Friday, March 6, 2015

Shot Clock Fever!

It’s all the rage these days within college basketball punditry to advocate for a shorter shot clock. The logic, as best as I can tell, is that this will speed up the pace of play and as a result increase scoring. As a fan, I am 100% in favor of faster play and higher scoring. I wonder, though, whether decreasing the shot clock from 35 to 30 (or fewer) seconds would actually achieve these goals.

It is undeniable that the game has slowed down. The average number of possessions per game in 2002 was 69.5. This season it is 64.9, the lowest it’s been over those 14 seasons. That’s a 6.6% decrease. The average offensive efficiency, however, has remained pretty close to 101 points per 100 possessions in that time period. The past two seasons have actually seen a spike in offensive efficiency, but because of the declining pace points per 40 minutes have dropped from 70.13 in 2002 to 66.26 this season, a 5.5% drop.



Shortening the shot clock might seem like a good idea, then, but what if we consider pace in terms of average possession length? Again, possession length has inarguably grown in the past 14 seasons, from 17.27 seconds in 2002 to 18.49 seconds this season. That’s a 7.09% increase.



And yet, 18.49 seconds is nowhere near 35 seconds.



Nevertheless, decreasing the shot clock from 35 to 30 seconds would potentially add up to 4 possessions per team per game. According to data from hoop-math.com, 10.6% of all FGAs in 2015 occur in the last five seconds of the shot clock. Setting aside turnovers that occur between 30 and 35 seconds on the shot clock, and assuming that all these FGAs occur at 35 seconds, the average possession length outside of those 10.6% possessions is 16.53 seconds. If we cut five seconds off the shot clock, and then assume that these 10.6% of FGAs would occur at 30 seconds, we get an extra 68.79 seconds. That makes for 4.16 extra possessions at 16.53 seconds/possession.

There are, however, turnovers late in the shot clock; not all late shot clock attempts occur at 35 seconds; and reducing the shot clock by five seconds would not necessarily mean these late shot clock attempts would also occur five seconds earlier. In other words, reducing the shot clock from 35 to 30 seconds is unlikely to create four extra possessions per game. A more realistic number is somewhere between two and three additional possessions per game, enough to boost team scoring by two to three points per game.

A closer look at late shot clock offense, though, further undercuts any hope that a shorter shot clock would actually boost scoring. According to hoop-math.com data teams shoot lower FG% at the rim, on 2P jumpers, and on 3P jumpers 30+ seconds into the possession. Teams also take fewer shots at the rim and more 2P and 3P jumpers 30+ seconds into the possession. This means that teams are taking more difficult, lower percentage shots late in the shot clock, which suggests that long possessions are due more to good defense than to intentionally slow offense. Teams score about 1.005 points/FGA on shots taken before 30 seconds left on the shot clock but only 0.847 points/FGA on shots late in the clock. That’s a decrease in scoring of 15.7%.


%total FGAs
eFG%
% Shots at rim
FG% at rim
% Shots 2pt Jumpers
FG% 2pt Jumpers
% Shots 3pt
FG% 3pt
Total
100.0%
49.4%
35.8%
58.6%
30.0%
35.7%
34.2%
34.5%
30 Seconds into Possession
10.6%
42.4%
24.7%
51.2%
36.7%
32.1%
38.6%
31.0%
Pre-30 Seconds
89.4%
50.2%
37.1%
59.2%
29.2%
36.3%
33.7%
34.9%


Shortening the shot clock from 35 to 30 seconds would only affect these late-shot clock shots. Teams would have even less to time to create high percentage shots. Desperation looks would now happen between 25 and 30 seconds as opposed to between 30 and 35 seconds.  Late-shot clock attempts would increase and likely drive down overall shooting percentages. Rather than increasing scoring, a shorter shot clock coupled with no other rule changes to encourage scoring would probably drive scoring down.


Davidson and Wisconsin

Fran Fraschilla suggested that Bob McKillop and Bo Ryan could get good looks for their teams with an 8-second shot clock. Returning to the goal of increasing scoring, Davidson and Wisconsin offer decent examples of the disconnect between pace and scoring.

Davidson has an average offensive possession length of 17.8 seconds this season, as do seven other teams. Only one of those teams is scoring more points per 40 minutes than Davidson, while the rest are scoring nine to 18 fewer points.

Team
Avg. Poss Length
PTs per 40/min
Duke
17.8
80.6145
Davidson
17.8
78.4718
Georgetown
17.8
69.9622
Illinois
17.8
69.3315
Portland
17.8
68.6724
Memphis
17.8
66.6496
Northern Illinois
17.8
64.7065
Western Illinois
17.8
60.997

Meanwhile, Wisconsin is scoring 71.34 points per 40 minutes. Of the 15 D1 teams scoring between 71 and 72 points/40, Wisconsin has the longest average offensive possession at 21.5. That means Wisconsin is the most efficient team in the bunch even though they are all scoring roughly the same amount of points.

Team
Avg. Poss Length
PTs per 40/min
Wisconsin
21.5
71.3415
Utah
20.4
71.278
Oakland
18.6
71.4875
Stanford
18.5
71.9415
Charleston Southern
18.1
71.28
UNC Asheville
18
71.2068
William & Mary
17.9
71.7416
South Dakota St.
17.9
71.412
Charlotte
17.2
71.3944
Kansas
16.8
71.6205
Oklahoma
16.5
71.7516
Hawaii
16.5
71.1018
East Tennessee St.
16.4
71.5666
UT Arlington
16.3
71.2448
Coppin St.
15.8
71.854


Fraschilla encouraged coaches to “get creative,” but I suspect the number one factor in Davidson’s and Wisconsin’s ability to score efficiently is the quality of their players not the creativity of their coaches. Tinkering with the shot clock certainly won’t change that.


Wrapping Up

Points scored per 40 minutes is strongly correlated with offensive efficiency.



Points scored per 40 minutes is not nearly as strongly correlated with offensive possession length.




Offensive efficiency decreases late in the shot clock. Shortening the shot clock would likely decrease overall offensive efficiency which in turn would decrease scoring. If people want more scoring then they should advocate for changes to increase shooting percentages and free throw attempts: move the 3P line in, eliminate the charge call, crack down on defensive contact (hand checks, bumping cutters, and the like), and possibly even eliminate the three-second rule.

Thursday, February 26, 2015

Chris Collins, Dutch Lonborg, and a Bunch of Really Old Dudes You Probably Never Heard Of


With last night's win over Indiana, Chris Collins became the 8th Northwestern coach and the 2015 squad became the 18th NU team to string together at least four consecutive B1G wins. Some folks like to make “tallest midget” jokes about real achievements in Wildcat basketball history. Not me. This is a savory moment. With three regular season conference games left, this group has a shot (2.1% per current KenPom data) to match Dutch Lonborg’s 7-game record conference winning streak. The CBI is now within reach, and the NIT is not impossible. Enjoy.


Season
Coach
B1G Winning Streak
B1G Place
Overall Record
B1G Record
*1931
Arthur Lonborg
7
1
16-1
11-1
1932
Arthur Lonborg
7
t2
13-5
9-3
1933
Arthur Lonborg
7
t1
15-4
10-2
1916
Fred Murphy
5
t2
14-5
9-3
1929
Arthur Lonborg
5
4
12-5
7-5
1943
Arthur Lonborg
5
3
8-9
7-5
1946
Arthur Lonborg
5
t3
15-5
8-4
1954
Waldo Fisher
5
t5
9-13
6-8
1966
Larry Glass
5
t5
12-12
7-7
1913
Dennis Grady
4
2
14-4
7-2
1918
Norman Elliott
4
3
6-4
5-3
1934
Arthur Lonborg
4
t2
11-8
8-4
1938
Arthur Lonborg
4
t3
10-10
7-5
1939
Arthur Lonborg
4
6
7-13
5-7
1944
Arthur Lonborg
4
t4
12-7
8-4
1959
William Rohr
4
t2
15-7
8-6
1967
Larry Glass
4
t7
11-11
7-7
**2015
Chris Collins
4
t10
14-14
5-10
*also had a 4-game B1G winning streak
**current