PDA

View Full Version : Our Defense vs. UNC - second game - including cumulative numbers



tommy
03-06-2012, 11:36 PM
Well, it is not exactly breaking news that we struggled mightily on defense against the Tar Heels on Saturday night. As a number of posters in the post-game thread talked about Andre Dawkins and his lack of PT and whether that was related to his defense, I'm going to put some more Andre-specific information in another thread.

When I went back and charted this game, I realized immediately something I had not been aware of when watching it live: UNC scored on its first nine -- nine! -- possessions of the game. Obviously they got off to a great start against us, but I didn't realize they were 9 for 9 (possessions, not 9 for 9 shooting) at the first TV timeout. And their buckets came in a variety of ways. Zeller on individual post moves; jumpers by Barnes and Zeller and Marshall and Bullock -- some due to defensive breakdowns, some just jumpers over our defenders who were in decent position; layups. They had 9 successful trips before we got our first defensive rebound. Wow.

We then righted the ship somewhat, but the damage was done, as our getting torched at the defensive end obviously caused us to tighten up offensively as we missed opportunity after opportunity and the margin just grew and grew. There was another stretch late in the first half where UNC scored on 5 of 6 possessions, and they scored on their last four possessions before halftime. Then in the second half, from just before the 16 minute timeout almost until the 12 minute TO, they had a stretch where they scored on 8 of 10 possessions. And again, at the end of the game they scored on their last 6 possessions. Obviously, you just can't have that many long stretches where you're not getting stops.

There was plenty of blame to go around. As you'll soon be able to see in the full chart over at dukehoopblog.com, our team's overall defensive rating was north of 106, which is pretty high, even for us. The guy who struggled the most, when these numbers are considered, was Miles Plumlee. He was very active, to be sure, and came with great energy to his last appearance in Cameron, but he just got beaten all night. He gave up eight baskets and was partially responsible for a ninth -- by far the most in any game this year. He was beaten one-on-one in the post by Zeller; by Henson on step-out jumpers, and by his own failure to block out, as I detailed in a separate thread.

Again, I'll discuss Andre in another thread, but his DRating was very high as well - 109.6, and Quinn Cook's was worse, at 111.6. Quinn just couldn't stop penetration, a continuing theme for him this year, as in only 13 minutes I charted him getting beaten off the dribble three times. And he gave up three hoops in that short time on the floor.

The best defensive numbers belonged to Seth Curry and Ryan Kelly. Mason's numbers were poor, so Ryan's numbers really stood out as being the best of the bigs. He only gave up 2.5 hoops and forced 2 turnovers in 28 minutes. Austin Rivers' numbers were not far off, and he was involved in a ton of plays. Not only was he active, but UNC appeared to be targeting him too, in an attempt to wear him down. Thornton's numbers were just slightly north of Seth's, and I give Ty a lot of credit here, as he had to guard the much bigger and stronger Barnes for much of his time on the floor.

In terms of stop %, the numbers are not pretty, as one would expect. Here they are:




Stops
Hoops
Stop%


Curry
23
33
41%


Rivers
25
36
41%


Dawkins
7
15
32%


Mason
12
27
31%


Thornton
16
23
41%


Kelly
20
27
43%


Miles
20
24
45%


Cook
7
13
35%


Gbinije
0
1
0%


Hairston
0
2
0%


TEAM
26
40
39.4%


























Pretty bad, obviously. I can't believe that we fared as badly as we did with Mason on the floor, but we did. UNC scored on 69% of its possessions when Mason was out there. Wow.

I do find interesting that, despite his worst-on-the-team DRating, the team actually did best in terms of getting stops when Miles was on the floor. Little hard to interpret that. I guess it could be that when Miles' man wasn't directly involved in the play, he was providing good help defense and communication -- and I did see some of that -- which helped us get stops.

I think this is going to be the last game that I chart for this year. The postseason games just come too fast and furious and I won't be able to keep up with them, and plus, I feel like we've gained the information we're going to gain from it. I did want to post a cumulative stop% chart here though for those of you interested in it. And over on dukehoopblog, I will get around to posting a cumulative Defensive Rating chart there hopefully soon. That'll be interesting to see, I think. This stop% charting includes all games since the Western Michigan game, with the exception of Penn, NC State (no TV access for me) and the home OT game vs. Va Tech (I just never got around to doing that one and don't feel like it now!). Anyway, the cumulative numbers for almost all the games starting with Western Michigan on December 30 are as follows:





Stops
Hoops
Stop%


Curry
439
398
52.4%


Rivers
522
460
53.2%


Dawkins
308
300
50.7%


Mason
387
382
50.3%


Thornton
303
274
52.5%


Kelly
402
353
53.2%


Miles
314
281
52.8%


Cook
213
183
53.8%


Gbinije
38
51
42.7%


Hairston
95
80
54.3%


TEAM
606
550
52.4%

davekay1971
03-07-2012, 07:57 AM
Absolutely phenomenal work, as always. It's been a great addition to the board this season. Thank you for all the effort you've put in.

(Apparently I need to spread the love some more, but in my heart there's pitchfork points for you!)

devildeac
03-07-2012, 08:53 AM
Absolutely phenomenal work, as always. It's been a great addition to the board this season. Thank you for all the effort you've put in.

(Apparently I need to spread the love some more, but in my heart there's pitchfork points for you!)

I covered him for you;).

Thanks again, tommy for your excellent work this season.

Steven43
03-07-2012, 09:22 AM
Tommy, if your numbers are accurate it doesn't seem to bode well for our chances in either the ACC T or the NCAA T. I don't see how individual players can suddenly become capable defenders if they have significant flaws in their defensive abilities this late in the season. You basically skewered Andre, Miles, Mason and Quinn. And I had thought Kelly and Curry were perhaps our two weakest defenders who played significant minutes! Of course, we are just talking about one game against a very strong opponent, but one cannot deny that the stats you provided are sobering. That being said, I'm going to support the team as strongly as ever and hope for the best.

jv001
03-07-2012, 10:22 AM
Thanks tommy for your hard work on those defensive charts. I look forward to better numbers next year. GoDuke!

tommy
03-07-2012, 11:57 AM
Thanks, fellas, for the kind words (and the pitchforks!). Nice to be acknowledged.

CDu
03-07-2012, 12:14 PM
Add me to the list complimenting your efforts on this, Tommy. Very much appreciated as I'm sure it took quite a bit of work (at least it's a labor of love).

One thing that struck me is that the stop% looks a bit too balanced in general for the guards compared to the bigs. I feel like the typical FG% for guards (40-45% is pretty common) is lower than for bigs (50+% is common, below 45% is pretty bad). As such, I'd expect the stop % to be much better for the guards than for the bigs, as for the most part I'd expect the guards to be more responsible than the bigs for opposing guard production (obviously it's not a perfect correlation as screens, switches, and double-team help defense do play in, but I'd say for the most part it should hold). The fact that they are pretty close leads me to believe that the guards aren't doing a terribly good job defensively. They appear to be allowing roughly a 48% scoring rate while the bigs are also allowing roughly a 48% scoring rate. Tommy, correct me if I'm misinterpreting your data. It could very well be that I'm missing a key piece of this.

tommy
03-07-2012, 02:27 PM
Add me to the list complimenting your efforts on this, Tommy. Very much appreciated as I'm sure it took quite a bit of work (at least it's a labor of love).

One thing that struck me is that the stop% looks a bit too balanced in general for the guards compared to the bigs. I feel like the typical FG% for guards (40-45% is pretty common) is lower than for bigs (50+% is common, below 45% is pretty bad). As such, I'd expect the stop % to be much better for the guards than for the bigs, as for the most part I'd expect the guards to be more responsible than the bigs for opposing guard production (obviously it's not a perfect correlation as screens, switches, and double-team help defense do play in, but I'd say for the most part it should hold). The fact that they are pretty close leads me to believe that the guards aren't doing a terribly good job defensively. They appear to be allowing roughly a 48% scoring rate while the bigs are also allowing roughly a 48% scoring rate. Tommy, correct me if I'm misinterpreting your data. It could very well be that I'm missing a key piece of this.

CDu - you are misinterpreting what this data is showing. I started tracking what I call "stop %" here in response to many posters' beliefs earlier in the year that Tyler Thornton, while he was being beaten pretty regularly off the dribble, nevertheless provided significant defensive "intangibles" that made our defense stronger, and that was why he was getting PT. I wanted to see if that was measurable in any way. So all my stop % is showing is the following: when a given player is on the floor, how often is the TEAM getting a stop and how often is the TEAM giving up a hoop. All five guys who are on the floor get "credit" for the stop and all five get "blame" when the opponent scores. That's probably why, when accumulated, the perimeter guys and the bigs come out pretty close to each other as a group.

The stats that you're looking for -- where each player's actual contributions to forcing missed shots, forcing turnovers, allowing baskets himself, that sort of thing -- that I'm doing over at dukehoopblog.com. Those are more individualized than what the stop% charts here at DBR have been. Over at dukehoopblog you can see an Excel on each game that I've charted, with each individual player's ratings on a number of metrics. I also did a cumulative for all games through the first UNC game and that is available there too. All those charts are available under the "Duke Hoops" tab and then I think it's under "Defensive Charting."

I haven't gotten the second UNC game up there yet - hopefully today or tomorrow - and I haven't done a final cumulative chart there either, but that will happen soon too.

Hope this helps.

CDu
03-07-2012, 02:36 PM
So all my stop % is showing is the following: when a given player is on the floor, how often is the TEAM getting a stop and how often is the TEAM giving up a hoop. All five guys who are on the floor get "credit" for the stop and all five get "blame" when the opponent scores. That's probably why, when accumulated, the perimeter guys and the bigs come out pretty close to each other as a group.

Ah, I see. Then disregard my commentary on guards vs bigs. Instead, what we can sort of see is the performance of a particular player relative to the team's alternatives at the same position (only unadjusted for quality of opposing players being faced). So we should expect the starters to have slightly deflated stop% and the reserves to have slightly better (assuming that reserves play proportionally more of their minutes against the opposing team's reserves, and assuming that our reserves are better than most teams' reserves). To some degree we see that (Cook's and Hairston's stop% are higher than average, Mason below average), though we've shuffled the lineup and rotation quite a bit so some of that is muted.

FellowTraveler
03-07-2012, 03:09 PM
So all my stop % is showing is the following: when a given player is on the floor, how often is the TEAM getting a stop and how often is the TEAM giving up a hoop.

Have you maintained the data in a way that would allow you to generate stop percentages (or efficiency ratings) for lineups, in addition to individual players?

tommy
03-08-2012, 12:05 AM
Have you maintained the data in a way that would allow you to generate stop percentages (or efficiency ratings) for lineups, in addition to individual players?

I have the raw, done by hand, possession by possession data on the scoresheets that I created, but in order to translate those into percentages and ratings for particular lineups I would essentially have to go back to those sheets and figure out the numbers you're asking about manually. Huge, huge amount of work to do by hand. So I guess the answer is no. Sorry.