Analytics

[sedz]

A spot to move analytics discussions that go off the rails.

[sedz]

What about double screens at the top of the key? I recently saw a pro game (Charlotte and Orlando) where Charlotte ran a lot of double screens at the top of the key. By running the ball handler around 2 screens, that action created open shots on the wings and opportunities at the rim when the players setting the screens broke to the rim. The double screen made it difficult for the defender to get around 2 bigs. Sometimes one of the screeners broke to the rim. In other cases, both players that set the screens broke to the rim. It was an interesting display of different possibilities on offense.

Do not expect UC to do anything differently from what it has been doing. That is the sad part.

Yeah, double screens at the top are easy to do out of a horns set, which is extremely common in the pros. It's really hard to defend the paint and both corners out of that. Drive and kick to the corner is still the most efficient action for most NBA teams.

Still, the most important lesson from that chart is that ball screens are out of date in modern college basketball. Only three teams score more using ball screens than off ball actions. We are at the bottom, and we should just stop. It's not working.

[leeraymond372@gmail.com]

Minor changes in screen placement or timing can have a major effect. We saw that in the second half of the Clemson game. Alabama either sets their screens on the elbows/wings, allowing the guard to come off and then downhill through a vacant lane, or they run actions on the wings before the ball screen, which gets the defense moving side to side and allows the guard to choose whether to accept or reject the screen based on the defender's momentum. We run a lot of stagnant ball screens right at the top of the key. So the ballhandler ends up coming off toward the elbow rather than the lane, and the defenders are balanced and ready to defend either direction. It's the 2000-2010 style of ball screen. This team would be so much better if we just updated our schemes.

Another thing to notice from that chart is almost every team (except Alabama, Stanford, and Georgia) is much more efficient scoring off the ball. The axes are centered at 1.20 points per possession off ball and 1.00 points per possession on ball. We score around 1.13 off ball compared to 0.7 on ball. If we can't execute ball screens well, we should abandon them. If we had any analytics guys on the staff, we would have done that a long time ago. We abandoned it against Clemson for about 15 minutes, but no one seems to have identified that.

What about double screens at the top of the key? I recently saw a pro game (Charlotte and Orlando) where Charlotte ran a lot of double screens at the top of the key. By running the ball handler around 2 screens, that action created open shots on the wings and opportunities at the rim when the players setting the screens broke to the rim. The double screen made it difficult for the defender to get around 2 bigs. Sometimes one of the screeners broke to the rim. In other cases, both players that set the screens broke to the rim. It was an interesting display of different possibilities on offense.

Do not expect UC to do anything differently from what it has been doing. That is the sad part.

[Oldtimer_UC_fan]

Sedz, I'd like to know what your basketball background is?

[sedz]

Minor changes in screen placement or timing can have a major effect. We saw that in the second half of the Clemson game. Alabama either sets their screens on the elbows/wings, allowing the guard to come off and then downhill through a vacant lane, or they run actions on the wings before the ball screen, which gets the defense moving side to side and allows the guard to choose whether to accept or reject the screen based on the defender's momentum. We run a lot of stagnant ball screens right at the top of the key. So the ballhandler ends up coming off toward the elbow rather than the lane, and the defenders are balanced and ready to defend either direction. It's the 2000-2010 style of ball screen. This team would be so much better if we just updated our schemes.

Another thing to notice from that chart is almost every team (except Alabama, Stanford, and Georgia) is much more efficient scoring off the ball. The axes are centered at 1.20 points per possession off ball and 1.00 points per possession on ball. We score around 1.13 off ball compared to 0.7 on ball. If we can't execute ball screens well, we should abandon them. If we had any analytics guys on the staff, we would have done that a long time ago. We abandoned it against Clemson for about 15 minutes, but no one seems to have identified that.

[leeraymond372@gmail.com]

We are the worst power conference team in the country at scoring off the dribble. Only 0.70 points per possession against top 100 opponents. At the top is Alabama, scoring more than 1.3 points per possession off the dribble, nearly twice as efficient as us. Their screens are designed for the ballhandler to get downhill all the way to the rim. No pull up jumpers.

https://x.com/jgtrends/status/2009682327341863024

Yeah. You can see that in how UC plays. I wonder what are the team objectives when UC practices. If you practice like a champion, you will play like a champion. If that is the case, what can be said about the practice habits of a last place team?. Surely UC is not concerned with getting to the rim.

[Bearcat1996]

We are the worst power conference team in the country at scoring off the dribble. Only 0.70 points per possession against top 100 opponents. At the top is Alabama, scoring more than 1.3 points per possession off the dribble, nearly twice as efficient as us. Their screens are designed for the ballhandler to get downhill all the way to the rim. No pull up jumpers.

https://x.com/jgtrends/status/2009682327341863024

That is eye opening! Seems like an easy problem to fix, in my opinion.

[sedz]

We are the worst power conference team in the country at scoring off the dribble. Only 0.70 points per possession against top 100 opponents. At the top is Alabama, scoring more than 1.3 points per possession off the dribble, nearly twice as efficient as us. Their screens are designed for the ballhandler to get downhill all the way to the rim. No pull up jumpers.

https://x.com/jgtrends/status/2009682327341863024

[sedz]

If someone like sedz - no offense intended - has access to these as a college coach outsider, I would imagine CWM has access to this data and even more data. The question is does CWM use it, or use it correctly.

Haha, no offense taken. I only use free public data. Coaching staffs have access to that and can pay for much more. I don't think Wes uses analytics as a primary tool though. There are a couple of obvious clues - players with poor metrics have the highest usage and we run a midrange heavy offense. Brannen was supposed to be an analytics guy, but there is no way he was using metrics to evaluate players. Chris McNeal and Rapolas Ivanauskas had terrible metrics even before they came here. Tari Eason had the best metrics and he rode the bench.

Maybe Corey Evans will help.

[leo from jersey]

This is another astute observation by the great Yogi

1. "In theory there is no difference between theory and practice. In practice there is."

Do they still play as they practice?

[leo from jersey]

If someone like sedz - no offense intended - has access to these as a college coach outsider, I would imagine CWM has access to this data and even more data. The question is does CWM use it, or use it correctly.

I was posting tongue in cheek. One other view. Just having stats doesn't automatically = wins or makes one a better coach. There is a major human factor. Stat prediction often disappear in the heat of a game and a coach must, adapt, improvise and overcome.

: "It's tough to make predictions, especially about the future."
Yogi Berra

[London 'Cat]

Does CWM have access to those analytics? If he does/did he really has no excuses for further failure except for coaching or injuries.

If someone like sedz - no offense intended - has access to these as a college coach outsider, I would imagine CWM has access to this data and even more data. The question is does CWM use it, or use it correctly.

[leo from jersey]

I thought about doing that, but final AP is a better proxy for what most people judge a season on, which is tournament results. Plus this shows that algorithms are better than human voters at predicting their own poll. There's just so much information at our disposal, and computer models are now better able to sort through it than humans are.

Does CWM have access to those analytics? If he does/did he really has no excuses for further failure except for coaching or injuries.

[sedz]

Here is how Torvik's preseason rankings compared to the AP last season for all teams in either the AP's preseason or final rankings. Torvik was closer on 21, the AP was closer on 12.

Team	AP Pre	Torvik Pre	AP Final
Kansas	1	4	37
Alabama	2	9	6
UConn	3	13	29
Houston	4	1	2
Iowa St	5	3	17
Gonzaga	6	8	23
Duke	7	2	3
Baylor	8	15	--
UNC	9	5	--
Arizona	10	10	15
Auburn	11	7	4
Tennessee	12	6	5
TX A&M	13	12	19
Purdue	14	17	14
Creighton	15	14	30
Arkansas	16	35	20
Indiana	17	30	--
Marquette	18	25	38
Texas	19	11	--
Cincinnati	20	27	--
Florida	21	36	1
UCLA	22	24	32
Kentucky	23	23	12
Ole Miss	24	38	18
Rutgers	25	19	--

Team	AP Pre	Torvik Pre	AP Final
St Johns	27	16	11
Texas Tech	29	20	8
Michigan St	32	18	7
Michigan	34	34	10
BYU	35	28	13
Clemson	41	44	22
Wisconsin	45	40	16
St Marys	46	39	24
Louisville	47	62	21
Maryland	55	48	9
Memphis	--	32	25

Re compring AP and Torvik Pre, it might be interesting to consider Torvik final as well.

I thought about doing that, but final AP is a better proxy for what most people judge a season on, which is tournament results. Plus this shows that algorithms are better than human voters at predicting their own poll. There's just so much information at our disposal, and computer models are now better able to sort through it than humans are.

[sedz]

I'll be analyzing the transfers we're connected with and using BPM as the main measuring stick. Here's a quick and dirty grading scale:

10+: All-American, top 20 nationally
8-9: All Conference
6-7: Star Player
4-5: Major Rotation Player
2-3: Role Player or Specialist
1 or Below: Project

In the portal I'm mainly looking to bring in 4+ BPM players. Mitchell was our only guy in the Star Player category. It would be nice to get another star, or for Mitchell to increase his usage and move up to All Conference Caliber.

If you add up the BPMs for all five positions, you get the team's efficiency margin. Kenpom uses that for his NetRtg that he sorts teams by. Duke's was 39.3 this year, or an average 8 BPM per position. They are basically running out an all conference player at all 5 positions.

Our NetRtg this year was 14.9, or an average 3 BPM per position. We're running out a bunch of role players. We had one star in Mitchell, but we also had a bunch of minutes from 1 or below BPM players that brought the average down. If we can average 4 BPM at each position, that will push our NetRtg up to 20, which would comfortably be in the field. That should be our goal.