AI for Scouting Reports, Game Predictions, and Analysis

[RetroVol]

* If you ever need AI to produce a more accessible graphical representation of information, try prompting it with "and show me 3 ways that Edward R. Tufte might have graphed this information."

Brilliant. I encountered Tufte's work some years ago, but had forgotten that. I'm going to find a chance to try that prompt!

 

[RetroVol]

I don't know if this idea has any merit, but at this hour, after a night of interrupted sleep and before any morning coffee... it makes sense in my brain.

If I wanted a kind of statistical model against which to track the progress, strengths, and weaknesses of this season's team, I would want to build that model using data from CKC's teams at Glenville State. That's where we know the system was performing as-intended. And if those stats are retrievable, then they would also offer a longitudinal sample of paths and patterns of improvement from year-to-year, as well as within each season.

From the research I did when her hiring was announced, I remember being impressed that the system allowed her teams to win games many different ways. It seemed to me it would allow teams to break free from the "rock-paper-scissors" (tall vs. quick vs. good shooting) dimension of predictability within which basketball played below the rim gets stuck.

If those impressions are accurate (???) then a model derived from her Glenville State seasons would be complicated with multiple relationships between factors that generated success. I imagine it might produce an amoebic scatterplot* sweetspot--or maybe even several distinct sweetspots.

But I'm already way past my mathematical limits. My gray matter is good at seeing patterns, but I learned late in life that I have an actual defect thThat makes numerals and numbers "slippery" and probably why I always hated arithmetic and math. So I hope you can find some useful thought in this post.


* If you ever need AI to produce a more accessible graphical representation of information, try prompting it with "and show me 3 ways that Edward R. Tufte might have graphed this information."

Thanks for the thoughts. I've got the stats for Kim's GS teams in my Codex Lady Vols workspace. I'll be looking for new insights. I already know that her first GS team shot took 3s for 50% of their attempts, but that trended down over the next few years. Also, her teams won in November and December, usually by big margins, despite the idea floated in these forums that the system is too hard to learn. This was even true of her last team, the one following her Natty, that she had to rebuild almost entirely with transfers.

Maybe the best thing to do is to turn Codex loose on all the data from her GS, Marshall, and LV teams and let it see what it can find!

 

[Volfan2012]

I know there are at least a few forum participants who are interested in the way I've been using AI to generate Scouting Reports to extract predictions, determine the winner, analyze substitution patterns, etc. Others, I'm sure, have no interest whatsoever. I'm starting this thread as a place to describe what I'm doing, interesting things I run into, and for others to discuss both this project and the implications for the use of AI in general and, in particular, how the availability of AI may affect how we educate kids in the future. It may well be a good place for folks to question the use of AI, rail against it, etc., etc. Whatever. recognize this is a bit off topic, but, I will stay focused on my use of AI on Lady Vol topics. I think the best way to learn is to try and that discussions about specific attempts to use these tools are far better than generalities. Plus, maybe we'll come up with some additional feasible uses for Lady Vol topics. Anyway, may be a bit before I get the first substantive post up, assuming there isn't a mass demand that this thread get taken down, which, if there is, certainly won't hurt my feelings!

Regardless, I'll still generate the kind of content I've been posting and put it in the proper threads, unless, again, folks don't want it.

I would sure like to see a projection based on actual players minutes of their proposed statistical output for next years team where you could also draw a projection of how many points this team might average per game and our final totals projection in every statistical category. I did something like that, but it was based on every players 40 minute totals and I guess I could have taken a day or two to break it all down with my own idea of projected players minutes and what stats playing those actual minutes would look like.
What I am hoping for this team is something above 83 ppg and overall averages of 75 percent free throw, 35 percent three point and 52 percent two point. The percentage of three pointers taken around 40 percent and two point 60 percent would give our team a total of 70 shots per game. 28 threes per game and 10 made for 30 points. 42 twos with 22 made for 44 points and then to add something like 9 free throws made per game. That comes out to 44.5 percent overall average for the team. Team would need to also get to the line 12 times per game while averaging 9 makes.
This would probably be big winning numbers over 20 wins in a 30 game season with any kind of decent defense at all.
I know these are fairly lofty numbers, but to me seem reachable if every player matches their past performance. That certainly may be hard to do in the SEC.

What kind of efficiency would that look like over last season. Our three point percentage was 30.5 on 30.5 attempts per game. So I am asking for an improvement of 4.5 percent from the three even taking close to 3 less attempts. We averaged 37 two pointers per game last season and made 49.7 percent of the two point attempts. Looking to up that by 3 percent. Free throw percentage was 67 percent last season and we got to the line an average of 17 times and 11.7 made. Looking for a whopping 8 percent improvement. 67.5 shots per game and our overall average was 41 percent. Looking for three more shots per game while making 3.5 percent more attempts.
Last year we seemed to be fighting against ourselves taking 45 percent of our shots from the three and only making 30.5 percent. 55 percent from the two while making 49.7 percent so last year could've been more successful taking less threes and more twos based on percentages.

Team average 76 ppg last season I am asking for 83 on far better efficiency numbers. I feel confident that we can reach the 35 percent from the three next season. Far less confident that we can shoot 52 percent from the two and 75 from the line. The plus seven points on offense should certainly give us a couple more wins still beyond that always comes down to better defensive play probably also needing to lower defensive points allowed by seven as well.

 

[RetroVol]

I would sure like to see a projection based on actual players minutes of their proposed statistical output for next years team where you could also draw a projection of how many points this team might average per game and our final totals projection in every statistical category. I did something like that, but it was based on every players 40 minute totals and I guess I could have taken a day or two to break it all down with my own idea of projected players minutes and what stats playing those actual minutes would look like.
What I am hoping for this team is something above 83 ppg and overall averages of 75 percent free throw, 35 percent three point and 52 percent two point. The percentage of three pointers taken around 40 percent and two point 60 percent would give our team a total of 70 shots per game. 28 threes per game and 10 made for 30 points. 42 twos with 22 made for 44 points and then to add something like 9 free throws made per game. That comes out to 44.5 percent overall average for the team. Team would need to also get to the line 12 times per game while averaging 9 makes.
This would probably be big winning numbers over 20 wins in a 30 game season with any kind of decent defense at all.
I know these are fairly lofty numbers, but to me seem reachable if every player matches their past performance. That certainly may be hard to do in the SEC.

What kind of efficiency would that look like over last season. Our three point percentage was 30.5 on 30.5 attempts per game. So I am asking for an improvement of 4.5 percent from the three even taking close to 3 less attempts. We averaged 37 two pointers per game last season and made 49.7 percent of the two point attempts. Looking to up that by 3 percent. Free throw percentage was 67 percent last season and we got to the line an average of 17 times and 11.7 made. Looking for a whopping 8 percent improvement. 67.5 shots per game and our overall average was 41 percent. Looking for three more shots per game while making 3.5 percent more attempts.
Last year we seemed to be fighting against ourselves taking 45 percent of our shots from the three and only making 30.5 percent. 55 percent from the two while making 49.7 percent so last year could've been more successful taking less threes and more twos based on percentages.

Team average 76 ppg last season I am asking for 83 on far better efficiency numbers. I feel confident that we can reach the 35 percent from the three next season. Far less confident that we can shoot 52 percent from the two and 75 from the line. The plus seven points on offense should certainly give us a couple more wins still beyond that always comes down to better defensive play probably also needing to lower defensive points allowed by seven as well.

Is this what you're looking for? Way Too Early 2026-27 LV Basketball Prognostications.

 

[BruisedOrange]

Thanks for the thoughts. I've got the stats for Kim's GS teams in my Codex Lady Vols workspace. I'll be looking for new insights. I already know that her first GS team shot took 3s for 50% of their attempts, but that trended down over the next few years. Also, her teams won in November and December, usually by big margins, despite the idea floated in these forums that the system is too hard to learn. This was even true of her last team, the one following her Natty, that she had to rebuild almost entirely with transfers.

Maybe the best thing to do is to turn Codex loose on all the data from her GS, Marshall, and LV teams and let it see what it can find!

Just thinking aloud with my fingers... (...because i should be busy elsewhere, you can bet!)

- Her GS teams won so many of their games (only 24 losses over seven seasons) that losses alone might not offer a large enough sample to develop reliable trends. Might have to encourage Codex to look for some "special categories" to help round out that number. Maybe add to the losses any games they only won by 3 points or less, or games they won in overtime?

- Would it be good to prompt Codex to consider scores and stats per quarter rather than just game totals--given the desired effects from platooning to promote opponents' fatigue? At the end of how many 1st, 2nd, and 3rd quarters did they trail on the scoreboard, I wonder?

- Confessing my bias to see each game as a movie whose script is being written in real time, it follows that each quarter might be its own unique act, with different statistical aspects being up or down for each quarter, regardless of the eventual outcome.
For example: there may be one quarter filled with forced turnovers, another quarter loaded with offensive rebounds, another when GS's 3-point shooting was hot, and even a quarter where Glenville wasn't proficient offensively or defensively, but the opponent was too gassed to produce much on their end, either.

All of which is another way of realizing the challenge of finding a statistical sweetspot in a system that promotes winning via a number of ways.


- Another argument for assessing stats by quarter would be because of last season's signature mystery--those sluggish 3rd quarter starts!
That's the question I'd like to hear asked of CKC: "Do you think y'all figured out why that kept happening, and how to keep it from happening with this roster?"

 

[BruisedOrange]

I would sure like to see a projection based on actual players minutes of their proposed statistical output for next years team where you could also draw a projection of how many points this team might average per game and our final totals projection in every statistical category. I did something like that, but it was based on every players 40 minute totals and I guess I could have taken a day or two to break it all down with my own idea of projected players minutes and what stats playing those actual minutes would look like.
What I am hoping for this team is something above 83 ppg and overall averages of 75 percent free throw, 35 percent three point and 52 percent two point. The percentage of three pointers taken around 40 percent and two point 60 percent would give our team a total of 70 shots per game. 28 threes per game and 10 made for 30 points. 42 twos with 22 made for 44 points and then to add something like 9 free throws made per game. That comes out to 44.5 percent overall average for the team. Team would need to also get to the line 12 times per game while averaging 9 makes.
This would probably be big winning numbers over 20 wins in a 30 game season with any kind of decent defense at all.
I know these are fairly lofty numbers, but to me seem reachable if every player matches their past performance. That certainly may be hard to do in the SEC.

What kind of efficiency would that look like over last season. Our three point percentage was 30.5 on 30.5 attempts per game. So I am asking for an improvement of 4.5 percent from the three even taking close to 3 less attempts. We averaged 37 two pointers per game last season and made 49.7 percent of the two point attempts. Looking to up that by 3 percent. Free throw percentage was 67 percent last season and we got to the line an average of 17 times and 11.7 made. Looking for a whopping 8 percent improvement. 67.5 shots per game and our overall average was 41 percent. Looking for three more shots per game while making 3.5 percent more attempts.
Last year we seemed to be fighting against ourselves taking 45 percent of our shots from the three and only making 30.5 percent. 55 percent from the two while making 49.7 percent so last year could've been more successful taking less threes and more twos based on percentages.

Team average 76 ppg last season I am asking for 83 on far better efficiency numbers. I feel confident that we can reach the 35 percent from the three next season. Far less confident that we can shoot 52 percent from the two and 75 from the line. The plus seven points on offense should certainly give us a couple more wins still beyond that always comes down to better defensive play probably also needing to lower defensive points allowed by seven as well.

One relevant data point that could probably be gathered would be to compare the before & after shooting percentages (or proficiencies) of her transfers into Glenville State. And then the same in their second season in CKC's system.

We'd expect the numbers to go down, given the new pattern of playing really hard for short periods before coming out. But we might discover that was offset by the freedom that Kim encouraged to shoot good shots whenever you're open, regardless of the shot clock.

We didn't see as much of that last season as we did the season before--and, thus, her emphasis on recruiting "floor spacers" this year.

 

[LVFL_Soccer]

This is a brilliant thread.

 

[RetroVol]

One relevant data point that could probably be gathered would be to compare the before & after shooting percentages (or proficiencies) of her transfers into Glenville State. And then the same in their second season in CKC's system.

We'd expect the numbers to go down, given the new pattern of playing really hard for short periods before coming out. But we might discover that was offset by the freedom that Kim encouraged to shoot good shots whenever you're open, regardless of the shot clock.

We didn't see as much of that last season as we did the season before--and, thus, her emphasis on recruiting "floor spacers" this year.

I think these three posts might give some of what you're looking for: Post 1, Post 2, Post 3.

 

[volsagain]

This is a brilliant thread.

( Kinda ) MAD we don't have one for soccer

 

[RetroVol]

Just thinking aloud with my fingers... (...because i should be busy elsewhere, you can bet!)

- Her GS teams won so many of their games (only 24 losses over seven seasons) that losses alone might not offer a large enough sample to develop reliable trends. Might have to encourage Codex to look for some "special categories" to help round out that number. Maybe add to the losses any games they only won by 3 points or less, or games they won in overtime?

- Would it be good to prompt Codex to consider scores and stats per quarter rather than just game totals--given the desired effects from platooning to promote opponents' fatigue? At the end of how many 1st, 2nd, and 3rd quarters did they trail on the scoreboard, I wonder?

- Confessing my bias to see each game as a movie whose script is being written in real time, it follows that each quarter might be its own unique act, with different statistical aspects being up or down for each quarter, regardless of the eventual outcome.
For example: there may be one quarter filled with forced turnovers, another quarter loaded with offensive rebounds, another when GS's 3-point shooting was hot, and even a quarter where Glenville wasn't proficient offensively or defensively, but the opponent was too gassed to produce much on their end, either.

All of which is another way of realizing the challenge of finding a statistical sweetspot in a system that promotes winning via a number of ways.


- Another argument for assessing stats by quarter would be because of last season's signature mystery--those sluggish 3rd quarter starts!
That's the question I'd like to hear asked of CKC: "Do you think y'all figured out why that kept happening, and how to keep it from happening with this roster?"

I asked Codex if we had the quarter-by-quarter data in the workspace. We didn't, but we do now, and I did virtually nothing. Codex pulled the data, even though Glenville State's data pages made it a bit difficult. I only included 24-25 for the LVs. Last year was simply too much of an outlier to stick it in with all Kim's prior seasons. If this year looks like last year... well, I doubt we'll be talking about Kim except in the past tense. That said, using her GS teams, Marshal, and her first season with the LVs, here's what Codex came up with as a summary analysis:

I built the quarter-by-quarter dataset from official school boxscores and validated all 282/282 targeted games, excluding Tennessee 2025-26. There were 0 parse failures; 3 overtime games are kept with OT separate from Q4.

Main read: the data does not support a simple “Q1 blitz” story. Across the full sample, Q3 was the best offensive quarter at 24.55 points, while Q2 had the best scoring margin at +6.38. In losses, the first three quarters were all negative on average, while Q4 was roughly even, so losses look more like the pressure trade going bad early/mid-game than a late fade.

The stark difference in Kim's teams 3rd quarter performances before last year with what we saw last year suggests two things to me. First, those that wanted to point to this as evidence that Kim isn't a good coach and doesn't make good half-time adjustments need to find more supporting data. Second, it does fit with other evidence that there was a HUGE disconnect between Kim and many players on last year's team. I doubt we will ever get much insight into exactly what happened, but that does seem to be the case. And it does seem to be an anomaly. Last year was just different, and I hope it looks even more like an anomaly after this season.

 

[RetroVol]

* If you ever need AI to produce a more accessible graphical representation of information, try prompting it with "and show me 3 ways that Edward R. Tufte might have graphed this information."

 

[LVFL_Soccer]

( Kinda ) MAD we don't have one for soccer

 

[RetroVol]

View attachment 839710

 

[chuckiepoo]

I don't know if this idea has any merit, but at this hour, after a night of interrupted sleep and before any morning coffee... it makes sense in my brain.

If I wanted a kind of statistical model against which to track the progress, strengths, and weaknesses of this season's team, I would want to build that model using data from CKC's teams at Glenville State. That's where we know the system was performing as-intended. And if those stats are retrievable, then they would also offer a longitudinal sample of paths and patterns of improvement from year-to-year, as well as within each season.

From the research I did when her hiring was announced, I remember being impressed that the system allowed her teams to win games many different ways. It seemed to me it would allow teams to break free from the "rock-paper-scissors" (tall vs. quick vs. good shooting) dimension of predictability within which basketball played below the rim gets stuck.

If those impressions are accurate (???) then a model derived from her Glenville State seasons would be complicated with multiple relationships between factors that generated success. I imagine it might produce an amoebic scatterplot* sweetspot--or maybe even several distinct sweetspots.

But I'm already way past my mathematical limits. My gray matter is good at seeing patterns, but I learned late in life that I have an actual defect that makes numerals and numbers "slippery" and probably why I always hated arithmetic and math. So I hope you can find some useful thought in this post.


* If you ever need AI to produce a more accessible graphical representation of information, try prompting it with "and show me 3 ways that Edward R. Tufte might have graphed this information."

Leave the analysis, take the cup of joe. ️

 

[chuckiepoo]

I would sure like to see a projection based on actual players minutes of their proposed statistical output for next years team where you could also draw a projection of how many points this team might average per game and our final totals projection in every statistical category. I did something like that, but it was based on every players 40 minute totals and I guess I could have taken a day or two to break it all down with my own idea of projected players minutes and what stats playing those actual minutes would look like.
What I am hoping for this team is something above 83 ppg and overall averages of 75 percent free throw, 35 percent three point and 52 percent two point. The percentage of three pointers taken around 40 percent and two point 60 percent would give our team a total of 70 shots per game. 28 threes per game and 10 made for 30 points. 42 twos with 22 made for 44 points and then to add something like 9 free throws made per game. That comes out to 44.5 percent overall average for the team. Team would need to also get to the line 12 times per game while averaging 9 makes.
This would probably be big winning numbers over 20 wins in a 30 game season with any kind of decent defense at all.
I know these are fairly lofty numbers, but to me seem reachable if every player matches their past performance. That certainly may be hard to do in the SEC.

What kind of efficiency would that look like over last season. Our three point percentage was 30.5 on 30.5 attempts per game. So I am asking for an improvement of 4.5 percent from the three even taking close to 3 less attempts. We averaged 37 two pointers per game last season and made 49.7 percent of the two point attempts. Looking to up that by 3 percent. Free throw percentage was 67 percent last season and we got to the line an average of 17 times and 11.7 made. Looking for a whopping 8 percent improvement. 67.5 shots per game and our overall average was 41 percent. Looking for three more shots per game while making 3.5 percent more attempts.
Last year we seemed to be fighting against ourselves taking 45 percent of our shots from the three and only making 30.5 percent. 55 percent from the two while making 49.7 percent so last year could've been more successful taking less threes and more twos based on percentages.

Team average 76 ppg last season I am asking for 83 on far better efficiency numbers. I feel confident that we can reach the 35 percent from the three next season. Far less confident that we can shoot 52 percent from the two and 75 from the line. The plus seven points on offense should certainly give us a couple more wins still beyond that always comes down to better defensive play probably also needing to lower defensive points allowed by seven as well.

…another check in before coffee?

 

[LVFL_Soccer]

.

 

[BruisedOrange]

I kind of miss the Lamborghini…

Guess we're goin' old school this year.