Assisted on Data [Archive]

pfrduke

05-15-2012, 12:46 AM

We've talked a lot on this site about the passing performance (or lack thereof) from this year's Duke team. All told, it was one of the least assist-heavy squad's in K's tenure, with only 48.4% of all field goals coming off of the pass. This analysis breaks down the passing a little bit to see how our guys scored the ball last year. Pomeroy took a look at this (http://kenpom.com/blog/index.php/weblog/a_look_at_assisted-on_data/) with respect to the final four squads earlier, so that's an interesting point of comparison to these numbers.

Numbers are below, but here are some general thoughts:

Austin almost never scored off the ball. Not a single two-point jump shot was assisted, and in total only 12 of his 116 shots inside the arc came off of the pass. I suspect (and will try to break this information down further) that a substantial portion of those 12 came off fast breaks, meaning that in the half court offense, he wasn't scoring inside the arc unless he could create it for himself. Even beyond the arc, he had by far the lowest percentage of assisted baskets, with over 60% of his 3s coming off the dribble. It's remarkable that he shot as well as he did (36.5%) given that he created more 3s for himself than were created for him.
Three pointers come off the pass. Ryan Kelly, Tyler Thornton, and Andre Dawkins combined for 134 3s. 128 of those were assisted. As a team, 78% of our 3s were assisted. Other than Austin Rivers, 7 out of every 8 3s came off of a pass. It's impossible to tell from play by play data which attempts are off the pass as opposed to self-created - the only way I can tell for makes is because assists are recorded - but odds are if you see a guy attempt a 3 off the dribble, he's not going to make it. Aside from Austin, we made 25 in 34 games.
Two point jump shots are essentially never assisted. This is especially notable if you exclude Miles and Hairston, for whom jump shots happened almost exclusively when they were found open on ball rotation (note that Miles's numbers are further skewed because everything that's not a dunk or layup is considered a jump shot, so his jumper numbers include shots from the paint off of post moves that weren't quite layups - if you're talking 12-15 footers, the assist numbers probably skew even higher). They combined for 20 assisted baskets on 31 jumpers - the rest of the team got 33 assisted baskets on 162 jumpers, about 1 in 5. And, as mentioned, Austin did all of these for himself, with not a single two point jumper off the pass. In a way, this makes sense - the offense should not be designed to result in two-point jump shots, because they're very low percentage (and thus low efficiency), so guys are more likely to take those shots when the offense breaks down and the only option is to create for himself.
Austin got into the paint so well. I know this is obvious by sight, but getting almost 3 layups/dunks a game, the vast majority of which were the result of his own penetration (almost always starting from beyond the 3-point line) is impressive for a guard. Curry was the next best perimeter player, and he had just 36.
Dawkins simply does not create for himself. Only 17 of his 92 field goals came off the bounce (and I suspect at least a few of those are fast break buckets). His 81.5% assisted on rate is the highest on the team.
The next step in this is to factor in the buckets off of offensive rebounds. The numbers for Mason, Miles, Ryan, and Josh are all skewed a little downward because they don't factor in put-back buckets. So all of the un-assisted baskets aren't really self-created in the sense that they got the ball in a half-court set and made something for themselves. Mason, in particular, needs improvement on this. He had just 48 unassisted baskets, a decent number of which must be put-backs. He needs to do better than one self-created hoop per game.

Here are the numbers:

Layups/Dunks

Two Point Jumpers

Three Pointers

Made

Assisted

Made

Assisted

Made

Assisted

Rivers

84

12

14.29%

32

0.00%

58

23

39.66%

174

35

20.11%

Curry

36

4

11.11%

41

8

19.51%

64

49

76.56%

141

61

43.26%

Mason

112

64

57.14%

31

7

22.58%

143

71

49.65%

Dawkins

10

5

50.00%

15

7

46.67%

67

63

94.03%

92

75

81.52%

Kelly

43

18

41.86%

24

8

33.33%

40

39

97.50%

107

65

60.75%

Miles

67

34

50.75%

22

12

54.55%

89

46

51.69%

Thornton

8

0.00%

4

2

50.00%

27

26

96.30%

39

28

71.79%

Hairston

22

15

68.18%

9

8

88.89%

31

23

74.19%

Cook

18

2

11.11%

15

1

6.67%

14

11

78.57%

47

14

29.79%

Gbinije

7

2

28.57%

4

3

75.00%

11

5

45.45%

407

156

193

53

274

214

874

423

48.40%

38.33%

27.46%

78.10%

Kedsy

05-15-2012, 10:46 AM

m g

05-15-2012, 11:02 AM

Year %asst
---- -------
2012 48.4%
2011 50.4%
2010 52.3%
2009 50.7%
2008 51.6%
2007 52.7%
2006 55.8%
2005 49.4%
2004 52.7%
2003 49.5%
2002 57.2%
2001 57.6%
2000 55.9%
1999 50.9%
1998 47.5%
1997 52.7%
1996 52.9%
1995 53.7%
1994 55.7%
1993 53.5%

Interesting that the three top assist % years in this set were the Jason Williams years.

pfrduke

05-15-2012, 11:21 AM

The next step in this is to factor in the buckets off of offensive rebounds. The numbers for Mason, Miles, Ryan, and Josh are all skewed a little downward because they don't factor in put-back buckets. So all of the un-assisted baskets aren't really self-created in the sense that they got the ball in a half-court set and made something for themselves. Mason, in particular, needs improvement on this. He had just 48 unassisted baskets, a decent number of which must be put-backs. He needs to do better than one self-created hoop per game.

I'm just realizing that I got the comment on Mason slightly wrong. He had just 48 unassisted dunks & layups, but when you factor in other 2s (including things like his hook shots), that number goes up to 72. Again, it certainly comes down a bit with put backs, but getting closer to two buckets of his own making per game is definitely better than one. If that number could go up closer to 3 per game (Sullinger was in that neck of the woods, with 109 unassisted twos), it would show good development in Mason's low post offense and ability to create for himself. Note that the ~6 points per game would be under-representative, since it would not include trips to the line that come from unconverted post moves.

Kedsy

05-15-2012, 11:22 AM

Interesting that the three top assist % years in this set were the Jason Williams years.

Bobby Hurley's four years were 56.6%, 56.1%, 58.9%, and 53.5%. Clearly we didn't have a Bobby Hurley or Jason Williams on the 2012 team. But not so many teams do.

Reilly

05-15-2012, 12:10 PM

... So maybe our team passing wasn't as awful as many people lamented. It was just that our scoring and shot leader strongly preferred to get his points off the bounce.

... and that may be why others lamented our scoring and shot leader's preference: it wasn't what folks may have been used to, or it wasn't the style folks prefer seeing, or both.

Kedsy

05-15-2012, 03:29 PM

... and that may be why others lamented our scoring and shot leader's preference: it wasn't what folks may have been used to, or it wasn't the style folks prefer seeing, or both.

Yeah, but you get what you get. Everybody wanted him. He was the #2 recruit in the country and very talented. Talented enough that Coach K allowed the offense to be structured in such a way so Austin could play the style he preferred.

UrinalCake

05-15-2012, 03:53 PM

So Austin rarely scored off an assist. That makes sense, and he'll definitely have to work on moving without the ball and being able to receive the ball and score as he transitions to the next level. So my question is, if the rest of the team clocks in at a 52.5% scoring-off-of-an-assist rate, then who is the one passing the ball to them in order to generate these assists? According to ESPN's numbers our assists were pretty evenly distributed:

Curry 83
Rivers 71
Thornton 69
Cook 63
Mason 55
Kelly 35
Dawkins 20
Miles 18

So I agree with Kedsy that our team did in fact share the ball pretty well with the glaring exception of Rivers and how he generated scoring.

Other observations from the assist totals shown above:
- During the season I heard a lot of people saying that Tyler doesn't create for others, but he put up a similar number of assists as Austin and Seth despite playing FAR fewer minutes. If you look at assists per minute you'd get Seth 0.081, Austin 0.063, Tyler 0.096, and Quinn 0.16. This also makes a pretty strong case for Quinn being our starting point guard next year if he can maintain this and improve his defense.
- I've always thought of Ryan as being an excellent passer and actually better than our guards at feeding the post. Yet his assist numbers are pretty paltry.
- Mason put up quite a few assists himself. His high school coach (who provided much drama this spring) commented on Mason's ability and willingness to pass out of the post. Hopefully we can take advantage of this next year as he should be the focal point of our offense.

CDu

05-15-2012, 04:00 PM

So, yes, our assisted-on rate is the 2nd lowest in 20 years. But if you take Austin out of the equation, the assisted-on rate on everyone else's buckets was 388 on 700 baskets, or 55.4%. And other than 2001 and 2002, that number (55.4%) is within half a percent of our highest rate over the 20 year span.

Of course, just taking Austin out isn't a fair comparison, because every team has a playmaker, but suppose we pushed Austin's assisted-on rate up to 40% (which would still be the lowest on the team, other than Quinn's). In that case, our team assisted-on rate would have been 52.4%, squarely in the middle on the above chart, higher than the 2010 and 1999 teams, and essentially even with the 2004 team.

So maybe our team passing wasn't as awful as many people lamented. It was just that our scoring and shot leader strongly preferred to get his points off the bounce.

I think one could still very reasonably argue that our passing was awful - just perhaps not in the exact same manner. As you said, simply taking Rivers out isn't appropriate. The fact that he was such a primary ballhandler in the half court and neither created nor received many assists, he is a drain on these stats. But he also doesn't address another issue, which is that the majority of our assists came on 3 pointers (a LOT of which I suspect were of the "pass the ball around the perimeter" type rather than the "draw and dish" variety).

Only 38% of our 2pt buckets inside were assisted. Under 30% of our 2pt jumpers were assisted. In spite of the fact that over 2/3 of our shots were 2pt attempts, less than half of our assists were on 2pt attempts. So it doesn't seem like our passing was doing that great a job of creating easy scoring chances. We were just a really good perimeter shooting team, had a guy who could get to the rim, and had another guy who could score some with his back to the basket. Not a great passing team, not a great dribbling team, and arguably our best dribbler was one of our worst passers/decision makers.

The fact that we remained one of the most efficient scoring teams in the country for most of the season (really right up until the last few games) is a testament to our team's ability to offensive rebound and hit 3s.

Kedsy

05-15-2012, 04:12 PM

I think one could still very reasonably argue that our passing was awful - just perhaps not in the exact same manner. As you said, simply taking Rivers out isn't appropriate. The fact that he was such a primary ballhandler in the half court and neither created nor received many assists, he is a drain on these stats. But he also doesn't address another issue, which is that the majority of our assists came on 3 pointers (a LOT of which I suspect were of the "pass the ball around the perimeter" type rather than the "draw and dish" variety).

Only 38% of our 2pt buckets inside were assisted. Under 30% of our 2pt jumpers were assisted. In spite of the fact that over 2/3 of our shots were 2pt attempts, less than half of our assists were on 2pt attempts. So it doesn't seem like our passing was doing that great a job of creating easy scoring chances. We were just a really good perimeter shooting team, had a guy who could get to the rim, and had another guy who could score some with his back to the basket. Not a great passing team, not a great dribbling team, and arguably our best dribbler was one of our worst passers/decision makers.

The fact that we remained one of the most efficient scoring teams in the country for most of the season (really right up until the last few games) is a testament to our team's ability to offensive rebound and hit 3s.

I think this is a reasonable explanation, with one exception. You don't state it but you imply that a "pass the ball around the perimeter" assist is not good passing. To the extent you believe that, I disagree. I think swinging it around the perimeter until you find a wide open shooter is very good passing, especially when the open shot is by a good shooter. Getting an assist without penetration can arguably be a better pass than the "draw and dish," because a draw-and-dish assist is at least as much due to good dribbling skills than it is to passing skills.

pfrduke

05-15-2012, 04:53 PM

One thing I'm hoping to put together over the next few days is a look at who assisted whom, and for what kinds of shots. Were our bigs hitting outside shooters more often than inside scorers (suggesting either offensive rebound kickouts or effective inside-outside work)? Or were our guards the more frequent perimeter distributors (for Austin, this could be the result of drive/dish; for everyone else, not as likely)?

I'm also interested in seeing when our makes and assists came with respect to the shot clock - I suspect (not surprisingly) that in the last 10 seconds of the clock we had a lot lower assist rate.

CDu

05-16-2012, 08:49 AM

I think this is a reasonable explanation, with one exception. You don't state it but you imply that a "pass the ball around the perimeter" assist is not good passing. To the extent you believe that, I disagree. I think swinging it around the perimeter until you find a wide open shooter is very good passing, especially when the open shot is by a good shooter. Getting an assist without penetration can arguably be a better pass than the "draw and dish," because a draw-and-dish assist is at least as much due to good dribbling skills than it is to passing skills.

I don't think we're talking about the same thing with regard to "passing around the perimeter." I completely agree that the quick passes around the perimeter (usually following penetration or a post up which draws the defense slightly out of position and into chase mode) is good passing. This happened a lot in the Duhon/Williams/Dunleavy/Battier era. And it was great basketball. So in that sense, I completely agree with what you said above.

But I don't think that's what happened last year. We very rarely had the quick passing game, as so many of our players would catch and look to shoot themselves. Instead it was much more frequently just bad defense/good shooting that resulted in "empty" assists (I use that term loosely to define a pass that did little to actually set up the shot but still resulted in an assist).

FellowTraveler

05-16-2012, 10:47 AM

I don't think we're talking about the same thing with regard to "passing around the perimeter." I completely agree that the quick passes around the perimeter (usually following penetration or a post up which draws the defense slightly out of position and into chase mode) is good passing. This happened a lot in the Duhon/Williams/Dunleavy/Battier era. And it was great basketball. So in that sense, I completely agree with what you said above.

But I don't think that's what happened last year. We very rarely had the quick passing game, as so many of our players would catch and look to shoot themselves. Instead it was much more frequently just bad defense/good shooting that resulted in "empty" assists (I use that term loosely to define a pass that did little to actually set up the shot but still resulted in an assist).

I agree with CDu. I'd have liked to have seen more and quicker passing (and less dribbling/screening) around the perimeter. I've been meaning to write a longer post about this for a long time, but haven't been able to bring myself to do it.

Here's the short version: Duke had a bunch of good shooters who weren't extraordinary at breaking defenses down off the dribble. How do you get open looks for such shooters? Well, you could play outside-in, but that wasn't happening. You can swing the ball around the perimeter, with quick, crisp passes, good spacing, and lots of ball reversal. Defenders can't move as quickly as a crisp pass, and reversing the direction of the ball can be particularly valuable. But Duke didn't do as much of this as I'd have liked. Instead, Rivers and Curry (and to a lesser extent Thornton and Cook) dribbled around the perimeter, probing the defense. When they passed the ball, they tended to do so after dribbling towards the recipient, shortening the distance of the pass and making it easier for the defense to keep up. Passes were neither decisive nor crisp. (One thing that was striking watching teams like Kansas and Kentucky deep in the NCAA tournament is how different their perimeter passes looked in comparison to Duke's: Crisp, hard, fast passes.) Bigs came out to set a lot of screens. Screens can be great, but when 1) The ball handler isn't a great penetrator and 2) Everybody knows the big isn't going to get the ball on the roll, setting a screen merely brings an extra defender out to the perimeter, further cluttering up an already poorly-spaced perimeter.

CDu

05-16-2012, 11:42 AM

Here's the short version: Duke had a bunch of good shooters who weren't extraordinary at breaking defenses down off the dribble. How do you get open looks for such shooters? Well, you could play outside-in, but that wasn't happening. You can swing the ball around the perimeter, with quick, crisp passes, good spacing, and lots of ball reversal. Defenders can't move as quickly as a crisp pass, and reversing the direction of the ball can be particularly valuable. But Duke didn't do as much of this as I'd have liked.

I completely agree (obviously, since your post was in agreement with mine) that the team didn't do much of that. The problem is that, for the quick passing game to work, you have to draw the defense out of position to begin with. If the opponent is playing man to man (or even zone in some cases), simply passing the ball quickly around the perimeter doesn't do anything. The defender is already guarding his man.

You have to get the defense in a rotating situation for the quick passing game to be effective. And that's done by drawing help defense. The ways to draw help defense are:
- screens (either on-ball or away from the ball)
- playing inside-out, drawing help defense on the post man
- having a playmaker beat his man off the dribble and draw defenders to him

We employed all three to some degree (mostly the 1st and 3rd option). But our playmakers typically didn't pass when they drove, so the quick passing game wasn't utilized. And when the ball was returned to the perimeter, nobody was looking to quickly swing the ball again (either an open shot went up, which is good; or there was hesitation, which is bad).

Jderf

05-16-2012, 12:04 PM

I agree with CDu. I'd have liked to have seen more and quicker passing (and less dribbling/screening) around the perimeter. I've been meaning to write a longer post about this for a long time, but haven't been able to bring myself to do it.

Here's the short version: Duke had a bunch of good shooters who weren't extraordinary at breaking defenses down off the dribble. How do you get open looks for such shooters? Well, you could play outside-in, but that wasn't happening. You can swing the ball around the perimeter, with quick, crisp passes, good spacing, and lots of ball reversal. Defenders can't move as quickly as a crisp pass, and reversing the direction of the ball can be particularly valuable. But Duke didn't do as much of this as I'd have liked. Instead, Rivers and Curry (and to a lesser extent Thornton and Cook) dribbled around the perimeter, probing the defense. When they passed the ball, they tended to do so after dribbling towards the recipient, shortening the distance of the pass and making it easier for the defense to keep up. Passes were neither decisive nor crisp. (One thing that was striking watching teams like Kansas and Kentucky deep in the NCAA tournament is how different their perimeter passes looked in comparison to Duke's: Crisp, hard, fast passes.) Bigs came out to set a lot of screens. Screens can be great, but when 1) The ball handler isn't a great penetrator and 2) Everybody knows the big isn't going to get the ball on the roll, setting a screen merely brings an extra defender out to the perimeter, further cluttering up an already poorly-spaced perimeter.

To your question (which I bolded), I would say that in the past Duke has done this exactly the way you laid it out: motion, passing, and spacing. As you continue to point out, however, this Duke team did not do this as well or as consitently as past teams. I think the counter-intuitive reason for this was (as has been noted in the thread) the success of Austin Rivers.

Austin was simply so good at one-on-one penetration that, rather than focusing on creating good team spacing, the team focussed more often on creating good space for Austin. Similarly, rather than having good ball movement and swinging the ball around the perimeter until the defense is thrown off-balance, we would swing the ball around the perimeter until it got to Austin -- who would then isolate his man and see if he had a driving opportunity before he would either drive or, more rarely, continue the chain of distribution. Whether or not Austin drove or passed, however, his pause for evaluation would interrupt our perimeter motion/passing/spacing, allowing defenses a moment to adjust, which somewhat undermined Duke's traditional offense.

Now, whether or not this is even a bad thing, I'm not so sure. Austin sure had a knack for getting that ball to the cup, and in my opnion, I consider his presence as a net gain for last year's team. However, I believe his "disruptive effect" on our motion offense was what many people were keying in on throughout the season when they worried that, despite our impressive efficiency, our offense was "not fluid at all," "had no rhythm," etc etc.

FellowTraveler

05-16-2012, 04:34 PM

To your question (which I bolded), I would say that in the past Duke has done this exactly the way you laid it out: motion, passing, and spacing. As you continue to point out, however, this Duke team did not do this as well or as consitently as past teams. I think the counter-intuitive reason for this was (as has been noted in the thread) the success of Austin Rivers.

I think that's a reasonable, and possibly correct conclusion, but I don't think I share it.

Austin was simply so good at one-on-one penetration that, rather than focusing on creating good team spacing, the team focussed more often on creating good space for Austin.

Austin was very good at one-on-one penetration. In my opinion, however, he was not good enough at it to justify such an approach, if indeed it was a conscious decision. That's not a knock on Rivers: The guy was a freshman, after all, and one of the best in the country. I was not at all disappointed in his play. But I always found talk of him being a truly elite offensive threat overblown. He wasn't. (Yet.)

Similarly, rather than having good ball movement and swinging the ball around the perimeter until the defense is thrown off-balance, we would swing the ball around the perimeter until it got to Austin -- who would then isolate his man and see if he had a driving opportunity before he would either drive or, more rarely, continue the chain of distribution.

And here's why I don't agree that Austin's individual brilliance was the primary cause of we both see as suboptimal ball movement: I don't think the team really behaved as you describe. When the ball got to Seth, he'd do the same thing Austin did -- pause to assess, probe the defense with the dribble, etc. Tyler did a bit of that, too, as did Cook. And guards, including Austin, often received the ball while passer and recipient were moving towards each other -- ball-handler dribbling towards the player he's about to pass to, recipient running to meet the pass. This isn't a good approach to isolating your man.

I think confidence in/reliance upon Austin's individual abilities probably contributed to the suboptimal ball movement we both perceived. But I think a bigger factor was that there didn't seem to be a clear offensive strategy, and there certainly wasn't consistent, competent execution. (We frequently saw two guards within 10 feet of each other on the perimeter, both without the ball, clearly not understanding or agreeing which of them should be where. We saw more passes delivered to 20 feet of empty space, as passer and recipient weren't on the same page, than I've ever seen from a Duke team. Neither of those things are indicative of a plan to get out of the way and let Austin isolate; rather, they suggest a less than clear strategic vision and poor execution.)

Whether or not Austin drove or passed, however, his pause for evaluation would interrupt our perimeter motion/passing/spacing, allowing defenses a moment to adjust, which somewhat undermined Duke's traditional offense.

Totally agree with this, though I again wouldn't single Austin out -- none of the perimeter players (or post players, for that matter) were particularly instinctive/decisive.

Now, whether or not this is even a bad thing, I'm not so sure.

Nor am I. Always important to keep perspective: The offense was the nation's 11th most efficient, after all, so they weren't exactly the Washington Generals out there. I don't think it was as efficient as it could've been (some perspective on that perspective: The offense was the nation's 11th most efficient, but it was also Duke's least efficient since 2007.) And I certainly don't think it was as consistent or as versatile as it could've been. But it was good enough that I can't be certain of either of those conclusions.

Austin sure had a knack for getting that ball to the cup

He did -- but, to me, 15 ppg on 53% true shooting and 104 offensive rating isn't enough of a knack to rely on if you can help it. Again: That's not a shot at Rivers, but he wasn't exactly senior year Redick -- or even senior year Nolan Smith.

I consider his presence as a net gain for last year's team.

As do I.

However, I believe his "disruptive effect" on our motion offense was what many people were keying in on throughout the season when they worried that, despite our impressive efficiency, our offense was "not fluid at all," "had no rhythm," etc etc.

To sum up: I agree it played a role, but I think that role was smaller than you do, and that other factors were quite significant.

throatybeard

05-16-2012, 06:57 PM

A couple of things about the table interest me. One, one of the worst numbers is 2003, with jives with my foggy memory, because I seem to remember a lot of running isos for Dahntay Jones. The other is 1998, also low. Which is surprising given that Wojo, who is like sixth or so in all time assists at Duke, was still on the team.