Player Ratings (or, decomposing our Kenpom rating)

fluoxetine · Dec 19, 2024

What am I doing here? Attempting to decompose our Kenpom rating of +10.01 into contributions of individual players.

How?
For each game Bart attempts to decompose the margin by player. See: the NET column here (https://barttorvik.com/box.php?muid=Saint+Peter'sRutgers11-11&year=2025).

I took this along with the minutes played and Kenpom rating of each opponent (and home/away/neutral) to compute an ADJUSTED NET.

Then some more math and you get the results. My estimates add up to +10.16 and our actual Kenpom rating is +10.01 so I think this is pretty close. There were also a few minutes played by walk-ons that I didn't account for which is probably some of the error. Also minutes rounding (i.e. the box score gives 201 minutes for the Wagner game, I haven't adjusted for this).

Overall, as a team, we have played 784 possession with a scoring margin of +44 and an adjusted scoring margin of (by my computations) +78.94. This should yield a rating of +10.07. The actual rating on the site is +10.01. That is close enough for me.

If you are averse to spreadsheet basketball, uh, don't read this.

All the usual caveats about how this is a huge approximation and how a lot of defensive stuff does not show up in the box score etc. This is just fun interesting numbers. It is not THE RIGHT ANSWER to anything in particular.

Note, for each possession each player on the court gets 0.2 possessions NOT 1 possession. That is also why I included Adj NET per 20 possessions as that corresponds to 100 team possessions.

Player	Total Adjusted NET Contribution	Minutes Played	Est Possessions Played	% of team total	Adj NET per 20 possessions	Adj NET per 40 minutes	Est Contribution to Kenpom Rating	Best Adj Game	Worst Adj Game
Dylan Harper	126.9	376	132.8	16.9%	19.1	13.5	16.19	Penn St. (+23.4)	Saint Peter's (-1.1)
PJ Hayes	10.5	160	56.4	7.2%	3.7	2.6	1.34	Monmouth (+6.1)	Notre Dame (-6.9)
Jordan Derkack	9.7	275	96.6	12.3%	2.0	1.4	1.24	Monmouth (+6.7)	Penn St. (-5.7)
Lathan Sommerville	5.6	215	75.3	9.6%	1.5	1.0	0.71	Penn St. (+9.6)	Monmouth (-8.3)
Jeremiah Williams	3.4	252	88.2	11.2%	0.8	0.5	0.43	Texas A&M (+13.1)	Ohio St. (-10.4)
Zach Martini	-1.5	176	61.6	7.9%	-0.5	-0.3	-0.19	Monmouth (+4.7)	Alabama (-6.1)
Ace Bailey	-2.1	305	108.7	13.9%	-0.4	-0.3	-0.27	Alabama (+12.2)	Ohio St. (-10.2)
Tyson Acuff	-4.0	133	46.5	5.9%	-1.7	-1.2	-0.51	Notre Dame (+3.9)	Alabama (-4.6)
Dylan Grant	-10.6	20	7.4	0.9%	-28.6	-21.2	-1.35	DNP (0)	Texas A&M (-4.4)
Emmanual Ogbole	-12.1	153	54.8	7.0%	-4.4	-3.2	-1.54	Ohio St. (+7.8)	Wagner (-8.3)
Jamichael Davis	-46.3	161	56.2	7.2%	-16.5	-11.5	-5.91	Wagner (+2.9)	Penn St. (-10.2)
Residual							-0.13
Total							10.01

fluoxetine · Dec 19, 2024

According to this analysis, such as it is:

Dylan Harper + the roster of #154 Ohio would be a tournament team (roughly #41)

This roster minus Dylan Harper would be a Q4 team.

Davis has single handedly moved us from #45 to #79

PhillyRU · Dec 19, 2024

Big moment for the JaMichael Davis isn’t actually good at defense truthers. This is great, my biggest issue/detraction maybe is that Torvik’s individual NET number seems to average their OBPM and DBPM almost 1:1, and I’m just super sketchy about individual defensive metrics.

PJ being a Top 5 defender on the team just doesn’t seem right, although I could really just be relying on my eyes too much.

fluoxetine · Dec 19, 2024

PhillyRU said:
Big moment for the JaMichael Davis isn’t actually good at defense truthers. This is great, my biggest issue/detraction maybe is that Torvik’s individual NET number seems to average their OBPM and DBPM almost 1:1, and I’m just super sketchy about individual defensive metrics.

PJ being a Top 5 defender on the team just doesn’t seem right, although I could really just be relying on my eyes too much.

Yes this will definitely overrate the worse defenders and vice versa

RUJMM78 · Dec 19, 2024

Hayes is the second most productive and he hardly plays or take shots.

lion1983 · Dec 19, 2024

If I understand the math of it correctly - not sure I do so feel free to correct me - Hayes, Derkack, Sommerville, Williams and Martini are all more valuable than Bailey?

If so, any system that puts Martini at the same level of contribution to the team, or more, than Bailey is seriously flawed. That is not even mentioning the other players it puts ahead of Bailey.

Sorry if I misread this.

fluoxetine · Dec 19, 2024

lion1983 said:
If I understand the math of it correctly - not sure I do so feel free to correct me - Hayes, Derkack, Sommerville, Williams and Martini are all more valuable than Bailey?

If so, any system that puts Martini at the same level of contribution to the team, or more, than Bailey is seriously flawed. That is not even mentioning the other players it puts ahead of Bailey.

Sorry if I misread this.

Yes, although it’s Bart’s math (and whoever designed box plus minus) that results in this more than mine (I did the adjustments and weighting but the raw sum still states this).

It is based on box plus minus and Bailey’s is not good.

While on the one hand I think you are correct (Bailey is way too far down this list) on the other hand I think it points in the direction of something real (despite Bailey’s huge potential he has not really been a hugely above replacement level player thus far).

fluoxetine · Dec 19, 2024

There are three stages here basically:

(1) Bart computes box plus minus
(2) Bart takes that + minutes played and the score of the game to compute NET which is an estimate of the players’ impact on the game in terms of points
(3) I take number 2 and adjust it for the opponent quality and location. That’s my raw adjusted contribution. I then weight that by the # of possessions they were in the game to estimate their contribution to the overall rating

PhillyRU · Dec 19, 2024

lion1983 said:
If I understand the math of it correctly - not sure I do so feel free to correct me - Hayes, Derkack, Sommerville, Williams and Martini are all more valuable than Bailey?

If so, any system that puts Martini at the same level of contribution to the team, or more, than Bailey is seriously flawed. That is not even mentioning the other players it puts ahead of Bailey.

Sorry if I misread this.

Most of that is Bart thinking Ace is a really bad defender, but he has not been efficient offensively either honestly. His shooting splits are fine, especially for his horrific shot diet lol, but he’s got a 1:5 assist to turnover ratio and shoots 60% at the line, albeit that’s very skewed by SHU.

RUChoppin · Dec 19, 2024

I really dislike box plus/minus as an individual stat, because it ignores how teams function together. For example, Harper is by pretty much all metrics and eye tests our best player so far this year.... which means players who share more minutes with him will have better box plus/minus than players that are used more often to give him rest.

I think box plus/minus should be used for lineups, not for individuals. How does a combination of 4 guys + Davis compare to all the combinations of those same 4 guys + "not Davis".

I tried to look at this somewhat a couple years back, but it's tough to get the data.

Individual box-plus/minus is an even more misleading stat than usage, imo.

trotter · Dec 19, 2024

@fluoxotine, great stuff! Just curious, what does your handle represent? Second, I dabble in a little sport gambling. Have you ever looked at any of these sites for a gambling edge? Currently playing around with Torvik and comparing it to actual point spread and adding in a min edge factor. Probably not going to be profitable, but I can’t pick my nose either.

fluoxetine · Dec 19, 2024

trotter said:
@fluoxotine, great stuff! Just curious, what does your handle represent?

A long time ago (before this handle, so 10+ years ago at least) I used to make new accounts for what I would describe as "OCD-like" reasons (though I do not have clinical OCD or anything like that). The joke behind this handle is that I was going to stop doing that so it is named after a drug that is used to treat OCD.

trotter said:
Second, I dabble in a little sport gambling. Have you ever looked at any of these sites for a gambling edge? Currently playing around with Torvik and comparing it to actual point spread and adding in a min edge factor. Probably not going to be profitable, but I can’t pick my nose either.

I have looked. It won't beat the market line on it's own. At their core Bart/Kenpom are actually pretty basic calculations and are kind of the baseline models that spreads are built on in the first place. I wouldn't be surprised if something like (https://shotquality.com/stats-explained) could be market beating for now but I think it generally requires that kind of specialized model.

RUDivision · Dec 20, 2024

fluoxetine said:
What am I doing here? Attempting to decompose our Kenpom rating of +10.01 into contributions of individual players.

How?
For each game Bart attempts to decompose the margin by player. See: the NET column here (https://barttorvik.com/box.php?muid=Saint+Peter'sRutgers11-11&year=2025).

I took this along with the minutes played and Kenpom rating of each opponent (and home/away/neutral) to compute an ADJUSTED NET.

Then some more math and you get the results. My estimates add up to +10.16 and our actual Kenpom rating is +10.01 so I think this is pretty close. There were also a few minutes played by walk-ons that I didn't account for which is probably some of the error. Also minutes rounding (i.e. the box score gives 201 minutes for the Wagner game, I haven't adjusted for this).

Overall, as a team, we have played 784 possession with a scoring margin of +44 and an adjusted scoring margin of (by my computations) +78.94. This should yield a rating of +10.07. The actual rating on the site is +10.01. That is close enough for me.

If you are averse to spreadsheet basketball, uh, don't read this.

All the usual caveats about how this is a huge approximation and how a lot of defensive stuff does not show up in the box score etc. This is just fun interesting numbers. It is not THE RIGHT ANSWER to anything in particular.

Note, for each possession each player on the court gets 0.2 possessions NOT 1 possession. That is also why I included Adj NET per 20 possessions as that corresponds to 100 team possessions.

Player Total Adjusted NET Contribution Minutes Played Est Possessions Played % of team total Adj NET per 20 possessions Adj NET per 40 minutes Est Contribution to Kenpom Rating Best Adj Game Worst Adj Game
Dylan Harper
126.9
376
132.8
16.9%
19.1
13.5
16.19
Penn St. (+23.4) Saint Peter's (-1.1)
PJ Hayes
10.5
160
56.4
7.2%
3.7
2.6
1.34
Monmouth (+6.1) Notre Dame (-6.9)
Jordan Derkack
9.7
275
96.6
12.3%
2.0
1.4
1.24
Monmouth (+6.7) Penn St. (-5.7)
Lathan Sommerville
5.6
215
75.3
9.6%
1.5
1.0
0.71
Penn St. (+9.6) Monmouth (-8.3)
Jeremiah Williams
3.4
252
88.2
11.2%
0.8
0.5
0.43
Texas A&M (+13.1) Ohio St. (-10.4)
Zach Martini
-1.5
176
61.6
7.9%
-0.5
-0.3
-0.19
Monmouth (+4.7) Alabama (-6.1)
Ace Bailey
-2.1
305
108.7
13.9%
-0.4
-0.3
-0.27
Alabama (+12.2) Ohio St. (-10.2)
Tyson Acuff
-4.0
133
46.5
5.9%
-1.7
-1.2
-0.51
Notre Dame (+3.9) Alabama (-4.6)
Dylan Grant
-10.6
20
7.4
0.9%
-28.6
-21.2
-1.35
DNP (0) Texas A&M (-4.4)
Emmanual Ogbole
-12.1
153
54.8
7.0%
-4.4
-3.2
-1.54
Ohio St. (+7.8) Wagner (-8.3)
Jamichael Davis
-46.3
161
56.2
7.2%
-16.5
-11.5
-5.91
Wagner (+2.9) Penn St. (-10.2)
Residual
-0.13
Total
10.01

lol JMike! Oh and you can rewatch the PSU game again he wasn’t good defensively. He had 2 to 3 moments which accounts to less then a minute of the 20 he was on the court.

fluoxetine · Dec 22, 2024

After Princeton:

Player	Total Adjusted NET Contribution	Minutes Played	Est Possessions Played	% of team total	Adj NET per 20 possessions	Adj NET per 40 minutes	Est Contribution to Kenpom Rating
Dylan Harper	129.5	412	146.1	17.03%	17.7	12.6	15.1
PJ Hayes	8.4	167	59.0	6.88%	2.8	2.0	1.0
Jeremiah Williams	6.5	274	96.3	11.23%	1.4	1.0	0.8
Jordan Derkack	6.2	298	105.1	12.25%	1.2	0.8	0.7
Lathan Sommerville	3.7	229	80.5	9.38%	0.9	0.7	0.4
Zach Martini	-0.4	196	69.0	8.05%	-0.1	-0.1	-0.1
Tyson Acuff	-2.8	138	48.4	5.64%	-1.2	-0.8	-0.3
Emmanuel Ogbole	-7.8	169	60.7	7.07%	-2.6	-1.8	-0.9
Ace Bailey	-8.8	338	120.9	14.09%	-1.5	-1.0	-1.0
Dylan Grant	-10.5	20	7.4	0.86%	-28.4	-21.1	-1.2
Jamichael Davis	-38.4	186	65.5	7.63%	-11.7	-8.3	-4.5
Residual							-0.2
Total							9.8

RuSnp · Dec 22, 2024

fluoxetine said:
After Princeton:

Player Total Adjusted NET Contribution Minutes Played Est Possessions Played % of team total Adj NET per 20 possessions Adj NET per 40 minutes Est Contribution to Kenpom Rating
Dylan Harper
129.5
412
146.1
17.03%
17.7
12.6
15.1
PJ Hayes
8.4
167
59.0
6.88%
2.8
2.0
1.0
Jeremiah Williams
6.5
274
96.3
11.23%
1.4
1.0
0.8
Jordan Derkack
6.2
298
105.1
12.25%
1.2
0.8
0.7
Lathan Sommerville
3.7
229
80.5
9.38%
0.9
0.7
0.4
Zach Martini
-0.4
196
69.0
8.05%
-0.1
-0.1
-0.1
Tyson Acuff
-2.8
138
48.4
5.64%
-1.2
-0.8
-0.3
Emmanuel Ogbole
-7.8
169
60.7
7.07%
-2.6
-1.8
-0.9
Ace Bailey
-8.8
338
120.9
14.09%
-1.5
-1.0
-1.0
Dylan Grant
-10.5
20
7.4
0.86%
-28.4
-21.1
-1.2
Jamichael Davis
-38.4
186
65.5
7.63%
-11.7
-8.3
-4.5
Residual
-0.2
Total
9.8

Wonder what Martini's would look like if the Monmouth game never happened.

fluoxetine · Dec 23, 2024

lion1983 said:
If I understand the math of it correctly - not sure I do so feel free to correct me - Hayes, Derkack, Sommerville, Williams and Martini are all more valuable than Bailey?

If so, any system that puts Martini at the same level of contribution to the team, or more, than Bailey is seriously flawed. That is not even mentioning the other players it puts ahead of Bailey.

Sorry if I misread this.

fluoxetine said:
Yes, although it’s Bart’s math (and whoever designed box plus minus) that results in this more than mine (I did the adjustments and weighting but the raw sum still states this).

It is based on box plus minus and Bailey’s is not good.

While on the one hand I think you are correct (Bailey is way too far down this list) on the other hand I think it points in the direction of something real (despite Bailey’s huge potential he has not really been a hugely above replacement level player thus far).

PhillyRU said:
Most of that is Bart thinking Ace is a really bad defender, but he has not been efficient offensively either honestly. His shooting splits are fine, especially for his horrific shot diet lol, but he’s got a 1:5 assist to turnover ratio and shoots 60% at the line, albeit that’s very skewed by SHU.

Actually I've thought about this a little more and it is not exactly saying those players are more valuable than Bailey. It says those players are hurting us less than Bailey, but if you look at Bart's usage adjusted rating thingy it has Bailey as the second best player on the team.

What it really says is that Bailey is hurting us more than those players because his usage is too high relative to what he is. Presumably if you asked those other players to force up 15 shots per game like Bailey is doing then their numbers would be even worse.

PhillyRU · Dec 23, 2024

fluoxetine said:
Actually I've thought about this a little more and it is not exactly saying those players are more valuable than Bailey. It says those players are hurting us less than Bailey, but if you look at Bart's usage adjusted rating thingy it has Bailey as the second best player on the team.

What it really says is that Bailey is hurting us more than those players because his usage is too high relative to what he is. Presumably if you asked those other players to force up 15 shots per game like Bailey is doing then their numbers would be even worse.

Right but the PRPG (Points over Replacement Per Game) figure just measures offensive replacement value I believe. Interestingly, Bailey has a team-best 3.7 D-PRPG despite having a -0.8 DBPM (Box +\-), which I can only assume is because Torvik assumes Dylan Grant (team-worst 0.6 D-PRPG/-3.1DBPM) would be his replacement.

I’m less certain about how Torvik calculates his individual NET value than I was a week ago, but it seems like most of the marks against Bailey are attributable to his perceived defensive shortcomings, although he’s not great on either end right now.

RUChoppin · Dec 23, 2024

fluoxetine said:
Actually I've thought about this a little more and it is not exactly saying those players are more valuable than Bailey. It says those players are hurting us less than Bailey, but if you look at Bart's usage adjusted rating thingy it has Bailey as the second best player on the team.

What it really says is that Bailey is hurting us more than those players because his usage is too high relative to what he is. Presumably if you asked those other players to force up 15 shots per game like Bailey is doing then their numbers would be even worse.

You are putting too much stock in individual box plus/minus. It doesn't "really mean" anything... At best, it's a weak indicator of something that ignores countless variables and should not be taken in a bubble.

PSAL_Hoops · Dec 23, 2024

I don’t know - I get that this is adjusted to competition level in some way, but Sommerville’s net positive contributions have mostly come against terrible teams. PJ’s data seems like a blending issue too as he played a lot more possessions against terrible teams. Again - I know it’s supposedly adjusted but there's only so much that can be the case.

For example - there are certain shortcomings that are not a problem against competition that completely sucks but becomes rate limiting when the O or D is even halfway decent. There’s no algorithm that can appropriately account for this.

fluoxetine · Dec 23, 2024

RUChoppin said:
You are putting too much stock in individual box plus/minus. It doesn't "really mean" anything... At best, it's a weak indicator of something that ignores countless variables and should not be taken in a bubble.

I'm not sure I'm putting that much stock in it. I'm just reporting it. I don't really think this conclusion about Ace is wrong though.

seansherm · Dec 23, 2024

I don't look at the analytics in too much depth, but they seem to like Sommerville

fluoxetine · Dec 23, 2024

PSAL_Hoops said:
I don’t know - I get that this is adjusted to competition level in some way, but Sommerville’s net positive contributions have mostly come against terrible teams. PJ’s data seems like a blending issue too as he played a lot more possessions against terrible teams. Again - I know it’s supposedly adjusted but there's only so much that can be the case.

For example - there are certain shortcomings that are not a problem against competition that completely sucks but becomes rate limiting when the O or D is even halfway decent. There’s no algorithm that can appropriately account for this.

Total adjusted NET in only the 6 games against power conference opponents:

Dylan Harper	90.0
Emmanuel Ogbole	8.9
Lathan Sommerville	7.5
Jordan Derkack	2.3
Tyson Acuff	0.5
Ace Bailey	0.5
Jeremiah Williams	-1.9
PJ Hayes	-6.5
Dylan Grant	-7.0
Zach Martini	-8.1
Jamichael Davis	-33.2

fluoxetine · Dec 23, 2024

	vs P5	Other	Total
Dylan Harper	90.0	39.5	129.5
Emmanuel Ogbole	8.9	-16.7	-7.8
Lathan Sommerville	7.5	-3.8	3.7
Jordan Derkack	2.3	3.9	6.2
Tyson Acuff	0.5	-3.3	-2.8
Ace Bailey	0.5	-9.4	-8.8
Jeremiah Williams	-1.9	8.5	6.5
PJ Hayes	-6.5	14.9	8.4
Dylan Grant	-7.0	-3.6	-10.5
Zach Martini	-8.1	7.7	-0.4
Jamichael Davis	-33.2	-5.2	-38.4

Ridge 22 · Dec 23, 2024

The eye test and record both say we aren't that good

RuSnp · Dec 23, 2024

fluoxetine said:
Total adjusted NET in only the 6 games against power conference opponents:

Dylan Harper
90.0
Emmanuel Ogbole
8.9
Lathan Sommerville
7.5
Jordan Derkack
2.3
Tyson Acuff
0.5
Ace Bailey
0.5
Jeremiah Williams
-1.9
PJ Hayes
-6.5
Dylan Grant
-7.0
Zach Martini
-8.1
Jamichael Davis
-33.2

Sigh

PSAL_Hoops · Dec 23, 2024

fluoxetine said:
Total adjusted NET in only the 6 games against power conference opponents:

Dylan Harper
90.0
Emmanuel Ogbole
8.9
Lathan Sommerville
7.5
Jordan Derkack
2.3
Tyson Acuff
0.5
Ace Bailey
0.5
Jeremiah Williams
-1.9
PJ Hayes
-6.5
Dylan Grant
-7.0
Zach Martini
-8.1
Jamichael Davis
-33.2

Thanks. Interesting data. Confirms my thinking on PJ. Jeremiah’s data is heavily skewed by the OSU game. I’m not saying it should be eliminated but having a bad game like that as one if only six your factoring is going to skew the output. Point being - he was solid in 5 of 6 games against major conference teams.

Probably the best sample of the population to use would be to only remove the games where we dominated the opponent. Princeton, Kennesaw, etc. data should count. Wagner should not.

fluoxetine · Dec 23, 2024

PSAL_Hoops said:
Probably the best sample of the population to use would be to only remove the games where we dominated the opponent. Princeton, Kennesaw, etc. data should count. Wagner should not.

I think it should be by quality of the opponent not on how we did against them. If, for example, we had beaten Penn State by 20 because a couple of our players had great games I don't think those great games should then be removed from the analysis.

Shelby65 · Dec 23, 2024

RUChoppin said:
You are putting too much stock in individual box plus/minus. It doesn't "really mean" anything... At best, it's a weak indicator of something that ignores countless variables and should not be taken in a bubble.

Doesn’t every metric have countless underlying variables ? Answer: yes.

RutgersRaRa · Dec 23, 2024

Decomposition is such a nasty process. Deconstruction, on the other hand, is less organic but much less pungent.

RUChoppin · Dec 23, 2024

So much of this is also based on who you are on the floor with.

For example, Derkack played 23 min against Princeton.... and 15 of them overlapped with Martini. Williams played 22 min, and only 8 of them were with Martini. Harper was only off the floor for 4 minutes (and 2 of those also saw Bailey on the bench)... and Sommerville and Derkack were on the floor for all of them (about 29% of Sommerville's time on the floor; about 17% for Derkack).

Any sort of analysis like this should take into account the five-man unit, rather than the individual player. Unfortunately, that data is difficult to come by (for Princeton, I had to wade through the box score on sk.com and mark every sub in/sub out and the time... and it's not available for every game).

Greene Rice FIG · Dec 23, 2024

Evan miya stats looks at what def and off adj eff is when a player is in the game. Not looking at stats.

We are dreadful on D when PJ plays. Sample size is suspect but data matches my eyeballs and Pikes

RUChoppin · Dec 23, 2024

Greene Rice FIG said:
Evan miya stats looks at what def and off adj eff is when a player is in the game. Not looking at stats.

We are dreadful on D when PJ plays. Sample size is suspect but data matches my eyeballs and Pikes

Hayes played just 7 minutes.... and Sommerville was on the floor for 6 of them.

Amazingly, for 2:14 (across two different rotations), we had Sommerville/Martini/Hayes on the floor together. For 1:00 it was with Acuff/Derkack, and for 1:14 it was with Harper/Derkack.

Greene Rice FIG · Dec 24, 2024

RUChoppin said:
Hayes played just 7 minutes.... and Sommerville was on the floor for 6 of them.

Amazingly, for 2:14 (across two different rotations), we had Sommerville/Martini/Hayes on the floor together. For 1:00 it was with Acuff/Derkack, and for 1:14 it was with Harper/Derkack.

For the season

PSAL_Hoops · Dec 24, 2024

RUChoppin said:
Hayes played just 7 minutes.... and Sommerville was on the floor for 6 of them.

Amazingly, for 2:14 (across two different rotations), we had Sommerville/Martini/Hayes on the floor together. For 1:00 it was with Acuff/Derkack, and for 1:14 it was with Harper/Derkack.

It's funny you put it this way. My feeling is that Lathan is an ideal option to be on the floor when Dylan and Ace are sitting because he’s arguably then appropriately positioned as the go to scoring option. I think he hurts our team potential quite a bit when playing with them because he’s not a part of the ball rotation and absolutely everything becomes iso ball - and he still wants to be the go-to. Not to mention his D is inferior to other options. None of that is reflected in the plus minus though. It’s more an observation that a score first BIG who plays soft on D is not a great fit to play big minutes with our 2 stars. At least in my opinion. He’s just not looking to find them and is instead looking to score himself.

RUChoppin · Dec 24, 2024

Greene Rice FIG said:
For the season

Yes, but Pike has rotational patterns. Some guys are going to play more with others. There will be a guy who played the lowest percentage of minutes with Harper, and a guy who played the highest percentage of minutes with Martini, etc - especially so for guys who aren't getting starter minutes. None of that is factored into box plus/minus, which is seemingly what Bart's NET contribution is based on.

I picked the Princeton game because it was the most recent, and the data was available.

Greene Rice FIG · Dec 24, 2024

For the season when you look at Evan Miya offensive and defensive adjusted efficiencies it shows when Hayes is on the ice we give up a lot of points per possession....now the question is who is on the floor when he plays and how is their D, but still.

Still not a ton of sample size to come to a conclusion.

There was one year where when you look at Caleb he brought down def eff a ton, but also brought down off eff down a ton too.

fluoxetine · Dec 24, 2024

Yes as far as I am aware the currently accepted best way to do this (and what gets computed for the NBA) is to essentially do this EXCEPT for every specific lineup combination in the game.

So for each different 5 on 5 you take the score during that stretch and then split that up based on BPM and adjust and repeat. So you have a bunch of mini 5-on-5 games instead of one bigger game.

I assume the NCAA data is not reliable enough to figure out what lineups are in all the time or someone would be doing this.

fluoxetine · Dec 24, 2024

Also while I will often argue against eye test type stuff.. that's only when you have a lot of data to go by. When we are talking about data that's only based on 200-300 minutes of play there will be a ton of variance and that is when the eye test is valuable.

fluoxetine · Dec 31, 2024

Post Columbia:

Player	Total Adjusted NET Contribution	Minutes Played	Est Possessions Played	% of team total	Adj NET per 20 possessions	Adj NET per 40 minutes	Est Contribution to Kenpom Rating	Change from previous
Dylan Harper	157.21	444	157.60	16.95%	19.95	14.16	16.90	1.80
PJ Hayes	5.87	178	62.99	6.77%	1.86	1.32	0.63	-0.37
Jordan Derkack	5.11	314	110.86	11.92%	0.92	0.65	0.55	-0.15
Lathan Sommerville	1.55	248	87.31	9.39%	0.35	0.25	0.17	-0.23
Jeremiah Williams	-0.27	296	104.24	11.21%	-0.05	-0.04	-0.03	-0.83
Ace Bailey	-3.35	369	132.03	14.20%	-0.51	-0.36	-0.36	0.64
Tyson Acuff	-5.56	153	53.79	5.78%	-2.07	-1.45	-0.60	-0.30
Zach Martini	-7.52	206	72.63	7.81%	-2.07	-1.46	-0.81	-0.71
Dylan Grant	-10.50	20	7.42	0.80%	-28.31	-20.99	-1.13	0.07
Emmanuel Ogbole	-10.71	183	65.71	7.07%	-3.26	-2.34	-1.15	-0.25
Jamichael Davis	-24.23	215	75.93	8.16%	-6.38	-4.51	-2.61	1.89
Residual							-0.63	-0.43
Total							10.94	1.14

RUChoppin · Dec 31, 2024

fluoxetine said:
Post Columbia:

Player Total Adjusted NET Contribution Minutes Played Est Possessions Played % of team total Adj NET per 20 possessions Adj NET per 40 minutes Est Contribution to Kenpom Rating Change from previous
Dylan Harper
157.21
444
157.60
16.95%
19.95
14.16
16.90
1.80
PJ Hayes
5.87
178
62.99
6.77%
1.86
1.32
0.63
-0.37
Jordan Derkack
5.11
314
110.86
11.92%
0.92
0.65
0.55
-0.15
Lathan Sommerville
1.55
248
87.31
9.39%
0.35
0.25
0.17
-0.23
Jeremiah Williams
-0.27
296
104.24
11.21%
-0.05
-0.04
-0.03
-0.83
Ace Bailey
-3.35
369
132.03
14.20%
-0.51
-0.36
-0.36
0.64
Tyson Acuff
-5.56
153
53.79
5.78%
-2.07
-1.45
-0.60
-0.30
Zach Martini
-7.52
206
72.63
7.81%
-2.07
-1.46
-0.81
-0.71
Dylan Grant
-10.50
20
7.42
0.80%
-28.31
-20.99
-1.13
0.07
Emmanuel Ogbole
-10.71
183
65.71
7.07%
-3.26
-2.34
-1.15
-0.25
Jamichael Davis
-24.23
215
75.93
8.16%
-6.38
-4.51
-2.61
1.89
Residual
-0.63
-0.43
Total
10.94
1.14

Any metric that has PJ Hayes as #2 on our team in value is immediately suspect. As is any that have Martini as more valuable than Ogbole.

Player Ratings (or, decomposing our Kenpom rating)

Heisman Winner

Heisman Winner

Junior

Heisman Winner

Hall of Famer

Junior

Heisman Winner

Heisman Winner

Junior

Heisman Winner

All Conference

Heisman Winner

Senior

Heisman Winner

All Conference

Heisman Winner

Junior

Heisman Winner

All American

Heisman Winner

All American

Heisman Winner

Heisman Winner

All American

All Conference

All American

Heisman Winner

All American

Hall of Famer

Heisman Winner

Hall of Famer

Heisman Winner

Hall of Famer

All American

Heisman Winner

Hall of Famer

Heisman Winner

Heisman Winner

Heisman Winner

Heisman Winner

Similar threads