepts the burden.

Posted by JohnMcG | November 30, 2007 2:53 PM

But just because it sounds reasonable doesn't make it true.

But in this case it probably is. Compare also: Tom Brady's completion percentage and average YPA last year with top targets Reche Caldwell, Troy Brown and Jabar Gaffney v.s. this year with Randy Moss, Wes Welker and Donte Stallworth.

Posted by Brad R. | November 30, 2007 2:59 PM

Much too small a sample (one case) to make any kind of reasonable argument. This may have numbers in it, but it's an anecdote, not any kind of statistical demonstration.

Posted by Antid Oto | November 30, 2007 2:59 PM

Jeez, they're all shooting 58% or better? Holy cow. In a league where a team is doing well to shoot 50%, this is big news. I had no idea. How often has this happened before in the NBA?

Posted by Kit Stolz | November 30, 2007 3:02 PM

Even more strikingly, all three are putting up career-high numbers in True Shooting Percentage thus far this season.

Um... really? Sure looks like Allen was higher in '06, and that Pierce has been at this level many times before and had one bad year. Also, you're aware all of these player have been in the league more than 6 years, right?

Posted by right | November 30, 2007 3:03 PM

I seem to remember Dean Oliver talking a lot about this in "Basketball on Paper", discussing how certain players and their efficient respond to increased (and by extension, decreased) usage, and this dovetails nicely with Ken Rosenbaum's main criticism of WoW (or one of the main criticisms - my disdain for all things Berri is pretty well known): that the method completely ignores 'shot creation' which is uhhh, important in basketball...

Posted by Pooh | November 30, 2007 3:07 PM

It might might be more apples-to-applesish to compare the first 15 some games of each season rather than full seasons vs the first 14 games as we have in the charts above.

Posted by talboito | November 30, 2007 3:09 PM

You also need to factor the effects of injuries into the equation. All three are healthy right now but that certainly hasn't been the case in the past.

Posted by Steven Donegal | November 30, 2007 3:20 PM

Kit Stoltz -

Not shooting percentage, TRUE shooting percentage, which is a statistical measure used to accomodate the fact that the traditional "shots made divided by shots taken" doesn't accomodate the greater value of three pointers. If Player A shot 3/6 from the painted area and Player B shot 2/6 from behind the arc, Player B has scored the same number of points with the same number of shots, but a traditional shooting percentage rubric doesn't capture this. Some measures of "true shooting percentage" or whatever else it's sometimes called also factor in the free throw line in a way I can't recall.

That explanation aside, Pierce, Allen, and Garnett are shooting (traditional percentage) .471 (equals a career high), .435, and .558 (career high by a LONG shot) respectively. So the old-school stats appear to capture somewhat the performance of Pierce and Garnett, but don't reflect an apparent improvement in Allen's shooting over the prior year. (And Allen's shot has been shaky enough that it's been regularly commented on the past few weeks, w/the assumption that he'll snap out of it.)

Posted by Quarterican | November 30, 2007 3:22 PM

I dunno, it looks like a statistically insignificant change, at least for Pierce and Allen.

Posted by Mr. Noah | November 30, 2007 4:12 PM

I'm glad to see that you're enjoying beautiful Amsterdam and are not distracted by ... by ...

Never mind.

Posted by Martin | November 30, 2007 4:20 PM

I would guess, perhaps counter-intuitively but accurately accourding to these stats, that the new "Big 3" scenario benefits KG much more than the other 2.

Pierce has always been able to score on his own when healthy, even on bad teams. Remember, before he got hurt last year the C's weren't the atrocious disgrace that ended the season, they were just bad.

Allen (who, btw, going into last night's game was 3rd in the league in minutes), hasn't been better than his usual self at all, despite the 2 amazing clutch 3's.

KG, however, has been a monster on the post, and he usually isn't doubled, due to Allen and Pierce, one would have to assume.

KG should be on the post a lot more, something Doc will figure out by March, hopefully (right around the time he figures out playing Allen more minutes than LeBron James is probably not a great idea).

Posted by Andruw | November 30, 2007 4:24 PM

KG, however, has been a monster on the post,

Life is better in the East. He goes against Shaq and Haslem again tonight, though. They held him to 26 and 11 and almost beat Boston two weeks ago.

Posted by Just Karl | November 30, 2007 4:50 PM

matt, love the graphs you have been using recently. just wanted to know what you use to make them.

Thanks,

Joe C

Posted by joe c | November 30, 2007 4:51 PM

matt, love the graphs you have been using recently. just wanted to know what you use to make them.

Thanks,

Joe C

Posted by joe c | November 30, 2007 4:51 PM

I don't blame someone for being excited about adding Garnett to their team, but how many playoff series has he won again?

Posted by Notorious P.A.T. | November 30, 2007 4:54 PM

this dovetails nicely with Ken Rosenbaum's main criticism of WoW...

Which is what? That darn bridge by the Alliance base in Alterac Valley?

Posted by Steve | November 30, 2007 4:57 PM

I think its kind of like a truism that 'good' teammates will increase your true shooting percentage. Whether these guys are helping each other play better is hard to say from the available data. But the Celts are winning games.

I would expect these kinds of effects to vary a lot by situation. If you are a post player who starts seeing less attention with the addition of better perimeter guys, you would hope to see a big increase in shooting percentage. On the other hand, if you are a jump shooter who can't get his own shot, your percentages might not change so much as your opportunity.

Never-the-less, this is probably an interesting thing to look at when big name players come together, in combination with the change in the number of shots they're taking. It would give you an idea of how well they complement each other (theoretically).

Posted by mpowell | November 30, 2007 4:58 PM

Wouldn't "offensive stool" be better used to describe the Knicks?

Posted by Jason | November 30, 2007 5:21 PM

Kit Stolz said: Jeez, they're all shooting 58% or better? Holy cow. In a league where a team is doing well to shoot 50%, this is big news. I had no idea. How often has this happened before in the NBA?

If I remember correctly, at one time during the Lakers' Showtime heyday, Magic Johnson, Byron Scott, Kurt Rambis, James Worthy, and Kareem were all shooting over 50%. If not the whole starting five, I think at least Scott, Rambis, and Kareem were over 50%.

Posted by shnooky | November 30, 2007 5:55 PM

This is a clever and informative slicing of the data from Matthew.

-----

In a certain sense, this effect points out the limitations of sabremetrics in hoops as follows:

Many of us would have suspected that the KG/Truth/Ray-Ray combination would have had this precise effect. But pretty much no data coming into this season would have allowed us to predict the effect with numbers.

It's only an understanding of how things work on the court that could let us predict this. You couldn't have predicted it with stats.

Posted by Petey | November 30, 2007 7:51 PM

Actually the graph is pretty lousy. The horizontal axis is quantitatively meaningless, conveying no information about the independent variable ("quality of teammates") and the scaling of the vertical scale is downright wretched. Seriously, who the hell draws tickmarks with increments of 2.75? Last, it's preferable to label the data lines themselves, rather than force the reader to refer to a legend.

Paging Dr Tufte. Dr Tufte, please answer an emergency page from Amsterdam...

Posted by Matt B | November 30, 2007 8:14 PM

Matt B is unduly harsh, I think. The cosmetic comments (tickmarks of 2.75; use of a legend) are irrelevant. The comment about the horizontal exis is well taken. However, I think most knowledgeable basketball fans will recognize that, for most of the years represented on the chart prior to the current year, the quality of each of the players' teammates was, at least to some extent, worse than this year (and last year significantly worse in all three cases). Accordingly, we should at least see some qualitative idea of the significance of the evidence in support of the proposition Matthew is discussing.

Interestingly, though, in the case of KG - the year that KG had his best teammates prior to this year - 2004 - he also had his worst TS% (or tied for worst with last year). So does this change our opinion???

Posted by Al | November 30, 2007 9:35 PM

Yeah, the legend thing was petty. But I stand by the scaling comment. There's no reason not to use whole numbers. And the limits of 51 and 62% are plain silly. Goes to show that even Apple (Joe C, this is Numbers) can make bad graphics if stick with the defaults.

Posted by Matt B | November 30, 2007 10:03 PM

Much too small a sample (one case) to make any kind of reasonable argument. This may have numbers in it, but it's an anecdote, not any kind of statistical demonstration.

Exactly. Voros's Law: Any major leaguer of any ability can hit for any average in 70 at bats. Applies equally well to basketball.

In other words, even Ichiro can go 5/70, and even Mario Mendoza can go 40/70 on a limited sample. Doesn't really tell us anything.

Posted by Seitz | November 30, 2007 10:20 PM

this dovetails nicely with Ken Rosenbaum's main criticism of WoW...

Which is what? That darn bridge by the Alliance base in Alterac Valley?

Sorry, Dan Rosenbaum...

Posted by Pooh | December 1, 2007 1:30 AM

Matt, all this graph shows is that the shooting percentage of RA, KG and PP is pretty much the same as it has been in the past. The difference in the Celtics now is that they are ALL ON THE SAME TEAM.

Shooting efficiency is a function of # of shots taken, closeness to the basket (easier shots v. longer distance) and baskets made.

Aside from the fact that the Celtics have three starters shooting over 50%, the Celtics are winning with these wonderful players because the opposing team cannot double-team any one or two of them without getting burned on the defensive end by the player left open.

A more interesting graph would also show the number of shots taken v. shots taken when they were on their old teams. Iverson doesn't take as many shots as before, his points are down v. other years, but because he has Carmelo on his team, his assist totals are way up. And his team is winning more games than the Sixers did when he was the sole offensive weapon. The Nuggets still will not win as many games as easily as the Celtics because they do not have KG (or a young shaq, etc).

Besides, Paul Pierce is a much better defender than Carmelo (Carmelo has only had, I think, about 3 or 4 double-doubles in his career.) Iverson has always been an indiferent defender although he is one of the top thiefs (steals) in NBA history.

If you were to interview Garnett, Allen, Pierce, Iverson, Anthony, they would all say that the game is easier to them. If you then interviewed Kobe, he would tell you that he is exausted!

Posted by Ellen hamm | December 1, 2007 3:45 AM

offensive stool?

Posted by james root | December 1, 2007 9:36 AM

He goes against Shaq and Haslem again tonight, though. They held him to 26 and 11 and almost beat Boston two weeks ago.
I don't think that KG scoring 26 means he's being 'held' to anything. Looks more like he had his way with the Heat back then.

Posted by kozlodoev | December 2, 2007 12:57 AM