PDA

View Full Version : "Fourth Quarter" Woes



pfrduke
03-08-2014, 08:51 AM
We've had a closing issue in all our losses:

Kansas - 73-72 Duke with 5:00 to play, Kansas closes 22-11 down the stretch
Arizona - 49-49 with 10:00 to play, Arizona closes 23-17
Notre Dame - 60-53 Duke with 12:00 to play, Notre Dame closes 26-17 down the stretch
Clemson - 50-49 Duke with 10:00 to play, Clemson closes 23-9 down the stretch
Syracuse is a bit of an exception (although we did give up 37 points in the last 16 minutes)
Carolina - 60-58 Duke with 5:00 to play, Carolina closes 16-6 down the stretch
Wake - 63-55 Duke with under 6:00 to play, Wake closes 27-9 down the stretch.

Yeesh.

I wanted to come back to the closing issues that we've had. Pomeroy has the "by quarter" scores for each of our games this season (treating each 10 minutes as a quarter). Not surprisingly, it shows that we have had a 4th quarter problem. Somewhat surprisingly, though, this has happened almost only in our losses.

Across all ACC games, we're +47 through the first 10 minutes, +67 in the second 10 minutes (by far our best performance by quarter), +47 for the third 10 minutes, and then +7 for the last 10 minutes and overtime. That's a big disparity - we're 40 points worse at the end of conference games than in any other quarter, and 60 points worse in the last 10 minutes of a game than in the last 10 minutes of the first half (where we are consistently excellent).

In only the losses, that disparity really rears its ugly head: we're -2 in the first 10 minutes, +11 in the second 10 minutes, +1 in the third 10 minutes, and then -45 in the last 10 minutes and overtime. Look at that again - in 5 conference losses we've collectively been ahead (by an average of 2 points, but still ahead) with 10 minutes to play, and then outscored by 45 points in the combined final 55 minutes. We break down on both offense and defense, but particularly on defense - by quarter, we scored 84, 95, 95, and 78 points and allowed 86, 84, 94, and 121(!). If we played a single 40 minute game the way we've played in 4th quarters in our losses, we'd be outscored roughly 88-57.* Yeesh.

*This not a perfectly representative approach. We've played at a much faster pace down the stretch in these games due to extra possessions from free throws at the end (and shortened offensive possessions as we try to score quickly) and the margin is also somewhat inflated by those free throws.

Looking at the wins only, the disparity completely goes away and we play consistent basketball across all segments: +49 through the first 10, +56 through the second 10, +46 through the third 10, and, critically, +52 in the last 10 minutes.

A more concerning way to look at the trend is to isolate out only those games that have been close (that is, within a single digit margin) with 10 minutes to play. 8 of our 17 conference games so far have been within single digits with 10 to play. Although we've been ahead at that point in the game in 7 of those 8 games - and in the 8th we were down just 2 (@ Syracuse) - our collective record in those games is 3-5 and we outscored our opponent in the 4th quarter only twice (@ Pitt, where we finished +12, and home vs Syracuse, where we finished +3*). Our collective 4th quarter performance in the 8 "close" games is -30.

In the 9 games that haven't been particularly close with 10 to play - that is, where we've had a double digit lead - we've not only won all of those games, but we've outscored our opponents in the 4th quarter in 6 of the 9 games, and only Virginia outscored us by more than 2 points (in the others, Wake cut our lead from 22 to 20 in the final ten minutes and BC cut our lead from 22 to 21).

*Note that we got 3 points from the Boeheim double technical, at which point Syracuse had been +1 over the 4th quarter.

So, we have had no problem (with one exception) closing out games where we're up 10 or more and had a big problem (with one exception) closing out games where we're not. The question is whether this is an actual flaw with the team (i.e., something likely to repeat) or just something that's happened to date. 17 games is not a huge sample size (and the sub-sets of 8 and 9 games are even smaller). And there doesn't seem to be a fundamental basketball reason that we play badly in close games down the stretch. But as we get to the point of the season where all the games are likely to be close, we need to reverse the trend or we're not going to get very far in either tournament.

Papa John
03-08-2014, 09:40 AM
I wanted to come back to the closing issues that we've had. Pomeroy has the "by quarter" scores for each of our games this season (treating each 10 minutes as a quarter). Not surprisingly, it shows that we have had a 4th quarter problem. Somewhat surprisingly, though, this has happened almost only in our losses.

Across all ACC games, we're +47 through the first 10 minutes, +67 in the second 10 minutes (by far our best performance by quarter), +47 for the third 10 minutes, and then +7 for the last 10 minutes and overtime. That's a big disparity - we're 40 points worse at the end of conference games than in any other quarter, and 60 points worse in the last 10 minutes of a game than in the last 10 minutes of the first half (where we are consistently excellent).

In only the losses, that disparity really rears its ugly head: we're -2 in the first 10 minutes, +11 in the second 10 minutes, +1 in the third 10 minutes, and then -45 in the last 10 minutes and overtime. Look at that again - in 5 conference losses we've collectively been ahead (by an average of 2 points, but still ahead) with 10 minutes to play, and then outscored by 45 points in the combined final 55 minutes. We break down on both offense and defense, but particularly on defense - by quarter, we scored 84, 95, 95, and 78 points and allowed 86, 84, 94, and 121(!). If we played a single 40 minute game the way we've played in 4th quarters in our losses, we'd be outscored roughly 88-57.* Yeesh.

*This not a perfectly representative approach. We've played at a much faster pace down the stretch in these games due to extra possessions from free throws at the end (and shortened offensive possessions as we try to score quickly) and the margin is also somewhat inflated by those free throws.

Looking at the wins only, the disparity completely goes away and we play consistent basketball across all segments: +49 through the first 10, +56 through the second 10, +46 through the third 10, and, critically, +52 in the last 10 minutes.

A more concerning way to look at the trend is to isolate out only those games that have been close (that is, within a single digit margin) with 10 minutes to play. 8 of our 17 conference games so far have been within single digits with 10 to play. Although we've been ahead at that point in the game in 7 of those 8 games - and in the 8th we were down just 2 (@ Syracuse) - our collective record in those games is 3-5 and we outscored our opponent in the 4th quarter only twice (@ Pitt, where we finished +12, and home vs Syracuse, where we finished +3*). Our collective 4th quarter performance in the 8 "close" games is -30.

In the 9 games that haven't been particularly close with 10 to play - that is, where we've had a double digit lead - we've not only won all of those games, but we've outscored our opponents in the 4th quarter in 6 of the 9 games, and only Virginia outscored us by more than 2 points (in the others, Wake cut our lead from 22 to 20 in the final ten minutes and BC cut our lead from 22 to 21).

*Note that we got 3 points from the Boeheim double technical, at which point Syracuse had been +1 over the 4th quarter.

So, we have had no problem (with one exception) closing out games where we're up 10 or more and had a big problem (with one exception) closing out games where we're not. The question is whether this is an actual flaw with the team (i.e., something likely to repeat) or just something that's happened to date. 17 games is not a huge sample size (and the sub-sets of 8 and 9 games are even smaller). And there doesn't seem to be a fundamental basketball reason that we play badly in close games down the stretch. But as we get to the point of the season where all the games are likely to be close, we need to reverse the trend or we're not going to get very far in either tournament.

I think the issue we have when games get tight at the end can be attributed to a lack of floor leadership and cohesion. I can't recall who posted it on the post-Wake thread, but I believe that the winning formula for your team involves a measure of talent, quality coaching, floor leadership, cohesion, experience, and toughness. Clearly, we have the first two elements in spades. Because of our relative youth, I think we're lacking a strong, assertive floor leader. This often translates into a lack of cohesion when games get tight, and devolves into a lack of toughness at those times as well. It's painful to watch us disintegrate, but that's how I see it. The upside is, we have a couple of guys talented enough to grab the leadership reigns and carry us as far as the ACC or even national championship. The question is whether that will actually happen. I'm still eager to watch and see, either way...

Mtn.Devil.91.92.01.10.15
03-08-2014, 10:10 AM
I think the issue we have when games get tight at the end can be attributed to a lack of floor leadership and cohesion. I can't recall who posted it on the post-Wake thread, but I believe that the winning formula for your team involves a measure of talent, quality coaching, floor leadership, cohesion, experience, and toughness. Clearly, we have the first two elements in spades. Because of our relative youth, I think we're lacking a strong, assertive floor leader. This often translates into a lack of cohesion when games get tight, and devolves into a lack of toughness at those times as well. It's painful to watch us disintegrate, but that's how I see it. The upside is, we have a couple of guys talented enough to grab the leadership reigns and carry us as far as the ACC or even national championship. The question is whether that will actually happen. I'm still eager to watch and see, either way...

I was having this same conversation with another manic Duke fan (my dad) and we were discussing how we have both veterans and talent, but not really a talented veteran who can instill the confidence and leadership. I do wonder if it is partly because our experience and our talent aren't residing in the same bodies that we are seeing a vacuum in this regard. It is still possible for Hood or Parker to step up, but it needs to be post-haste. I'd love to see one of them basically put the team on their back for a few possessions and will the team to victory.

I wouldn't even mention this, but since there's currently an article on ESPN about Parker's lack of certainty about the draft I could absolutely see him coming back and developing into that take charge guy with the killer instinct, and leading a talented young group to some serious victories.

At any rate, I think that if anyone is going to step forward and claim that role this year, it will be Hood or Parker and it needs to happen immediately. I hope that the coaches are doing a little poking in this regard. There might possibly only be three games left of this season.

Go Duke! Beat UNC!

CR9
03-08-2014, 10:11 AM
Thing is, this only outlines games they've lost. This is a trend, even in wins. They gave second-half, double-digit leads away against Vermont, ECU, 'Bama, Maryland, and UVA. I know a win is a win, etc but it's slightly unnerving, going into the tourney, that this can happen against bad teams as well as good ones.

pfrduke
03-15-2014, 01:34 PM
Updating this with our last two games (in which we gave up 50 combined points in the final 10 minutes of both games):

Across all ACC games, we're +60 through the first 10 minutes, +61 in the second 10 minutes, +64 for the third 10 minutes, and then -4 for the last 10 minutes and overtime.

Look at that again - we outscore opponents by, collectively, 185 points through 30 minutes, and then get collectively outscored at the end.

Isolating the close games again (which now includes the Clemson game), we've been ahead with 10 to play in 8 of the 9 games - and in the 9th we were down just 2 (@ Syracuse) - our collective record in those games is 4-5 and we outscored our opponent in the 4th quarter only twice (@ Pitt, where we finished +12, and home vs Syracuse, where we finished +3*). Our collective 4th quarter performance in the 9 "close" games is -37.

*Note that we got 3 points from the Boeheim double technical, at which point Syracuse had been +1 over the 4th quarter.

All told, we're 8-11 in our 4th quarter performance.

davekay1971
03-16-2014, 09:44 AM
First of all, great thread and breakdown, pfr! This is THE critical issue for this Duke team. This team scares me more with a lead in the last 10 minutes of the game than any other Duke team I can remember.

I have no idea why they lose these 4th quarter leads. The tendencies which I think I've picked up are:

1) Inability to get stops. I'd be interested to see defensive efficiency stats in the last 10 minutes of our games, and I'm willing to guess they're pretty ugly. But why? Clearly our guys are focused during the closing minutes. It's almost like they defensively "choke" - having critical communication lapses that allow other teams crucial baskets.

2) Offensive hesitancy. I put this down to PG play. We don't have a guy who seems able to reliably set up the offense in crunch time. Therefore, some of our late game offensive possessions, when the opponent is bearing down defensively, turn into one guy trying to create against set defense.

3) Go to guy. We have two of the best scorers in the ACC, but neither Parker nor Hood have been able to "take over" or "put the team on his shoulders" when we need those late buckets, at least not reliably. K has talked about Jabari's conditioning, which has improved over the season. Is Jabari gassed by end of game? FWIW, late in the season I've found myself wanting the ball in Sulaimon's hands when we need a score late in the game. If he can improve his ability to dish out of a drive, given his ability to get to the rim against virtually anyone, he could become that guy that absolutely terrifies opponents when he gets the ball in his hands late in a close game.

One thing interesting, and frustrating, is that this is a deep Duke team. Our guys should be as strong, or stronger, than most opponents by the end of the game. But, especially in our losses and in a couple of close wins, when the opponent picks up the intensity, our guys seem to be unable to respond. Are they gassed, tight, or what? I had hoped we'd sorted that out, but the UNC and Wake losses, and the Clemson win, all within the last month, have shown that the team still can suffer from these strange end-game struggles.

At this point of the season, every game is a championship game. If they win seven more games, leading each one by 10 points with 6 minutes left and barely clinging on to win each by 1, I won't care too much about these disturbing "fourth quarter" differentials. But if we end our season on a loss in a game where we gave up a late lead, this tendency will be THE problem we couldn't quite figure out this season.

nyesq83
03-16-2014, 11:06 AM
I have seen or heard or read K stating - after multiple games this year - that the team was "tired" down the stretch.

Johnny Dawkins has been gone for many years now. My understanding is that he was a task master as far as building the players' stamina up through distance running etc.

I believe Nate Dogg took over the conditioning role, but has different responsibilities, recruiting perhaps, but who is the strength and conditioning coach, or is there one?

I also wonder if today's video game culture, and the likelihood that the players all have cars or friends with cars, has an impact on each player's endurance.

I didn't have a car in four years at Duke. I was a distance runner until junior year. I would bet that Duke players do not walk as much as I did, and after hearing K say the team is "tired" I get the impression they do not run for distance all that often.

TruBlu
03-16-2014, 11:22 AM
Here goes, along with my suggestions for correcting the problem: (Tongue planted in cheek, somewhat)

1) Poor substitution pattern - - Coaches should do better

2) Practices too hard - - Coaches should do better

3) Conditioning - - Too late to correct this year. Coaches should institute better conditioning next year

4) Nutrition - - Coaches should stop players from eating/drinking what I eat/drink

5) Ankle weights - - Coaches should inspect players before games to insure they aren't wearing them

I am still amazed that we were tired at the end of the Clemson game (according to coach K) after 5 days off, but Clemson seemed to have a lot of bounce in their step after 2 consecutive overtime games.:confused:

pfrduke
03-19-2014, 02:29 PM
Updating this through the end of the ACC season:

Across all ACC games, we're +52 through the first 10 minutes, +67 in the second 10 minutes, +74 for the third 10 minutes, and then -13 for the last 10 minutes and overtime.

We outscore opponents by, collectively, 193 points through 30 minutes, and then get collectively outscored at the end. We appear to improve each set of 10 minutes and then crater down the stretch.

Isolating just the "close" games again (i.e., where there was a single-digit margin at the 30-minute mark - which now includes all ACCT games), we've been ahead with 10 to play in 9 of the 11 games - and in the others we were down just 2 (@ Syracuse), and 1 (UVA ACCT) - our collective record in those games is 5-6 and we outscored our opponent in the 4th quarter only twice (@ Pitt, where we finished +12, and home vs Syracuse, where we finished +3*). Our collective 4th quarter performance in the 11 "close" games is -46; in the 6 losses it's -53 (we've lost those 6 games by a total of 44 points).

*Note that we got 3 points from the Boeheim double technical, at which point Syracuse had been +1 over the 4th quarter.

All told, we're 8-13 in our 4th quarter performance in ACC play.

weezie
03-19-2014, 04:15 PM
Nutrition - - Coaches should stop players from eating/drinking what I eat/drink

Ankle weights - - Coaches should inspect players before games to insure they aren't wearing them


Great insight. Thoughtful, balanced and deeply pondered. No eye-rolling necessary.

Zeb
03-19-2014, 06:22 PM
I also wonder if today's video game culture... has an impact on each player's endurance.


Do you say this when you get tired of yelling "Get off my lawn?"

Conditioning may play a part in 4th qtr letdowns, but video games play no part in this. Unless you honestly think Duke players' video game habits impact their conditioning more than those of players at other Div I schools. Which just sounds stupid.

superdave
03-21-2014, 02:47 PM
I believe we got outscored 20 to 8 over the final 4:30 today. Can anyone confirm?

It really is amazing that all out losses ended the same way. There's a fundamental problem with this particular mix of guys and coaches, and we never figured out how to identify it and correct it.

94duke
03-21-2014, 02:52 PM
I believe we got outscored 20 to 8 over the final 4:30 today. Can anyone confirm?

It really is amazing that all out losses ended the same way. There's a fundamental problem with this particular mix of guys and coaches, and we never figured out how to identify it and correct it.

We were up 63 - 58 with 4:53 to go.

CDu
03-21-2014, 03:04 PM
I believe we got outscored 20 to 8 over the final 4:30 today. Can anyone confirm?

It really is amazing that all out losses ended the same way. There's a fundamental problem with this particular mix of guys and coaches, and we never figured out how to identify it and correct it.

Yeah, it is uncanny. It shouldn't be a minutes issue (at least not in this one), as only 3 players topped 30 mpg. Maybe it is a leadership issue? A "not able to handle pressure" issue? Hard to say. But it really is amazing.

We got just 3 defensive stops in the last 7:42 of the game...

dukelifer
03-21-2014, 03:12 PM
Yeah, it is uncanny. It shouldn't be a minutes issue (at least not in this one), as only 3 players topped 30 mpg. Maybe it is a leadership issue? A "not able to handle pressure" issue? Hard to say. But it really is amazing.

We got just 3 defensive stops in the last 7:42 of the game...

It is the enigma of this specific team. Duke teams usually understand how to build a lead - how to get and make good shots at critical moments and how to get stops. This team has had leads in most every game this year at money time (3-4 minutes to go) and they went south in a hurry. I think it is a pressure management thing. Duke players also have the weight of the entire program on them. When Duke loses - it is a big deal. Some players learn to handle it and some don't. Some players refuse to lose. We have had our share of those over time.

mr. synellinden
03-21-2014, 04:06 PM
It is the enigma of this specific team. Duke teams usually understand how to build a lead - how to get and make good shots at critical moments and how to get stops. This team has had leads in most every game this year at money time (3-4 minutes to go) and they went south in a hurry. I think it is a pressure management thing. Duke players also have the weight of the entire program on them. When Duke loses - it is a big deal. Some players learn to handle it and some don't. Some players refuse to lose. We have had our share of those over time.

In our last 7 tournament losses, we have been outscored 332-247 in the second half for an average of 47.4 - 33.4. In all but one of those games, Duke was seeded higher. There was a late game problem this year, but there is a second half problem for Duke overall. I posted some thoughts as to why in the post-game thread which I won't repeat here.

_Gary
03-21-2014, 05:13 PM
In our last 7 tournament losses, we have been outscored 332-247 in the second half for an average of 47.4 - 33.4. In all but one of those games, Duke was seeded higher. There was a late game problem this year, but there is a second half problem for Duke overall. I posted some thoughts as to why in the post-game thread which I won't repeat here.

And you were spot on with those comments! The bottom line is that Duke generally hasn't been getting as good a look at the basket late as the other team has. And I believe that boils down to something really simple that I've mentioned from the beginning of this season - lack of a PG that can break down the defense with penetration. As much as I love Tyler's grit, that certainly has never been his strong suit. And as much as I love Quinn it's not his strong suit either. He has shown flashes of this in times past, but he's really not a guard that penetrates consistently in the half court. And when you don't have a player that can do this, you are forced into what we did all season long - depend on individual talent to create their own shot. We saw it game in and game out from Jabari, Rodney and Rasheed. The vast majority of their points came off their own "work", not moving without the ball and receiving a well-timed and well-placed assist. That kind of play generally tightens up and becomes less available as the game comes to a close. The need for "team" offense, facilitated normally (but not always) by a guard that can beat his man off the dribble, is critical down the stretch. We never really had this the entire year. And that's what killed us time and time again late in games on the offensive end of the court. Defense is a whole another animal that I'll refrain from posting about right now.

bedeviled
03-25-2014, 07:03 PM
Well, I said in another thread I wasn't going to do this....I lied. Here is a breakdown of the offensive and defensive efficiencies for the first 35 minutes of games and the last 5 minutes of games (Syracuse OT was ignored) (please, please don't tell me Pomeroy already calculates efficiency for his "quarters." I'll be sad).

I was interested in Defensive Efficiency in the first 35 minutes vs the last 5 minutes (I would have chosen the last official timeout, but it comes at different times in the games). Here are the averages for Offensive Efficiency in the first 35 vs last 5 minutes and Defensive Efficiency in the first 35 vs last 5 minutes:



35OEff
5OEff
35DEff
5DEff


Full Season
119.7
127.5
100.5
120.6


Since ACC play
119.8
119.7
103.5
120.9



I don't know what sorting or subsets people would be interested in (eg conference, Top25, games we lost, away games). So, I'll leave you with the full table so you can manipulate it however you'd like. If you are want of ideas, though, maybe you'd like to see a graphical representation of the differences in OEff and DEff per game. So, here you go:
4032

And, here's the full table. Possessions were calculated using Pomeroy's formula, including average Duke & Opponents possessions. Possessions in the first 35 minutes, though, was calculated by subtracting the possessions I tabulated for the last 5 minutes from the total possessions listed on Statsheet. My calculations were done using ESPN play-by-play data.
35Pos = Possessions in the first 35 minutes
Duke35Pts = Points Duke scored in the first 35 minutes
Opp35Pts = Points Opponent scored in the first 35 minutes
5Pos = Possessions in the last 5 minutes
and so on.



Team
35Pos
Duke35Pts
Opp35Pts
5Pos
Duke5Pts
Opp5Pts

35OEff
5OEff
35DEff
5DEff


Dav
59
95
69
11
16
8

161.0
145.5
116.9
72.7


KU
64
73
72
11
10
22

114.1
90.9
112.5
200.0


FAU
61
79
50
11
18
14

129.5
163.6
82.0
127.3


UNCA
61
79
49
8
12
6

129.5
150.0
80.3
75.0


ECU
63
67
64
9
16
10

106.3
177.8
101.6
111.1


UVM
55
75
78
9
16
12

136.4
177.8
141.8
133.3


ALA
62
58
51
12
16
13

93.5
133.3
82.3
108.3


Ariz
56
51
57
11
15
15

91.1
136.4
101.8
136.4


Mich
54
60
44
12
19
25

111.1
158.3
81.5
208.3


Webb
56
77
58
7
8
8

137.5
114.3
103.6
114.3


UCLA
61
65
57
11
15
6

106.6
136.4
93.4
54.5


EMU
60
71
50
9
11
9

118.3
122.2
83.3
100.0


Elon
63
75
37
9
11
11

119.0
122.2
58.7
122.2


ND
54
64
65
10
13
14

118.5
130.0
120.4
140.0


GT
54
66
55
7
13
2

122.2
185.7
101.9
28.6


Clem
56
57
64
8
2
8

101.8
25.0
114.3
100.0


UVA
51
56
47
10
13
18

109.8
130.0
92.2
180.0


NCSU
62
84
52
7
11
8

135.5
157.1
83.9
114.3


MIA
50
58
40
7
9
6

116.0
128.6
80.0
85.7


FSU
57
67
54
6
11
2

117.5
183.3
94.7
33.3


PITT
52
67
57
9
13
8

128.8
144.4
109.6
88.9


SU
58
76
81
9
13
10

131.0
144.4
139.7
111.1


WFU
56
74
57
6
9
6

132.1
150.0
101.8
100.0


BC
55
80
56
7
9
12

145.5
128.6
101.8
171.4


MARY
60
59
60
8
10
7

98.3
125.0
100.0
87.5


GT
51
60
42
7
8
9

117.6
114.3
82.4
128.6


UNC
58
57
56
10
9
18

98.3
90.0
96.6
180.0


SU
50
53
49
10
13
11

106.0
130.0
98.0
110.0


VT
55
60
44
6
6
4

109.1
100.0
80.0
66.7


WFU
53
66
61
13
6
21

124.5
46.2
115.1
161.5


UNC
55
79
65
13
14
16

143.6
107.7
118.2
123.1


CLEM
50
56
50
6
7
12

112.0
116.7
100.0
200.0


NCSU
46
65
54
9
10
13

141.3
111.1
117.4
144.4


UVA
50
53
57
10
10
15

106.0
100.0
114.0
150.0


MERC
50
60
58
13
11
20

120.0
84.6
116.0
153.8

Troublemaker
03-25-2014, 07:48 PM
So, I'll leave you with the full table so you can manipulate it however you'd like. If you are want of ideas, though, maybe you'd like to see a graphical representation of the differences in OEff and DEff per game. So, here you go:
4032


Nice graph (and great work overall)! Looks like towards the end of the season, both offense and defense would crumble at the end of games. Whereas earlier in the season, it was usually just the defense.

OldSchool
03-25-2014, 08:02 PM
Well, I said in another thread I wasn't going to do this....I lied. Here is a breakdown...

Wow - over the entire last six weeks of the season, there were only three games in which we outscored our opponents in the final five minutes of the game.

sagegrouse
03-25-2014, 08:25 PM
In our last 7 tournament losses, we have been outscored 332-247 in the second half for an average of 47.4 - 33.4. In all but one of those games, Duke was seeded higher. There was a late game problem this year, but there is a second half problem for Duke overall. I posted some thoughts as to why in the post-game thread which I won't repeat here.

It's not kosher statistics to just use "losses." You should use "all games" of a certain type over a certain time period, such as all tournament games since 2009. Reason is, random events can lead to losses, not just poor play, and using just losses will make Duke's play look worse than it actually is.

I still think you will get same result, just a bit smaller.

Faustus
03-25-2014, 08:53 PM
Some games (extremely few this year) when there was a big Duke lead, the "call off the dogs" and bench-clearing factor would also tend towards getting outscored in the final 5 minutes, but those would have been in 'meaningless' final minutes with the game well in hand, and should essentially be tossed out from consideration. Including those stats by using "all games" would actually muddy the picture a little bit to my mind. True, such numbers could also be distorted in losses as well - a team behind late in the game starts throwing up bad shots in a hurry, and fouls intentionally, often giving up inflated numbers that way. But Duke's losses this year were virtually all in games closely contested with 5 minutes remaining. I think these 'last 5 minutes stats' for games only in which Duke lost this year are quite revealing myself.

Disclaimer --- I am far from being a trained stats guy.

-jk
03-25-2014, 08:58 PM
Better would be games within - say - 10 points at the 5 minute mark.

-jk

sagegrouse
03-25-2014, 09:03 PM
Some games (extremely few this year) when there was a big Duke lead, the "call off the dogs" and bench-clearing factor would also tend towards getting outscored in the final 5 minutes, but those would have been in 'meaningless' final minutes with the game well in hand, and should essentially be tossed out from consideration. Including those stats by using "all games" would actually muddy the picture a little bit to my mind. True, such numbers could also be distorted in losses as well - a team behind late in the game starts throwing up bad shots in a hurry, and fouls intentionally, often giving up inflated numbers that way. But Duke's losses this year were virtually all in games closely contested with 5 minutes remaining. I think these 'last 5 minutes stats' for games only in which Duke lost this year are quite revealing myself.

Disclaimer --- I am far from being a trained stats guy.

Sorry, Faustus, but one can't reason backwards from the results. One reasons "forward" from the total sample of games. There is a "selectivity bias" in starting with just losses, and it took me years of study (many decades ago) to really understand that point, and, of course, I won't give it up for anything.

Newton_14
03-25-2014, 09:08 PM
Well, I said in another thread I wasn't going to do this....I lied. Here is a breakdown of the offensive and defensive efficiencies for the first 35 minutes of games and the last 5 minutes of games (Syracuse OT was ignored) (please, please don't tell me Pomeroy already calculates efficiency for his "quarters." I'll be sad).

I was interested in Defensive Efficiency in the first 35 minutes vs the last 5 minutes (I would have chosen the last official timeout, but it comes at different times in the games). Here are the averages for Offensive Efficiency in the first 35 vs last 5 minutes and Defensive Efficiency in the first 35 vs last 5 minutes:



35OEff

5OEff

35DEff

5DEff



Full Season

119.7

127.5

100.5

120.6



Since ACC play

119.8

119.7

103.5

120.9








Team

35Pos

Duke35Pts

Opp35Pts

5Pos

Duke5Pts

Opp5Pts


35OEff

5OEff

35DEff

5DEff



Dav

59

95

69

11

16

8


161.0

145.5

116.9

72.7



KU

64

73

72

11

10

22


114.1

90.9

112.5

200.0



FAU

61

79

50

11

18

14


129.5

163.6

82.0

127.3



UNCA

61

79

49

8

12

6


129.5

150.0

80.3

75.0



ECU

63

67

64

9

16

10


106.3

177.8

101.6

111.1



UVM

55

75

78

9

16

12


136.4

177.8

141.8

133.3



ALA

62

58

51

12

16

13


93.5

133.3

82.3

108.3



Ariz

56

51

57

11

15

15


91.1

136.4

101.8

136.4



Mich

54

60

44

12

19

25


111.1

158.3

81.5

208.3



Webb

56

77

58

7

8

8


137.5

114.3

103.6

114.3



UCLA

61

65

57

11

15

6


106.6

136.4

93.4

54.5



EMU

60

71

50

9

11

9


118.3

122.2

83.3

100.0



Elon

63

75

37

9

11

11


119.0

122.2

58.7

122.2



ND

54

64

65

10

13

14


118.5

130.0

120.4

140.0



GT

54

66

55

7

13

2


122.2

185.7

101.9

28.6



Clem

56

57

64

8

2

8


101.8

25.0

114.3

100.0



UVA

51

56

47

10

13

18


109.8

130.0

92.2

180.0



NCSU

62

84

52

7

11

8


135.5

157.1

83.9

114.3



MIA

50

58

40

7

9

6


116.0

128.6

80.0

85.7



FSU

57

67

54

6

11

2


117.5

183.3

94.7

33.3



PITT

52

67

57

9

13

8


128.8

144.4

109.6

88.9



SU

58

76

81

9

13

10


131.0

144.4

139.7

111.1



WFU

56

74

57

6

9

6


132.1

150.0

101.8

100.0



BC

55

80

56

7

9

12


145.5

128.6

101.8

171.4



MARY

60

59

60

8

10

7


98.3

125.0

100.0

87.5



GT

51

60

42

7

8

9


117.6

114.3

82.4

128.6



UNC

58

57

56

10

9

18


98.3

90.0

96.6

180.0



SU

50

53

49

10

13

11


106.0

130.0

98.0

110.0



VT

55

60

44

6

6

4


109.1

100.0

80.0

66.7



WFU

53

66

61

13

6

21


124.5

46.2

115.1

161.5



UNC

55

79

65

13

14

16


143.6

107.7

118.2

123.1



CLEM

50

56

50

6

7

12


112.0

116.7

100.0

200.0



NCSU

46

65

54

9

10

13


141.3

111.1

117.4

144.4



UVA

50

53

57

10

10

15


106.0

100.0

114.0

150.0



MERC

50

60

58

13

11

20


120.0

84.6

116.0

153.8





Great work. Thanks. I have been thinking a lot on this since Friday's painful exit. For me the meltdowns start with a lack of poise under extreme pressure. That was big for this team and it showed in every loss. There just wasn't anyone who was a calming influence on the floor, who could settle them down, keep them confident and able to execute on both ends. So that kind of goes hand in hand with the Leadership aspect. Hood and Jabari were the best players but neither of them fit the bill of "calming presence". Hood, bless his heart got butterflies so badly he often threw up before and during games. Having a weak stomach myself, I can relate, and please don't mistake that as choking. Hood is not a choker at all. Still, he does not qualify as a calming presence either. Jabari will probably grow into that type guy but not there yet. Realistically, the only other candidates were Quinn and Rasheed, and both of those guys are highly emotional, and wear their feelings on their sleeve's.

Mike Decourcy commented on this yesterday on the Adam Gold show and stated it could not be Tyler Thornton because he was not one of the top talented players and was not going to be on the floor 32+ minutes a game night in and night out. He felt, and I agree, that leader/calming presence, has to be a veteran guy who is on the floor those amount of minutes and one of your better players. While Tyler was a really solid player for us and a top level defender, he was not a go to guy on offense that could get you a bucket when you desperately needed it.

The two other contributing factors were more simplistic.

Contributing Factor Number 2: Defense. For whatever reason, this group just could not defend well and could not force turnovers. Like OldPhiCap stated in another thread, K did in fact try multiple defenses throughout the year. At the start he tried pressing fullcourt with both mantoman and zone presses, then abandoned it when they used a ton of energy and created no turnovers. He tried half court presses, altered where we picked up with man to man, played some zone, etc. Nothing worked all that well. The rotations, communication, ball screen d, etc etc just never came together fully for reasons unknown. We gave up more good looks from 3 this year than I ever remember doing, and could not defend the paint either off the bounce or in the post, and could not block shots.

Contributing Factor Number 3- Offense- Efficiency was good in those first 30 minutes as yours and PFR's data showed, but with the game on the line or with the opponent making a run, our offense could not generate buckets when we needed them the most. Part of that is a direct correlation to contributing factor 1 (Poise/Calming Presence/Leadership), and part of it is lack of execution. I also don't think we did the best job in the world this year of designing the offense to get guys the ball in the best spot on the floor for them. I think that could have been better. Watching UVA vs Memphis Sunday Night had me wishing we ran their offense. Lots of movement, cutting, screening, and they keep constant pressure on the defense. They also are very sneaky with their fastbreaking. They will run on you. I wish we had pushed the ball much more than we did. The theme on the poster this year was "The Fast and The Furious" afterall. Generating easy buckets through fast breaks and secondary fast breaks helps take pressure off your halfcourt offense. You dont have to go breakneck speed for 40 minutes or nothing but you can certainly run more.

Anyway, those are the 3 key things in my mind that hurt these guys the most this season, and led to the meltdowns we saw in each loss. So I anxiously await next season, and see what changes are in store. Having a top notch big man and pure point guard will cure a lot of ills. Rasheed, Quinn, and Amile will be older, wiser, and improved. Two of those three guys have to be Captains and Leaders. Dont care which two. MP3 and Semi should each be improved with the former hopefull entering the season fully healthy and with a complete healthy off season of work behind him. The latter is uber athletic and if he can learn how to play the game at a high level to go with that talent can be scary good. Winslow looks to be a really good defender and rebounder which should get him minutes. Were Jabari to come back, I think he would be in much better shape than this season, and the jump in his game will be tremendous. I would definitely pick us preseason number 1 if he comes back. I recognize the chances of that happening are not great but who knows, maybe the kid surprises us. I have seen Grayson play some but not enough to tell if he can play his way into the rotation or not. If yes, thats a plus.

Can't wait to see Okafor and Jones in Duke uni's. I just hope we somehow get to see them in the Pro-Am summer league this summer so we don't have to wait until that first open practice next Fall.

Go Duke!

Furniture
03-25-2014, 09:22 PM
I think this year we didn't have a 'go to' guy like Schyer or Nolan. Very early on in the season and at times last year it was Cook. This year since Cook lost confidence, or whatever, I would like to suggest that K was trying to mould Sheed. Although Sheed was for the most part outstanding he wasn't that guy at the end of a game or shot clock.

sagegrouse
03-25-2014, 09:45 PM
Sorry, Faustus, but one can't reason backwards from the results. One reasons "forward" from the total sample of games. There is a "selectivity bias" in starting with just losses, and it took me years of study (many decades ago) to really understand that point, and, of course, I won't give it up for anything.


I did all the tournament games (1) since 2010 and (2) since 2011. In the former case, our 1H margin was +4.3 and our 2H margin was +2.2.

Since 2011, our 1H margin was +4.1 and our 2H was -0.2. This is far different from the -14 2H margin achieved by Mr. Synellinden using only our seven losses.

Kindly, Sage
'Sorry to be so grousy, but I think it's genetic'

mr. synellinden
03-25-2014, 09:48 PM
It's not kosher statistics to just use "losses." You should use "all games" of a certain type over a certain time period, such as all tournament games since 2009. Reason is, random events can lead to losses, not just poor play, and using just losses will make Duke's play look worse than it actually is.

I still think you will get same result, just a bit smaller.


I'm not sure I completely agree with this. To put the point another way, excluding the 2010 championship run, in each NCAA tournament Duke has had a terrible (and in some cases monumental) second half collapse against a lower seeded team (exception being last year against Louisville). Here are the numbers:

2014: Mercer 44 Duke 36
2013: Louisville 50 Duke 31
2012: Lehigh 47 Duke 40
2011: Arizona 55 Duke 33
2009: Villanova 51 Duke 31
2008: West Virginia 44 Duke 33
2007: VCU 41 Duke 37

Some of those second half scores are big blowouts. The only games in which we were able to stay within 10 points (in a HALF) were against double digit seeds. The point is not really how we did overall in NCAA games - (by the way, we lost the second half to Michigan in 2011 (1 vs. 8 seeds) and Texas in 2009 (2 vs. 7 seeds) but only by 2 points in each game) - it is just the fact that a complete second half collapse has happened every year for a long stretch of time with a complete turnover of the roster. It's a very disturbing pattern which I think is attributable to the style of play and coaching.

gotoguy
03-25-2014, 10:00 PM
I think this year we didn't have a 'go to' guy like Schyer or Nolan. Very early on in the season and at times last year it was Cook. This year since Cook lost confidence, or whatever, I would like to suggest that K was trying to mould Sheed. Although Sheed was for the most part outstanding he wasn't that guy at the end of a game or shot clock.


Our goto guy was Jabari who was/is a load in the paint. There were very few ways to get him the ball in the paint at end game situations where he could be effective. Content to pass it around the perimeter and shoot rushed 3's or drive and turn it over, Duke lacked the offensive ingenuity to best player the ball where he could use his considerable offensive talents. A coaching failure IMHO.

wallyman
03-25-2014, 10:28 PM
I'm not sure I completely agree with this. To put the point another way, excluding the 2010 championship run, in each NCAA tournament Duke has had a terrible (and in some cases monumental) second half collapse against a lower seeded team (exception being last year against Louisville). Here are the numbers:

2014: Mercer 44 Duke 36
2013: Louisville 50 Duke 31
2012: Lehigh 47 Duke 40
2011: Arizona 55 Duke 33
2009: Villanova 51 Duke 31
2008: West Virginia 44 Duke 33
2007: VCU 41 Duke 37

Some of those second half scores are big blowouts. The only games in which we were able to stay within 10 points (in a HALF) were against double digit seeds. The point is not really how we did overall in NCAA games - (by the way, we lost the second half to Michigan in 2011 (1 vs. 8 seeds) and Texas in 2009 (2 vs. 7 seeds) but only by 2 points in each game) - it is just the fact that a complete second half collapse has happened every year for a long stretch of time with a complete turnover of the roster. It's a very disturbing pattern which I think is attributable to the style of play and coaching.le

Yes, a very disturbing pattern -- over a significant period of time. But I'm curious: In what specific ways do you think it's attributable to the style of play and coaching?

bedeviled
03-25-2014, 10:32 PM
The late-game efficiency data may be able to speak a little to the idea of 'tiredness leads to defensive collapse.'

Looking at correlations using multiple data sets, there aren't good correlations between [DEff in the last 5 minutes] and [Number of possessions in the first 35 minutes]. So, if number of possessions is a semi-reasonable proxy for running up and down the court, then this data suggests that we are not running ourselves into poor late defense.

Of course, sometimes the half-court grind is what wears you out, not the running. Well, there is no correlation between [DEff in the last 5 minutes] and [OEff in the first 35 minutes]
or between [DEff in the last 5 minutes] and [DEff in the first 35 minutes].
So, if OEff and DEff are semi-reasonable proxies for how hard Duke has to work, then late defensive collapses are not due to having to work too hard in the preceding 35 minutes.

Definitely not the most sound extrapolations, but I think they make sense enough to keep in mind when we are conjecturing :)
Hmmmm, I suppose some curious fan could also investigate distribution of minutes vs late-game efficiency if he/she is one who insists on platooning or otherwise resting the stars.

Kedsy
03-25-2014, 11:10 PM
Well, I said in another thread I wasn't going to do this....I lied. Here is a breakdown of the offensive and defensive efficiencies for the first 35 minutes of games and the last 5 minutes of games (Syracuse OT was ignored) (please, please don't tell me Pomeroy already calculates efficiency for his "quarters." I'll be sad).

Wow, very cool stuff. Thanks for doing all that work.


To put the point another way, excluding the 2010 championship run, in each NCAA tournament Duke has had a terrible (and in some cases monumental) second half collapse against a lower seeded team (exception being last year against Louisville).

Actually it's even worse. Did you know that in every loss Duke has suffered in the Coach K era, the opponent has outscored us over the course of the game? I'm not kidding, every one. In fact, if you exclude Duke's wins, we've lost every single game. I think it must be a coaching problem.

Sorry, couldn't resist. :p But Sagegrouse is right. You can't single out Duke losses and expect the data to be anything except bad for Duke.

mr. synellinden
03-26-2014, 03:11 AM
le

Yes, a very disturbing pattern -- over a significant period of time. But I'm curious: In what specific ways do you think it's attributable to the style of play and coaching?

I don't want to repeat it all here but take a look at post 66 in the Mercer post-game thread. That summarizes my thoughts on the issues - playing style-wise which is a direct function of coaching - that lead to poor tournament game performances - especially in the second half

Wow, very cool stuff. Thanks for doing all that work.



Actually it's even worse. Did you know that in every loss Duke has suffered in the Coach K era, the opponent has outscored us over the course of the game? I'm not kidding, every one. In fact, if you exclude Duke's wins, we've lost every single game. I think it must be a coaching problem.

Sorry, couldn't resist. :p But Sagegrouse is right. You can't single out Duke losses and expect the data to be anything except bad for Duke.

Again, I really think is missing the point. I think you can and SHOULD single out losses because ALL of our recent tournament losses have at least one significant thing in common - we get beat very badly in the second half. Look at the numbers. The margins are astonishing. So many people here like to rely on KenPom and it's predictive value. Our second half performances in the 7 recent NCAA losses produced an average margin of -14. I'll bet that the KenPom predictive values would expect Duke to have won those halves by an average of about 3 points per half. That's a pretty incredible disparity from the expected value and enough of a sample size to make the numbers shocking. It's not like the average margin of second half loss was a one point per game - which would have been somewhat close to the predicted margin. It was about 17 points PER GAME WORSE. That's not something you can dismiss or joke away with a cute comment about us being outscored in all of our losses. We have lost BADLY in every NCAA loss for the past seven tournaments. You know what makes the point even stronger - in 5 of the 7 losses we led at the half. And the two we didn't - Louisville and Villanova - we were down by only 3 -- and then proceeded to get blown out in the second half. Do you not think it's remarkable that we've had such a disparity in second half performance in 7 straight NCAA tournament losses? You know what game also fits this pattern? Belmont in 2008 - won the first half by 7 and lost the second half by 6. We were extremely close to having 4 first round losses to double digit seeded teams in the last 7 years. We can make jokes about other teams scoring more points than us in all our losses. But they're not just outscoring us - they're killing us in the second half and in most cases they are huge underdogs - like double digit underdogs - that are killing us in the second half. I'm not even going to try to address the talent disparity Duke has had in most of these games because that will open up a separate can of worms. But anyone who thinks Duke's pattern of losing in the NCAA tournament during the past 8 years is not incredibly remarkable and disturbing is not facing reality. Two more games for reference - remember the first game of the 2011-12 season? - in Cameron - Maui Invitational - against Belmont again. Duke ranked #6 with at least 4 future NBA players playing significant minutes - what happened? Won the first half by 9. Lost the second half by 8. How about Vermont? - of course we remember that game from early this season. Duke again coincidentally ranked #6 - in Cameron - won the first half by 8. Lost the second half by 7.

Look, I am as big of a Duke fan as anyone here. But there is something systemic causing these types of losses and near-losses - all of which involve major first and second half margin disparities - to teams that we would expect to beat with relative ease.

I'll also point out the following:

2002. First half. Duke 42 IU 29. Second half. IU 45 - Duke 31
2004. First half. Duke 41 UCONN 34. Second half. UCONN 45. Duke. 37
2005. First half. Duke 32. MSU 32. Second half. MSU. 46. Duke 36

Furniture
03-26-2014, 06:25 AM
Our goto guy was Jabari who was/is a load in the paint. There were very few ways to get him the ball in the paint at end game situations where he could be effective. Content to pass it around the perimeter and shoot rushed 3's or drive and turn it over, Duke lacked the offensive ingenuity to best player the ball where he could use his considerable offensive talents. A coaching failure IMHO.

I am not going to disagree. I also wonder what the psychological effect on the rest of the team was when they were told by the coaching staff that the only guys that can win us the game (maybe in different words) are JP and Hood? Do you think to yourself ok, I'll just pass them the ball and watch them do the business?

sagegrouse
03-26-2014, 07:21 AM
I'm not sure I completely agree with this. To put the point another way, excluding the 2010 championship run, in each NCAA tournament Duke has had a terrible (and in some cases monumental) second half collapse against a lower seeded team (exception being last year against Louisville). Here are the numbers:

2014: Mercer 44 Duke 36
2013: Louisville 50 Duke 31
2012: Lehigh 47 Duke 40
2011: Arizona 55 Duke 33
2009: Villanova 51 Duke 31
2008: West Virginia 44 Duke 33
2007: VCU 41 Duke 37

Some of those second half scores are big blowouts. The only games in which we were able to stay within 10 points (in a HALF) were against double digit seeds. The point is not really how we did overall in NCAA games - (by the way, we lost the second half to Michigan in 2011 (1 vs. 8 seeds) and Texas in 2009 (2 vs. 7 seeds) but only by 2 points in each game) - it is just the fact that a complete second half collapse has happened every year for a long stretch of time with a complete turnover of the roster. It's a very disturbing pattern which I think is attributable to the style of play and coaching.

Losses are often ugly for both teams. Here are selected tournament wins over the past five years:



Year Rank SCh 1H 2H Tourn
2010 3 D 39 39 N-Sem
2010 6 WVa 31 26

2010 3 D 32 46 R-Fin
2010 19 Bay 35 36

2010 3 D 24 46 R-Sem
2010 10 Pur 23 34

2010 4 D 27 30 ACC-Q
2010 NA UVa 27 19

2013 6 D 32 39 Reg - S
2013 9 Mich St 31 30

2013 6 D 29 37 2nd Rd
2013 NA Creigh 23 27


Games get out of control. Some losses are ugly. Ask WVa in Indianapolis -- or Mich State in 2013. And, as I remember, Louisville and Arizona broke open very close games at the end. It happens. FWIW, if Arizona doesn't make two impossible threes at the end of the first half (I mean, the defense was perfect), Duke wins the game by double-digits and probably goes to the FF in 2011.

GGLC
03-26-2014, 07:47 AM
I don't want to repeat it all here but take a look at post 66 in the Mercer post-game thread. That summarizes my thoughts on the issues - playing style-wise which is a direct function of coaching - that lead to poor tournament game performances - especially in the second half


Again, I really think is missing the point. I think you can and SHOULD single out losses because ALL of our recent tournament losses have at least one significant thing in common - we get beat very badly in the second half. Look at the numbers. The margins are astonishing. So many people here like to rely on KenPom and it's predictive value. Our second half performances in the 7 recent NCAA losses produced an average margin of -14. I'll bet that the KenPom predictive values would expect Duke to have won those halves by an average of about 3 points per half. That's a pretty incredible disparity from the expected value and enough of a sample size to make the numbers shocking. It's not like the average margin of second half loss was a one point per game - which would have been somewhat close to the predicted margin. It was about 17 points PER GAME WORSE. That's not something you can dismiss or joke away with a cute comment about us being outscored in all of our losses. We have lost BADLY in every NCAA loss for the past seven tournaments. You know what makes the point even stronger - in 5 of the 7 losses we led at the half. And the two we didn't - Louisville and Villanova - we were down by only 3 -- and then proceeded to get blown out in the second half. Do you not think it's remarkable that we've had such a disparity in second half performance in 7 straight NCAA tournament losses? You know what game also fits this pattern? Belmont in 2008 - won the first half by 7 and lost the second half by 6. We were extremely close to having 4 first round losses to double digit seeded teams in the last 7 years. We can make jokes about other teams scoring more points than us in all our losses. But they're not just outscoring us - they're killing us in the second half and in most cases they are huge underdogs - like double digit underdogs - that are killing us in the second half. I'm not even going to try to address the talent disparity Duke has had in most of these games because that will open up a separate can of worms. But anyone who thinks Duke's pattern of losing in the NCAA tournament during the past 8 years is not incredibly remarkable and disturbing is not facing reality. Two more games for reference - remember the first game of the 2011-12 season? - in Cameron - Maui Invitational - against Belmont again. Duke ranked #6 with at least 4 future NBA players playing significant minutes - what happened? Won the first half by 9. Lost the second half by 8. How about Vermont? - of course we remember that game from early this season. Duke again coincidentally ranked #6 - in Cameron - won the first half by 8. Lost the second half by 7.

Look, I am as big of a Duke fan as anyone here. But there is something systemic causing these types of losses and near-losses - all of which involve major first and second half margin disparities - to teams that we would expect to beat with relative ease.

I'll also point out the following:

2002. First half. Duke 42 IU 29. Second half. IU 45 - Duke 31
2004. First half. Duke 41 UCONN 34. Second half. UCONN 45. Duke. 37
2005. First half. Duke 32. MSU 32. Second half. MSU. 46. Duke 36

Thanks very much for your excellent post. I think sage and Kedsy are being a little too quick to snark and as a result are talking past what you're saying a bit. I would be interested in everyone's responses to the Pomeroy point made above -- there's something going on with Duke's second-half performances in games where Duke "should" be in a position to do at least as well in the second half as it did in the first. We're Duke; between us and Mercer (or VCU, or Lehigh, or Wake, or...) we're not supposed to be the one running out of gas down the stretch. Not with our guys. And not over a period of years.

Kedsy
03-26-2014, 10:18 AM
Again, I really think is missing the point. I think you can and SHOULD single out losses because ALL of our recent tournament losses have at least one significant thing in common - we get beat very badly in the second half.

Well, first of all, in only three of your seven games did we get beat "badly" in the second half -- if you're losing at the end of the game you foul. If the other team makes their free throws, losing by 4 (VCU) or 7/8 (Lehigh/Mercer) or even 11 (WVa) is not a blowout. I haven't looked it up, but what was the 2nd half score in those games before we started fouling? My guess is most or all of those second halves were moderately close.

More importantly, the reason you can't single out losses is all teams play worse when they lose, not just Duke. I bet if you looked up every NCAA tournament loss since since 2007 (and I'm not doing this research either, at least right now), you'd find the vast majority of the losing teams lost by at least 7 in the second half, and probably a similar percentage (3 out of 7) got blown out in the second half. It's the nature of losses. If you had that research available, and it showed other teams typically won or played close in the second half of their losses, but Duke was different, then I'd agree you might be on to something. But without that sort of backup, you can't cherry pick 7 games and reasonably say they equal a trend or are evidence of bad coaching.

CDu
03-26-2014, 10:54 AM
Well, first of all, in only three of your seven games did we get beat "badly" in the second half -- if you're losing at the end of the game you foul. If the other team makes their free throws, losing by 4 (VCU) or 7/8 (Lehigh/Mercer) or even 11 (WVa) is not a blowout. I haven't looked it up, but what was the 2nd half score in those games before we started fouling? My guess is most or all of those second halves were moderately close.

More importantly, the reason you can't single out losses is all teams play worse when they lose, not just Duke. I bet if you looked up every NCAA tournament loss since since 2007 (and I'm not doing this research either, at least right now), you'd find the vast majority of the losing teams lost by at least 7 in the second half, and probably a similar percentage (3 out of 7) got blown out in the second half. It's the nature of losses. If you had that research available, and it showed other teams typically won or played close in the second half of their losses, but Duke was different, then I'd agree you might be on to something. But without that sort of backup, you can't cherry pick 7 games and reasonably say they equal a trend or are evidence of bad coaching.

I agree with your overall point in general. But I thought I'd provide the data you requested for completeness:

We never intentionally fouled in the VCU game. It was a back-and-forth game with Maynor hitting the winning basket with 5 seconds left.

We didn'tstart intentionally fouling in the WVU game until under 2 minutes to go. By that point, we were already down 11 (meaning we had been outscored by 18 in the second half). We actually cut into their lead by fouling, outscoring them 12-7 down the stretch. So the fouling at the end actually served to lessen the damage.

We didn't start fouling Lehigh intentionally until under a minute to go, down 6 (outscored by 8 in the second half). We again cut into that lead (barely) by outscoring Lehigh 13-12 in the final minute.

We didn't start intentionally fouling Mercer until under a minute to go (down 5; down 6 for the second half). We then were outscored 10-8 over the last minute. So that game may have slightly magnified the difference. Although given that Mercer was scoring every time down court by that point anyway, I'm not sure that this argument would have held had we just played them straight up.

freshmanjs
03-26-2014, 11:05 AM
I'm not sure I completely agree with this. To put the point another way, excluding the 2010 championship run, in each NCAA tournament Duke has had a terrible (and in some cases monumental) second half collapse against a lower seeded team (exception being last year against Louisville). Here are the numbers:

2014: Mercer 44 Duke 36
2013: Louisville 50 Duke 31
2012: Lehigh 47 Duke 40
2011: Arizona 55 Duke 33
2009: Villanova 51 Duke 31
2008: West Virginia 44 Duke 33
2007: VCU 41 Duke 37

Some of those second half scores are big blowouts. The only games in which we were able to stay within 10 points (in a HALF) were against double digit seeds. The point is not really how we did overall in NCAA games - (by the way, we lost the second half to Michigan in 2011 (1 vs. 8 seeds) and Texas in 2009 (2 vs. 7 seeds) but only by 2 points in each game) - it is just the fact that a complete second half collapse has happened every year for a long stretch of time with a complete turnover of the roster. It's a very disturbing pattern which I think is attributable to the style of play and coaching.

i think you can do this with most teams. i mean, if you lose the game, you had to lose at least one of the halves. take michigan state and their terrible halves in tournament losses:

2013: -9 vs Duke (2nd)
2012: -8 vs louisville (2nd)
2011: -18 vs ucla (1st)
2010: no bad half vs butler
2009: -21 vs UNC (1st)
2008: -30 vs memphis (1st)
2007: -8 vs UNC (1st)

Kedsy
03-26-2014, 11:06 AM
I agree with your overall point in general. But I thought I'd provide the data you requested for completeness:

We never intentionally fouled in the VCU game. It was a back-and-forth game with Maynor hitting the winning basket with 5 seconds left.

We didn'tstart intentionally fouling in the WVU game until under 2 minutes to go. By that point, we were already down 11 (meaning we had been outscored by 18 in the second half). We actually cut into their lead by fouling, outscoring them 12-7 down the stretch. So the fouling at the end actually served to lessen the damage.

We didn't start fouling Lehigh intentionally until under a minute to go, down 6 (outscored by 8 in the second half). We again cut into that lead (barely) by outscoring Lehigh 13-12 in the final minute.

We didn't start intentionally fouling Mercer until under a minute to go (down 5; down 6 for the second half). We then were outscored 10-8 over the last minute. So that game may have slightly magnified the difference. Although given that Mercer was scoring every time down court by that point anyway, I'm not sure that this argument would have held had we just played them straight up.

Thanks. I guess West Virginia should slide into the blowout category then. But losing by 4 or 6 or 8 in the second half is not a blowout and thus don't really support the OP's point. Because even 4 out of 7 blowouts, looking only at games that you lose, doesn't show any horrifying trend.

GGLC
03-26-2014, 11:08 AM
Well, first of all, in only three of your seven games did we get beat "badly" in the second half -- if you're losing at the end of the game you foul. If the other team makes their free throws, losing by 4 (VCU) or 7/8 (Lehigh/Mercer) or even 11 (WVa) is not a blowout. I haven't looked it up, but what was the 2nd half score in those games before we started fouling? My guess is most or all of those second halves were moderately close.

More importantly, the reason you can't single out losses is all teams play worse when they lose, not just Duke. I bet if you looked up every NCAA tournament loss since since 2007 (and I'm not doing this research either, at least right now), you'd find the vast majority of the losing teams lost by at least 7 in the second half, and probably a similar percentage (3 out of 7) got blown out in the second half. It's the nature of losses. If you had that research available, and it showed other teams typically won or played close in the second half of their losses, but Duke was different, then I'd agree you might be on to something. But without that sort of backup, you can't cherry pick 7 games and reasonably say they equal a trend or are evidence of bad coaching.

Well, let's start with a self-evident statement: In 100 percent of basketball games, the losing team gives up more points than the winning team. But there are different shapes to those losses. 1) In some, the losing team loses both halves on point differential. 2) In some, the losing team is losing at halftime and outscores the other team in the second half, but not enough to erase the deficit. 3) And in some, the losing team is winning or tied at half time and gives up more points than the other team in the second half. Right?

Mr. synellinden's data purports to show that Duke's NCAA losses (and close games like Belmont) consistently fall within the third category, and not the first or second. pfrduke's data to begin this thread further isolates the problem during this season to the "fourth quarter" of losses and close games, the final ten minutes (see Wake, Maryland, Clemson, Notre Dame, Arizona, UNC I, Virginia I, etc.).

These may or may not be meaningful observations, but they are NOT rebutted by the pat response that obviously teams that lose will give up more points. Because that it not what they're addressing. It's not the fact of the loss...it's the SHAPE of the loss.

And for a program that reliably has a talent advantage on its opponents, seeing pervasive evidence that we tend to wilt down the stretch during our losses and close calls relative to the other teams we play (even less talented teams) is not something that can or should be hand-waved away with glib generalities.

What does it mean? That's a separate discussion. But you're addressing the wrong thing if your response is, well, obviously teams play worse when they lose. That's not the point of what the data demonstrate, and the data deserve to be engaged on the level at which synellinden and pfrduke intend, since they've done a considerable amount of work to look at the issue.

GGLC
03-26-2014, 11:12 AM
i think you can do this with most teams. i mean, if you lose the game, you had to lose at least one of the halves. take michigan state and their terrible halves in tournament losses:

2013: -9 vs Duke (2nd)
2012: -8 vs louisville (2nd)
2011: -18 vs ucla (1st)
2010: no bad half vs butler
2009: -21 vs UNC (1st)
2008: -30 vs memphis (1st)
2007: -8 vs UNC (1st)

Right, and you can look at that data and say that Michigan doesn't have a "second-half" problem in its losses.

By contrast, Duke absolutely does, whether it's meaningful or not.

bedeviled
03-26-2014, 11:15 AM
Ha! I was working on something else and, like other posters, decided to do the homework. It does appear that Duke consolidates its negative differential to the 2nd half, whereas other teams are equal opportunity tankers (combine problems in 1st and 2nd halves). Here's some data I collected:

Duke



Opponent
Duke 1H
Duke 2H
Opp 1H
Opp 2H
1H Diff
2H Diff


2014
Mercer
35
36
34
44
1
-8


2013
UL
32
31
35
50
-3
-19


2012
Lehigh
30
40
28
47
2
-7


2011
Arizona
44
33
38
55
6
-22


2010

Champion






2009
Nova
23
31
26
51
-3
-20


2008
WVU
34
33
29
44
5
-11


2007
VCU
40
37
38
41
2
-4


2006
LSU
27
27
31
31
-4
-4


2005
MSU
32
36
32
46
0
-10



Kansas



Opponent
KU 1H
KU 2H
Opp 1st
Opp 2nd
1H Diff
2H Diff


2014
Stanford
24
33
22
38
2
-5


2013
Michigan
40
36
34
42
6
-6


2012
Kentucky
27
32
41
26
-14
6


2011
VCU
41
34
27
30
14
4


2010
N Iowa
28
39
36
33
-8
6


2009
MSU
36
26
29
38
7
-12


2008

Champion






2007
UCLA
31
24
35
33
-4
-9


2006
Bradley
27
46
37
40
-10
6


2005
Bucknell
31
32
36
32
-5
0



Michigan State



Opponent
MSU 1H
MSU 2H
Opp 1st
Opp 2nd
1H Diff
2H Diff


2013
Duke
31
30
32
39
-1
-9


2012
UL
18
26
23
34
-5
-8


2011
UCLA
24
52
42
36
-18
16


2010
Butler
28
22
28
24
0
-2


2009
UNC
34
38
55
34
-21
4


2008
Memphis
20
54
50
42
-30
12


2007
UNC
33
34
41
33
-8
1


2006
Gmason
30
35
33
42
-3
-7


2005
UNC
38
33
33
54
5
-21



UNC



Opponent
UNC 1H
UNC 2H
Opp 1st
Opp 2nd
1H Diff
2H Diff


2014
ISU
37
46
40
45
-3
1


2013
Kansas
30
28
21
49
9
-21


2012
Kansas
47
20
47
33
0
-13


2011
Kentucky
30
39
38
38
-8
1


2010
Dayton
32
36
45
34
-13
2


2009

Champion






2008
Kansas
27
39
44
40
-17
-1


2007
Georgetown
50
31
44
37
6
-6


2006
Gmason
27
33
20
45
7
-12

Kedsy
03-26-2014, 11:40 AM
Mr. synellinden's data purports to show that Duke's NCAA losses (and close games like Belmont) consistently fall within the third category, and not the first or second. pfrduke's data to begin this thread further isolates the problem during this season to the "fourth quarter" of losses and close games, the final ten minutes (see Wake, Maryland, Clemson, Notre Dame, Arizona, UNC I, Virginia I, etc.).

I thought pfrduke's data was more revealing than Mr. S's data for several reasons: first, he showed that we gave away leads in the last 10 minutes, rather than looking at entire halves and I think the more narrow time frame gives the stat more credence; second, he confined it to one season, where a team's "personality" might be revealed by something like this, rather than over multiple years where the makeup of the team would be very, very different; and third, he looked at wins as well as losses, and compared the first three "quarters" to the fourth. Finally, pfrduke presented the data and didn't presume to make a conclusion as to why it happened or even if it was meaningful, in part because "17 games is not a huge sample size."

In contrast, Mr. S cherry-picked 7 games (or even 10 or 12, adding in games like the Belmont win in 2009 or the Indiana loss in 2002) over 7 years (or more, if you count, e.g., 2002), without looking at wins (and expressly excluding 2010), and without much if any context or comparison to anything (regular season losses, other team's performance in NCAA tournaments, how close was the 2nd half score before we started fouling, etc.), and then argued that his data showed a terrible trend which means we've had poor coaching. And I just don't think his data shows much of anything, other than teams that lose tend to get outscored.

Also, to your point, I think there are many more "shapes" of losses than three. For example, in some games, a team is close (or winning) at halftime and the opponent rips off 12 straight points to start the second half, then they play even for the rest of the half but the first team can't ever dent the lead significantly. In others, a team might be nursing a lead before giving up the last 12 points of the game. In others still, there is no big run -- Team A is up 2 at halftime, then goes down 2 with five minutes to play, the teams trade baskets until, with two minutes to play, Team B hits a three to go up 5, Team A misses a shot, and then has to foul when Team B goes into a stall. A bunch of made free throws later, combined with several missed desperation threes, and Team A "lost" the half by 12 points. Those three games show completely different things, but the score by halves is similar.

mr. synellinden
03-26-2014, 03:59 PM
I thought pfrduke's data was more revealing than Mr. S's data for several reasons: first, he showed that we gave away leads in the last 10 minutes, rather than looking at entire halves and I think the more narrow time frame gives the stat more credence; second, he confined it to one season, where a team's "personality" might be revealed by something like this, rather than over multiple years where the makeup of the team would be very, very different; and third, he looked at wins as well as losses, and compared the first three "quarters" to the fourth. Finally, pfrduke presented the data and didn't presume to make a conclusion as to why it happened or even if it was meaningful, in part because "17 games is not a huge sample size."

In contrast, Mr. S cherry-picked 7 games (or even 10 or 12, adding in games like the Belmont win in 2009 or the Indiana loss in 2002) over 7 years (or more, if you count, e.g., 2002), without looking at wins (and expressly excluding 2010), and without much if any context or comparison to anything (regular season losses, other team's performance in NCAA tournaments, how close was the 2nd half score before we started fouling, etc.), and then argued that his data showed a terrible trend which means we've had poor coaching. And I just don't think his data shows much of anything, other than teams that lose tend to get outscored.

Also, to your point, I think there are many more "shapes" of losses than three. For example, in some games, a team is close (or winning) at halftime and the opponent rips off 12 straight points to start the second half, then they play even for the rest of the half but the first team can't ever dent the lead significantly. In others, a team might be nursing a lead before giving up the last 12 points of the game. In others still, there is no big run -- Team A is up 2 at halftime, then goes down 2 with five minutes to play, the teams trade baskets until, with two minutes to play, Team B hits a three to go up 5, Team A misses a shot, and then has to foul when Team B goes into a stall. A bunch of made free throws later, combined with several missed desperation threes, and Team A "lost" the half by 12 points. Those three games show completely different things, but the score by halves is similar.

Hold on. I didn't cherry pick anything. We've only lost 7 NCAA games since 2007. There's only 1 loss per season in the NCAA tournament (unless of course ...). I analyzed them all for a several year period. I considered every game that could be considered. How is that cherry picking? You know what? We played terribly in every second half. Period. That's the point I'm trying to make. We significantly underperformed relative to KenPom expectations our even reasonable expectations in EVERY NCAA loss for 7 straight years. What other context would be helpful here? Do you not agree that it is pretty remarkable that in all 7 games we lost the second half by a substantial margin? And that it has happened three other times in the last 12 years? Of course you are more likely to lose the second half in losses. But to lose each and every second half by such a high margin? When three of them were double digit seeds? When we won the first half of 5 of the 7 (losing by 3 in the other 2). I don't think this is a statistical anomaly or the product of cherry picking. And I really don't understand the argument that this pattern can be explained away. Also, I only threw in references to the Vermont and Belmont regular season games because those were similar in two senses (1) they were games against teams that were similar to 14 or 15 seeds; and (2) in both cases we won the first half by several points and lost the second half by several points (and in neither one were we fouling at the end). I added in the references to Indiana, Michigan St. and UCONN because those were other fairly recent tournament losses that fit the PATTERN. And it is the PATTERN that I think is noteworthy.

Finally, I don't believe I said it was due to poor coaching. I said it was due to our playing style which is a function of coaching. I do believe it is a product of how Duke has been coached and I'll refer again to my post in the Mercer post-game thread for how that affects our second half performance practically.





Well, first of all, in only three of your seven games did we get beat "badly" in the second half -- if you're losing at the end of the game you foul. If the other team makes their free throws, losing by 4 (VCU) or 7/8 (Lehigh/Mercer) or even 11 (WVa) is not a blowout. I haven't looked it up, but what was the 2nd half score in those games before we started fouling? My guess is most or all of those second halves were moderately close.

More importantly, the reason you can't single out losses is all teams play worse when they lose, not just Duke. I bet if you looked up every NCAA tournament loss since since 2007 (and I'm not doing this research either, at least right now), you'd find the vast majority of the losing teams lost by at least 7 in the second half, and probably a similar percentage (3 out of 7) got blown out in the second half. It's the nature of losses. If you had that research available, and it showed other teams typically won or played close in the second half of their losses, but Duke was different, then I'd agree you might be on to something. But without that sort of backup, you can't cherry pick 7 games and reasonably say they equal a trend or are evidence of bad coaching.

If you don't think losing by 7 or 8 points in a half to a 14 or 15 seed constitutes losing badly, then I don't know what I can say. And I'm certain that losing by 19, 20 or 24 points in a half regardless of seeding is losing badly. I don't think any of those margins had anything to do with fouling at the end of games, and others have pointed to the actual facts of the way those losses played out.

GGLC
03-26-2014, 04:17 PM
Mr. synellinden eloquently said what I wanted to say.

It doesn't matter how many discrete shapes of losses there are; I focused on the three general categories that I did because the discussion centered around first-half perfomance vs second-half performances, and not anything more granular.

Can we at least agree that IF there's a persistent pattern of playing the first half close (or having a lead) and getting blown out the second half, then a more general reference to losing teams scoring fewer points than winning teams over the game as a whole is not responsive as a rejoinder to the point that is being made?

If team A consistently "loses" first halves and outscores their opponent in second halves, while team B consistently "wins" first halves and gets outscored during second halves (and team C has no particular pattern in which halves it's outscored in its losses), I think that's a valid observation and an interesting thing to discuss and speculate on. Waving it away by talking about how "teams that lose tend to get outscored" seems to be missing the point.

Duvall
03-26-2014, 05:01 PM
Hold on. I didn't cherry pick anything. We've only lost 7 NCAA games since 2007. There's only 1 loss per season in the NCAA tournament (unless of course ...). I analyzed them all for a several year period. I considered every game that could be considered. How is that cherry picking? You know what? We played terribly in every second half. Period. That's the point I'm trying to make. We significantly underperformed relative to KenPom expectations our even reasonable expectations in EVERY NCAA loss for 7 straight years. What other context would be helpful here? Do you not agree that it is pretty remarkable that in all 7 games we lost the second half by a substantial margin? And that it has happened three other times in the last 12 years? Of course you are more likely to lose the second half in losses. But to lose each and every second half by such a high margin? When three of them were double digit seeds? When we won the first half of 5 of the 7 (losing by 3 in the other 2). I don't think this is a statistical anomaly or the product of cherry picking. And I really don't understand the argument that this pattern can be explained away. Also, I only threw in references to the Vermont and Belmont regular season games because those were similar in two senses (1) they were games against teams that were similar to 14 or 15 seeds; and (2) in both cases we won the first half by several points and lost the second half by several points (and in neither one were we fouling at the end). I added in the references to Indiana, Michigan St. and UCONN because those were other fairly recent tournament losses that fit the PATTERN. And it is the PATTERN that I think is noteworthy.

Okay, but I still think it's incomplete to pick out seven losses and conclude that there's a PATTERN that is deeply meaningful. Shouldn't this PATTERN of weaker second halves also be showing up in Duke's NCAA wins? In March games before the NCAAs? Maybe it does, I don't know.

Kedsy
03-26-2014, 05:17 PM
Do you not agree that it is pretty remarkable that in all 7 games we lost the second half by a substantial margin? And that it has happened three other times in the last 12 years? Of course you are more likely to lose the second half in losses. But to lose each and every second half by such a high margin?

First of all, according to your data we lost to VCU by 4 points in the 2nd half, so even by your own definition we didn't lose "by a substantial margin" in "all 7 games." Second, no, I do not think losing a half by 7 or 8 points in a half to anybody constitutes "losing badly." If you lose at all, there's a decent chance something like that might happen. Finally, no, I don't think it's pretty remarkable.

It's seven games across eight years. Or 10 or 12 games over 13 years. It's not a pattern, or at least it's not a meaningful pattern. In large part because it doesn't take into account all the games over that time period when it didn't happen. Which is also why I still think it's cherry picking.

mr. synellinden
03-26-2014, 05:34 PM
I bet if you looked up every NCAA tournament loss since since 2007, you'd find the vast majority of the losing teams lost by at least 7 in the second half, and probably a similar percentage (3 out of 7) got blown out in the second half. It's the nature of losses. If you had that research available, and it showed other teams typically won or played close in the second half of their losses, but Duke was different, then I'd agree you might be on to something. But without that sort of backup, you can't cherry pick 7 games and reasonably say they equal a trend or are evidence of bad coaching.

Again - I did not cherry pick anything. And I didn't say it was bad coaching. I said it was due to how we are coached, which I happen to think leads to bad second half performances in NCAA tournament games. If you want to extrapolate that into bad coaching, I wouldn't disagree.

I just did the research for the 52 games that have been played so far in the NCAA tournament. Here are the results:

Average margin of second half loss for the losers = 4.4 points.

Percentage of losers who lost the second half by 7 or more = 35% (18/52). Not quite a vast majority

Percentage of losers who lost the second half by 10 or more = 21% (11/52) It it is worth pointing out that that only two of those (Villanova and Creighton) were the higher seeded team. What do both of those teams have in common? Perimeter oriented offenses that rely heavily on the 3 point shot.

Percentage of losers who were "blown out" in second half (I used 15 as an arbitrary number for a blowout in a half - even though our three "blowouts" were 19, 20 and 22 points - I'm happy to run the numbers with a different threshold) = 5.7% (3/52 - and two of those were 15 seed American and 16 seed Coastal Carolina. Oregon was the other.). Not quite close to 3/7.

Since all three of our second half "blowouts" were against relatively high seeds (1, 3, and 5) and were in the regionals. I looked at the sweet 16 games for the past five years. Here are the results:

2013 (did not include Duke - Lville)

LOSER 2nd H Margin

MSU -9
Ore +6
Kans -6
FGC -8
Fla -3
Miami -3
IU +1
Marq. -10
LaSalle +2
Arizona -7
Ohio St. +11

AVERAGE MARGIN = -2.4 points

2012

LOSER 2nd H Margin

Wisc +5
Cinc. -3
Syr -7
Ohio +7
NCSU -4
UNC -13
IU -9
Xavier +2
Baylor +8
MSU -8
Marq. -4
Fla. -12

AVERAGE MARGIN = -3.1 points


2011 (did not include Duke - Zona

LOSER 2nd H Margin

OSU -2
Marq +7
UNC +1
Wisc +2
BYU 0
Fla -1
Richm -1
FSU +5
Kans +4
SDSU +2
Ariz +5

(note - 7/11 losers WON the second half)

AVERAGE MARGIN = +2.0 points

2010 (included all Duke games)

LOSER 2nd H Margin

Cornell -1
Wash -15
Ky -5
NIowa -14
OSU -6
Tenn. -3
Purdue -12 (to Duke)
St. Mary's +6
Baylor -10 (to Duke)
Syr. +6
Xavier -1
KSt. 0

AVERAGE MARGIN = -4.6 points


2009 (did not include Duke - Nova)

LOSER 2nd H Margin

Xav -13
Pitt -4
Ariz -18
Kans -12
Louisv -9
Gonz -10
Syr 0
Okla -3
Purd -7
Memph +2
Missouri -1

AVERAGE MARGIN = -7.4 points


FIVE YEAR AVERAGE: = -3.1 points

A few things to note. There were 60 Sweet 16 games during the past five years. Of those 60, 14 teams lost the second half by 10 points or more (23%). 21/60 lost the second half by 7 or more (35%). Only 4 out of 60 lost the second half by more than 15 points - and 75% of those were Duke. In other words, in the 54 regional games during the past 5 years not involving duke only 1 team lost the second half by more than 15 points (1/54 = 1.8%). And Duke has been blown out in the second half in 50% of their regional games during the past 5 years. And Duke has ALL three of the largest second half losses in regional games during the past five years. And all three of those losses where when we were seeded #1 or #2.

mr. synellinden
03-26-2014, 05:44 PM
First of all, according to your data we lost to VCU by 4 points in the 2nd half, so even by your own definition we didn't lose "by a substantial margin" in "all 7 games." Second, no, I do not think losing a half by 7 or 8 points in a half to anybody constitutes "losing badly." If you lose at all, there's a decent chance something like that might happen. Finally, no, I don't think it's pretty remarkable.

It's seven games across eight years. Or 10 or 12 games over 13 years. It's not a pattern, or at least it's not a meaningful pattern. In large part because it doesn't take into account all the games over that time period when it didn't happen. Which is also why I still think it's cherry picking.

I think Coach K would consider losing a half by 7 or 8 points to a 14 or 15 seed, which we've done three times during the last 7 years (including Belmont) to be losing badly.

Second, I guess we're going to disagree that this is a meaningful pattern.

I don't think pointing out that the second half loss to VCU was only 4 points really contributes much to the debate. But yes, I agree that was not a substantial margin of second half loss.

CR9
03-26-2014, 05:49 PM
A few things to note. There were 60 Sweet 16 games during the past five years. Of those 60, 14 teams lost the second half by 10 points or more (23%). 21/60 lost the second half by 7 or more (35%). Only 4 out of 60 lost the second half by more than 15 points - and 75% of those were Duke. In other words, in the 54 regional games during the past 5 years not involving duke only 1 team lost the second half by more than 15 points (1/54 = 1.8%). And Duke has been blown out in the second half in 50% of their regional games during the past 5 years. And Duke has ALL three of the largest second half losses in regional games during the past five years. And all three of those losses where when we were seeded #1 or #2.

Wow.

mr. synellinden
03-26-2014, 06:46 PM
I'm going to offer one more piece of evidence.

Since 2007 we have played 15 NCAA tournament games (not including 2010 here for purposes of the discussion). Our record in those games is 8-7. With that kind of record you might expect our first and second half performances to kind of even out. Well, in the first half we are collectively +72 (average +4.8) and in the second half we are collectively -33 (average -2.2).

That's an average 7.0 swing from first half to second in ALL 15 NCAA games for the past seven years, 2010 not included. I think that's a meaningful pattern.

Since I know people will be curious, in 2010, we were +41 first half; +48 second half. That's about what I would expect for a national championship contender.

Kedsy
03-26-2014, 08:58 PM
I'm going to offer one more piece of evidence.

Since 2007 we have played 15 NCAA tournament games (not including 2010 here for purposes of the discussion). Our record in those games is 8-7. With that kind of record you might expect our first and second half performances to kind of even out. Well, in the first half we are collectively +72 (average +4.8) and in the second half we are collectively -33 (average -2.2).

That's an average 7.0 swing from first half to second in ALL 15 NCAA games for the past seven years, 2010 not included. I think that's a meaningful pattern.

Since I know people will be curious, in 2010, we were +41 first half; +48 second half. That's about what I would expect for a national championship contender.

What do you think was different about 2010 that merits excluding that year from the analysis?

Also, I apologize but I either missed or don't remember your post in the Mercer game thread that explains your theory on why our coaching strategy is prone to 2nd half collapses. So the following are serious questions, not meant to get a rise or anything like that.

-- Is this theory only applicable to NCAA tournament games, or would it apply to all games? If the latter, why are you confining your analysis to NCAAT losses?

-- Do you believe we used a different system prior to 7 years ago that wasn't prone to these collapses, or do you think Coach K's philosophy has always contained this flaw?

-- Does your theory explain why this only happens in losses and "bad wins" like this year's Vermont and Duke/Belmont in 2009? Why wouldn't the same thing happen in our other wins?

-- Does your theory distinguish between "losing" the 2nd half at different points in the half? By that, I mean since giving up a big run early in half feels different from giving up a big run in the middle of the half, which feels different from giving up a run at the end of the game, which in turn feels different from a back-and-forth game in which one team takes a small lead and ultimately pads it with free throws at the end, I wouldn't expect one theory to cover all those situations. Have you controlled for that?

Finally, I took a quick look at the halftime scores of every game Duke played this season. We only trailed at halftime in three games out of 35, and never trailed at halftime by more than 3 points (which coincidentally is the same maximum points behind as in the 7 NCAA games you tracked). Don't you think that if we never trail at halftime (or by no more than 3 points behind, anyway) then pretty much the only way we're going to lose at all is if we lose the second half? And since I assume you don't expect to win them all, why is it remarkable that we lose the second half in our losses?


I don't think pointing out that the second half loss to VCU was only 4 points really contributes much to the debate. But yes, I agree that was not a substantial margin of second half loss.

If you're claiming something happened every time, especially if "every time" is only 7 games, then the fact that it didn't happen once seems meaningful to me.

sagegrouse
03-26-2014, 09:04 PM
Hold on. I didn't cherry pick anything. We've only lost 7 NCAA games since 2007. There's only 1 loss per season in the NCAA tournament (unless of course ...). I analyzed them all for a several year period. I considered every game that could be considered. How is that cherry picking? You know what? We played terribly in every second half. Period. That's the point I'm trying to make. We significantly underperformed relative to KenPom expectations our even reasonable expectations in EVERY NCAA loss for 7 straight years. What other context would be helpful here? Do you not agree that it is pretty remarkable that in all 7 games we lost the second half by a substantial margin? And that it has happened three other times in the last 12 years? Of course you are more likely to lose the second half in losses. But to lose each and every second half by such a high margin? When three of them were double digit seeds? When we won the first half of 5 of the 7 (losing by 3 in the other 2). I don't think this is a statistical anomaly or the product of cherry picking. And I really don't understand the argument that this pattern can be explained away. Also, I only threw in references to the Vermont and Belmont regular season games because those were similar in two senses (1) they were games against teams that were similar to 14 or 15 seeds; and (2) in both cases we won the first half by several points and lost the second half by several points (and in neither one were we fouling at the end). I added in the references to Indiana, Michigan St. and UCONN because those were other fairly recent tournament losses that fit the PATTERN. And it is the PATTERN that I think is noteworthy.

Finally, I don't believe I said it was due to poor coaching. I said it was due to our playing style which is a function of coaching. I do believe it is a product of how Duke has been coached and I'll refer again to my post in the Mercer post-game thread for how that affects our second half performance practically.






If you don't think losing by 7 or 8 points in a half to a 14 or 15 seed constitutes losing badly, then I don't know what I can say. And I'm certain that losing by 19, 20 or 24 points in a half regardless of seeding is losing badly. I don't think any of those margins had anything to do with fouling at the end of games, and others have pointed to the actual facts of the way those losses played out.


Mr. synellinden eloquently said what I wanted to say.

It doesn't matter how many discrete shapes of losses there are; I focused on the three general categories that I did because the discussion centered around first-half perfomance vs second-half performances, and not anything more granular.

Can we at least agree that IF there's a persistent pattern of playing the first half close (or having a lead) and getting blown out the second half, then a more general reference to losing teams scoring fewer points than winning teams over the game as a whole is not responsive as a rejoinder to the point that is being made?

If team A consistently "loses" first halves and outscores their opponent in second halves, while team B consistently "wins" first halves and gets outscored during second halves (and team C has no particular pattern in which halves it's outscored in its losses), I think that's a valid observation and an interesting thing to discuss and speculate on. Waving it away by talking about how "teams that lose tend to get outscored" seems to be missing the point.

Sorry, guys. You can't just use losses. That practice causes a time-space warp and the statistical universe will explode and kill you both. There. Is that clearer than my previous ten posts on the same subject? LOOK AT ALL THE GAMES OF A CERTAIN TYPE AND NOT JUST LOSSES. Excuse me, I have to quiet down -- I am getting hoarse from all this yelling.

Sage

Newton_14
03-26-2014, 09:41 PM
Right, and you can look at that data and say that Michigan doesn't have a "second-half" problem in its losses.

By contrast, Duke absolutely does, whether it's meaningful or not.

But Michigan surely has a 1st half problem, whether it's meaningful or not...

mr. synellinden
03-26-2014, 09:56 PM
What do you think was different about 2010 that merits excluding that year from the analysis?

Also, I apologize but I either missed or don't remember your post in the Mercer game thread that explains your theory on why our coaching strategy is prone to 2nd half collapses. So the following are serious questions, not meant to get a rise or anything like that.

-- Is this theory only applicable to NCAA tournament games, or would it apply to all games? If the latter, why are you confining your analysis to NCAAT losses?

-- Do you believe we used a different system prior to 7 years ago that wasn't prone to these collapses, or do you think Coach K's philosophy has always contained this flaw?

-- Does your theory explain why this only happens in losses and "bad wins" like this year's Vermont and Duke/Belmont in 2009? Why wouldn't the same thing happen in our other wins?

-- Does your theory distinguish between "losing" the 2nd half at different points in the half? By that, I mean since giving up a big run early in half feels different from giving up a big run in the middle of the half, which feels different from giving up a run at the end of the game, which in turn feels different from a back-and-forth game in which one team takes a small lead and ultimately pads it with free throws at the end, I wouldn't expect one theory to cover all those situations. Have you controlled for that?

Finally, I took a quick look at the halftime scores of every game Duke played this season. We only trailed at halftime in three games out of 35, and never trailed at halftime by more than 3 points (which coincidentally is the same maximum points behind as in the 7 NCAA games you tracked). Don't you think that if we never trail at halftime (or by no more than 3 points behind, anyway) then pretty much the only way we're going to lose at all is if we lose the second half? And since I assume you don't expect to win them all, why is it remarkable that we lose the second half in our losses?

If you're claiming something happened every time, especially if "every time" is only 7 games, then the fact that it didn't happen once seems meaningful to me.

Even including 2010, there is a 4.5 point disparity between 1st and 2nd half performance. Is that statistically significant over 21 games?

To answer your questions relative to my other post - the main point of that post is that the second half of NCAA tournament games involve more mental and physical fatigue. My theory or argument is that what makes teams successful is an ability to get easy points when the game becomes more of an intense half court battle and there is more pressure and fatigue. Easy points include putbacks, layups and FTs. An offense that encourages a lot of 3 point shots does not allow for those easy points because FTs rarely result and offensive rebounds are usually longer. Of course, if you're taking a 3, you're not getting a layup. I think a major problem has been that since Duhon graduated we have not had a point guard that can break down a defense and create easy shots/layups and/or draw fouls and/or get teammates in a good position to draw a foul. While that is not necessarily a coaching strategy issue, I do think our offensive system encourages 3 pointers more than attacking the defense which is far more likely to result in easy points. In the Mercer game, we got into the bonus with about 11:35 left in the game. Do you know how many FTs we shot after that point? 5. And three of them were Tyler's after he was fouled on a 3 pointer. That means in 11:35 we drew exactly two fouls by attacking the basket. In that same period of time, we took 11 3 point shots - 12 if you include Tyler's. Yes, we made 4 of them and TT hit his FTs, but that is besides the point of whether there is an offensive strategy issue. Yes, I think this is a problem - and yes, I think it is a problem in all games, but it becomes magnified in the NCAA tournament when there is usually more pressure and fatigue. Why did the coaching staff not direct the team to attack the basket more in the last 11:35? When a team is in the bonus the other team can't play defense as aggressively because any foul is likely to result in points. Easy points. So attacking the defense is easier. Especially when there is a talent gap. Hood, Parker, Cook and Sulaimon all were capable of attacking, but they didn't. Maybe they were tired. Maybe they had just been coached too much to think they should take any open 3 pointer they could. In his press conference today Coach K talked about how you resort to bad habits when you're tired. Maybe that's what happens to Duke in the second half - they resort to habits (namely jump shooting and lack of aggressiveness) that are bad for NCAA tournament games.

I do think our strategy has changed during the last 10 years or so, but it may be due to the fact that we really have not had a true point guard that could penetrate and create opportunities for other players since Duhon. Kyrie was the best I've ever seen at Duke in that regard, and I think we would have won the title had he not been injured. I'm not including him because he played so few games and his role changed by the time he came back - Nolan was the point guard then.

I do happen to think this theory is applicable in all games, including wins but it is magnified in the NCAA tournament for the reasons mentioned above. I especially think it's relevant for road games where teams are likely to be more fatigued due to travel. Again, I don't think it's just an NCAA tournament issue, but I think the nature of the games in the NCAA tournament exposes this issue more.

I really don't understand what you mean about distinguishing losses in the second half and when a run may happen. But hopefully how I've explained what I think the issues are helps answer that even if you don't agree with the theory. I do think you should read that post in the Mercer thread (#66) because I explained it more coherently there.

Finally, I do think the fact that we rarely trail at the half but have a greater tendency to lose in the second half helps support the point I'm trying to make. I really don't know why you would try to make the point that we have to lose the second half sometimes. No we don't. We could win them all. Or at the same rate as the first half. If we win almost every first half, shouldn't we expect to win the same number of second halves? Doesn't that make the point? What am I missing? Isn't it a concern that we had 9 losses this year and only trailed at the half 3 times? How many second halves did we lose?

GGLC
03-26-2014, 10:03 PM
But Michigan surely has a 1st half problem, whether it's meaningful or not...

Sure. Don't disagree with that at all.

Kedsy
03-27-2014, 12:12 AM
Sure. Don't disagree with that at all.

Well, you should, because the original post was about Michigan State, not Michigan.


Even including 2010, there is a 4.5 point disparity between 1st and 2nd half performance. Is that statistically significant over 21 games?

I don't know, but your analysis would be robust if you included all the games and not just the ones that prove your point. If it's still significant including 2010, then why exclude that year at all?



To answer your questions relative to my other post - the main point of that post is that the second half of NCAA tournament games involve more mental and physical fatigue. My theory or argument is that what makes teams successful is an ability to get easy points when the game becomes more of an intense half court battle and there is more pressure and fatigue. Easy points include putbacks, layups and FTs. An offense that encourages a lot of 3 point shots does not allow for those easy points because FTs rarely result and offensive rebounds are usually longer. Of course, if you're taking a 3, you're not getting a layup. I think a major problem has been that since Duhon graduated we have not had a point guard that can break down a defense and create easy shots/layups and/or draw fouls and/or get teammates in a good position to draw a foul. While that is not necessarily a coaching strategy issue, I do think our offensive system encourages 3 pointers more than attacking the defense which is far more likely to result in easy points.

When I get the chance (probably not for a couple days), I will search out post #66 in the Mercer thread and read it. In the mean time, I did this little bit of research:



Year 3pt pt % NCAAT 2nd half loss margin
2014 34.5 -8
2013 29.2 -19
2012 31.3 -7
2010 29.3 NONE (last loss: -5)
2009 27.9 -20
2008 32.7 -11
2007 25.6 -4
2006 28.1 -4
2005 33.7 -10
2004 26.7 -8
2003 27.3 -4
2002 29.0 -14
2001 34.5 NONE (last loss: -18)
2000 28.5 -2
1999 24.5 -7
1998 26.6 -22
1997 31.7 -15
1996 29.7 -15
1994 19.6 -3
1993 20.0 +5
1992 16.2 NONE (last loss: -6)
1991 15.4 NONE (last loss: -9)
1990 13.8 -18
1989 14.0 -22
1988 17.2 +4
1987 17.4 +4


3-pt point % is the percentage of points from 3-pointers out of the team's total points. First thing I noticed is we've lost the 2nd half in every NCAAT loss since 1994. Second thing I noticed is our 2013-14 team tied the team record for reliance on points from threes, so maybe your theory might help explain our late woes this season. But after that, there doesn't seem to be a lot of correspondence. The other team tied for Duke's most reliant on the three was the 2001 national champs, and obviously that team didn't fall prey to the dreaded second half collapse. The 2011 and 2013 teams that each got pommeled in the 2nd half of their last games relied on the three just a little bit less than the 2010 national champs, while the 1998 team that relied on the three a lot less than any of them lost their last second half by 22 points. And the Duke teams with the least reliance on the three, the 1989 and 1990 teams, lost their last second halves by 22 and 18 points, respectively. So while your theory is intriguing, I'm not sure the data entirely supports it.



I do think our strategy has changed during the last 10 years or so, but it may be due to the fact that we really have not had a true point guard that could penetrate and create opportunities for other players since Duhon.

Actually, looking at the data above, it appears the strategy changed in 1996, after Coach K returned from his season-ending injury. It therefore appears we used the same rely-on-the-three strategy while Chris Duhon, Jason Williams, and Will Avery ran the team. In fact, while JWill and Duhon manned the backcourt together, in the 2001 season, we jacked up more threes than any other Duke team.



I really don't understand what you mean about distinguishing losses in the second half and when a run may happen. But hopefully how I've explained what I think the issues are helps answer that even if you don't agree with the theory. I do think you should read that post in the Mercer thread (#66) because I explained it more coherently there.

What I mean is, for example, if we lost the second half because the other team opened the half with a 15-0 run, then your theory doesn't make as much sense. Why would we be more fatigued right after halftime than we were at the end of the first half. The second half is pretty long, a team can "lose" it at many different points and in many different ways. Your theory would only explain losing it in certain ways at certain times.



Finally, I do think the fact that we rarely trail at the half but have a greater tendency to lose in the second half helps support the point I'm trying to make. I really don't know why you would try to make the point that we have to lose the second half sometimes. No we don't. We could win them all. Or at the same rate as the first half. If we win almost every first half, shouldn't we expect to win the same number of second halves? Doesn't that make the point? What am I missing? Isn't it a concern that we had 9 losses this year and only trailed at the half 3 times? How many second halves did we lose?

We'll have to just disagree on this. It's unreasonable to expect a team to win every game.

Put another way, Duke has been outscored in the 2nd half of its final loss of the season for the last 21 years, and in 25 of the last 26. I guess that's a trend, but I still don't think it's meaningful. Whether you use 21 years or 26 years, we've had an awful lot of different styles as well as an awful lot of success, but in all those years we still "lost" the 2nd half of our final loss. And how does getting clobbered in our last loss of 1989 or 1998 relate to 2011 or 2013? I repeat that I think your theory is interesting, but I also think we need a lot more data before we have any idea whether it's true or not.

bedeviled
03-27-2014, 03:06 AM
Looks like the convo has shifted from 2014 to a full 7-year itch. I'm not excited to participate, but I did some work as the thread developed, so I thought I'd share in case Mr. Synellinden wants to explore further.

Mr. Synellinden, I’m okay with your choosing lost tournament games for the data set...it just severely limits what the data can say. (I should note, however, that I don't recall what the rationale was for choosing 2007 as the start date). It must be clear that any conjectures are constrained by:
A) when Duke plays (with no reference or comparison to other teams/programs)
B) within the NCAA tournament
C) when Duke loses
D) those games have a systematic pattern (the presence of a systematic pattern would be what makes it possible to investigate across years, despite the changes in personnel....thus allowing you to investigate Duke's style/System. But, with the other data constraints, you can't really say much about Duke's System as compared to any other).

For example, if the pattern was that, in the 7 games Duke lost in the tournament since 2007, Duke made zero 3FG attempts, that might be a pattern to explore for meaning (rules, style of play, personnel, etc) or predictive value (should I bet on 3P's made, will Andre Dawkins reach 1050 points,etc). It does NOT, necessarily, though:
A) indicate that taking zero 3FGs (the systematic pattern) is uniquely part of the Duke System. It may be a new NCAA rule or may otherwise be exhibited by other programs, thus having little meaning to Duke’s System specifically. (**I do note that, now, the thread has expanded to compare patterns to other teams**)
B) indicate that taking zero 3FGs (the systematic pattern) is the Duke System in general. During the regular season, Duke may have taken all their shots from 3pt land.
C) indicate that taking zero 3FGs (the systematic pattern) is associated with losing. Duke may also exhibit the same pattern in tournament wins. So, we may take zero 3FGs in all our losses, but that does not mean that taking 3FGs is associated with losing vs winning.

So, anyway, just because it is a systematic pattern, it does not mean that it can be equated with the Duke System. Nor can it be used to explain or predict losing (all your data points are losing data points!). From the Mr. Synellinden subset, the only pronouncement that can be made is that "when we lose in the tournament, the games have X pattern."

That being said, spying and investigating a pattern can be fruitful for entertainment and to give you ideas to look into further, so I’m down with it. Once you notice the pattern, then you can compare it to other teams to see if it is associated to the Duke System in particular. You can look for it in won games to see if it has any explanatory or predictive value for winning/losing. Etc. And, if you want to make a statement about the Duke System in general, you should include data from all games using the Duke System (or a fully representative subset). I agree with others that the current data set is very limited in derivable meaning….but you can test your theories in more appropriate data sets if you find something of interest.


Second, I guess we're going to disagree that this is a meaningful patternI addressed some of the limitations to meaning above. But, more importantly, I found it hard to know where to go in this inquiry because I'm not quite sure what your pattern is specifically as you spoke of variations of the scoring margin concern.

***** If the pattern is “Duke has a negative 2nd half scoring margin in NCAAT games they lose”…
Yes, we have negative 2nd half scoring margin in 7 out of 7 lost games since 2007. That’s a clear pattern.
But, Kentucky has negative 2nd half scoring margin in 5 out of 6 lost games since 2007.
UCLA has negative 2nd half scoring margin in 5 out of 5 lost games since 2007.
Louisville has negative 2nd half scoring margin in 5 out of 6 lost games since 2007.
Uconn and Florida both have negative 2nd half scoring margin in 3 out of 4 lost games since 2007.
So, I’m not sure this pattern will have much meaning.
You likely won’t be able to make comments about the Duke System versus other styles of play.
Nor will you be able to offer substantial insight into explanations or predictions of losing. After all, scoring margin and losing are pretty darn related; it’s to be expected.

***** If you are sad that "when Duke loses in the NCAAT, they lose big"...
Heck yeah we do! Our average deficit in losses is 11.57.
But, MSU's is 9.86. That's also a 4 possession game.
UNC's deficit is 9. Louisville's is 8.17.
Oh, and UCLA's is 14.6!! Losers!

*+*+*+ If I had to bet money on it, I would say that the pattern you are espousing is “When Duke loses in the NCAAT, Duke significantly lost the 2nd half scoring battle”
Well, we do have a negative 2nd half scoring margin. AND, on average, it is worse than other schools. Maybe there’s something there.



1H Diff

2H Diff



Duke

1.43

-13.00



Kansas

0.43

-2.29



MSU

-11.86

2.00



UNC

-3.71

-5.29



UK

-6.14

-0.14



UCLA

-6.80

-7.80



UL

-5.50

-2.67



Uconn

-3.00

-3.00



UF

-2.50

-3.50



But, does this have any meaning? What are the cut-offs used and why were they chosen?
How bad or how different does a value have to be to indicate that Duke is unique?
How bad or how different does a value have to be to have an effect on winning/losing?
Is there something underlying the pattern or is it just random or otherwise uncontrollable events?

Note that this pattern does not imply that losing the 2nd half is the reason why Duke lost!!! It’s like saying, “When Duke loses in the NCAAT, Mr. Synellinden was not in the starting lineup.” Indeed, Duke could lose the 2nd half in all their games (games lost AND won), just like Mr. Synellinden is always out of the starting lineup. The data only covers games in which Duke has lost and, therefore, does not speak to winning vs losing – see the next bullet point pattern if that’s the pattern you are interested in.

It is understandable to think, “Well, we lost the 2nd half scoring battle in all the games we lost and, on average, have worse 2nd half margins than other schools. Maybe that is a ripe area for improvement. And, maybe, maybe, if we won that battle, we wouldn’t lose – though my data can’t speak to that right now.” And, if you did work out some method of meaningful cut-offs and tests for an association to winning/losing, I’m sure there would be an association….simply because points (scoring margin) are related to total points (winning/losing). As such, it really doesn’t seem very meaningful as it stands.

It would be best to look at variables other than points. You have ideas about *why* Duke has a poor 2nd half scoring margin in lost NCAAT games. You believe that Duke loses because they shoot 3FGs in the second half while the other team is taking advantage of layups, putbacks, and FTs, right? You can see if these things are true. If true, you can then look for them in other data sets. That will give you a better sense of whether or not they are factors associated with losing.

Well, if that’s what you’re interested in, then you can skip the whole rest of this post and head to my next one, where I give you data to play with.

***** If you want to make some kind of statement explaining or predicting losing, you need a data set that includes the won games, too, so that your conclusions can speak about losing vs winning.
Perhaps you are hoping for the pattern, “When Duke has a negative 2nd half scoring margin, they lose.”
10 out of 21 NCAAT games since 2007, we had a negative 2nd half scoring margin
Of those 10 games, we lost 7.
That doesn’t seem very promising or interesting to me. Isn’t that what one would expect?

***** Maybe you are struck by how soundly we are beaten in the second half. Well,
6 out of 21 NCAAT games since 2007, we were beaten by >2 possessions (7+pts) in the 2nd half
Of those 6 games, we lost 6.
Where's my shocked face? :O There it is! But, now I need my wink :) LOL It just makes sense. But, referencing a recent post of yours, maybe you are interested in how often it happens Duke compared to other teams? I don’t have that data, but ESPN, Statsheet, and BasketballReference do!

***** Maybe you are impressed by the thought that, ‘when we are beaten, we are BEATEN!’
6 out of the 10 times were outscored in the 2nd half, it was by >2 possessions (7+pts)

***** Sometimes, the things you bring up make me think that you are most concerned with the difference between 1st half performance and 2nd half performance.
13 out of 21 games, our 2nd half margin was worse than the 1st half
Of those 13 games, we lost 7.
Seems pretty random; no clearly overwhelming pattern.

***** How about talking about how soundly bad the 2nd half was compared to the 1st?
10 out of 21 games, our 2nd half margin was worse than the 1st half margin by >2 possessions (7+pts)
Of those 10 games, we lost 6.
IDK. Is that exciting?

***** What about, “when we drop off in the 2nd half, we DROP OFF IN THE 2nd HALF!”
10 out of the 13 times our 2nd half margin was worse than the 1st half margin, it was by >2 possessions (7+ pts)

These are all related to the things you’ve been saying. But, I keep thinking that you are chiefly just describing things rather than defining and testing them. Thus, I’m not really sure what pattern you are most impressed by. I agree that the scoring margin events appear remarkable (and frustrating), but I'm not sure they are objectively remarkable…or can tell us anything particularly meaningful. But, I suppose it depends on what your pattern is and what questions you ask about it. I've worn myself out with it, so I'm just going to pass you my data in my next post :)

bedeviled
03-27-2014, 03:19 AM
Here's my raw data:





1st Half




2nd Half




1st Half





2nd Half









In Paint
Off T/O
2nd Chance
Fast Break
Bench
In Paint
Off T/O
2nd Chance
Fast Break
Bench
FG
FGA
3FG
3FGA
FT
FTA
FG
FGA
3FG
3FGA
FT
FTA


2014
Mercer
10
7
0
5
13
16
11
2
2
2
11
21
3
9
9
12
14
24
2
4
14
16


2014
Duke
6
3
14
5
8
4
4
9
0
15
12
34
8
21
3
4
10
28
7
16
9
9


2013
Louisville
22
6
4
4
5
20
3
12
2
15
13
28
0
5
9
13
16
27
2
8
16
23


2013
Duke
16
5
4
0
10
14
5
1
2
2
10
24
2
6
10
10
9
28
2
10
11
18


2012
Lehigh
12
8
0
8
0
16
5
8
6
2
10
26
2
7
6
11
12
26
4
11
19
26


2012
Duke
16
2
4
2
4
14
4
11
0
10
11
26
1
10
7
10
13
32
5
16
9
13


2011
Arizona
6
5
5
0
3
24
14
12
2
16
13
27
6
8
6
8
21
36
3
7
10
13


2011
Duke
16
12
11
0
18
6
5
2
0
21
18
34
3
8
5
6
9
24
2
6
13
16


2009
Nova
18
4
11
4
2
26
6
14
4
19
10
29
1
7
5
5
17
35
3
11
14
18


2009
Duke
6
2
2
3
4
12
4
5
2
7
7
25
2
11
7
10
9
35
3
16
10
14


2008
WVU
18
2
4
4
8
14
7
13
3
13
13
36
0
16
3
3
12
26
4
5
16
19


2008
Duke
14
6
0
0
12
10
3
5
0
10
10
25
2
9
12
12
9
25
3
13
12
20


2007
VCU
10
9
5
11
7
16
13
8
10
8
11
31
5
9
11
13
15
30
4
7
7
11


2007
Duke
20
5
9
4
7
20
9
10
2
0
14
26
1
2
11
16
13
32
2
9
9
16



And here's a pic of derived data:4034 and another: 4035
2FG% is FG% for only 2-pt FGs
eFG% is effective FG%
TS% is true shooting
%3P is the Percent of FG that were taken from 3P
FTA/FGA is free throw rate
%Pts 2P, %Pts 3P, and %Pts FT are the Percentage of total points gotten from 2P, 3P, and FT, respectively
%2P Paint is the Percentage of total 2P points that were gotten from inside the paint
%Pts Paint is the Percentage of total points that were gotten from inside the paint
%Pts PaintFT is the Percentage of total points that were gotten from (inside the paint + free throws)
Note, this is kind of a cool measure for what you are looking at. Unfortunately, the box score
stats don't list 'Missed FG in the Paint,' so the only calculations involving the Paint are for made FGs
%OR is the Percentage of available rebounds from the offensive glass that were obtained by the offensive team

bedeviled
03-27-2014, 03:21 AM
There's too much derived data to put here, but here are the averages for Duke and Opponents, by game half, for the 2007-present time period.



Opp1H
Opp2H
Duke1H
Duke2H


Pts In Paint
13.71
18.86
13.43
11.43


Off T/O
5.86
8.43
5.00
4.86


2nd Chance
4.14
9.86
6.29
6.14


Fast Break
5.14
4.14
2.00
0.86


Bench Pts
5.43
10.71
9.00
9.29


2FGA
19.57
21.57
18.14
16.86


2P Pts
18.29
24.29
18.00
13.71


3FGA
8.71
7.57
9.57
12.29


3P Pts
7.29
9.43
8.14
10.29


FTA
9.29
18.00
9.71
15.14


FT Pts
7.00
13.71
7.86
10.43


OR
5.00
7.14
5.57
6.14






Opp1H
Opp2H
Duke1H
Duke2H


2FG%
47.90
56.19
47.90
40.31


3FG%
29.54
45.52
29.90
27.48


eFG%
46.20
57.88
46.61
41.40


FT%
79.77
76.10
81.01
71.32


TS%
58.70
72.72
61.11
58.28


%3P
30.70
26.01
33.78
41.96


FTA/FGA
34.72
64.16
37.01
53.39


%Pts 2P
57.97
50.71
51.89
39.83


%Pts 3P
20.90
20.25
24.14
29.41


%Pts FT
21.13
29.04
23.96
30.76


%2P Paint
73.93
79.69
76.11
81.61


%Pts Paint
44.08
39.41
39.16
33.22


%Pts PaintFT
65.21
68.45
63.12
63.98


%OR
28.80
39.94
31.90
30.04

bedeviled
03-27-2014, 03:25 AM
And, finally, the information that I found most pleasing to the eye:
40364037




1H
2H
Opp 1H
Opp 2H


Duke
34.00
34.43
32.57
47.43


Kansas
32.43
32.00
32.00
34.29


MSU
26.86
36.57
38.71
34.57


UNC
36.14
34.14
39.86
39.43


UK
27.14
40.00
33.29
40.14


UCLA
29.40
35.80
36.20
43.60


UL
29.67
33.33
35.17
36.00


Uconn
33.00
32.00
36.00
35.00


UF
34.25
31.25
36.75
34.75


Phew, I'm glad to get rid of all this stuff. Knock yourself out.

Kedsy
03-27-2014, 07:01 AM
Phew, I'm glad to get rid of all this stuff. Knock yourself out.

Wow, thanks for all your work in this thread. Really amazing.

Also, I would add that teams that are behind tend to shoot more threes to try to cut into the lead, so even the fact that Duke attempted more threes in the 2nd half of games that we lost (apparent from one of your fabulous tables) doesn't necessarily prove that those are responsible for the loss.

I totally agree with you that this data is a lot less meaningful because we don't know if our wins in the period exhibited the same patterns we see in the losses. I also think that since (as I mentioned in my earlier post) we've been outscored in our final loss every year but one going all the way back to 1988 (the only exception being 1993), the eight year (minus one) time period seems arbitrary. I assume it was chosen originally because the period 2007 to 2014 (excluding 2010, which Mr. S explicitly did in his analysis) was arguably the worst 7 year period for NCAA tournament success since Coach K got here, and presumably Mr. S was looking for a pattern that might explain why that was.

bedeviled
03-27-2014, 09:04 AM
Also, I would add that teams that are behind tend to shoot more threes to try to cut into the leadYeppers. In addition, the FTA disparity is rather small and likely fully explained by the fact that trailing teams foul to extend the game....it doesn't seem to support that we were less aggressive at getting to the line. The reason I like the info in post #62 is that it looks apparent to me that the Duke offense in the 2nd half is comparable to the first half.