Giving Credit Where Credit is Due

[PurpleWhiteBoy]

The following table was re-created from statistics in Evan Miyakawa's excellent website.
It tell us which players were most important to each of the most recent Northwestern teams.
It also tells us which players improved or flatlined - and when.
Poss (Possessions) tells us who was getting minutes..
It is important to understand that Miyakawa measures productivity while each player is in the game. He is normalizing for minutes played.
The ratings represent Points per Game the player would add to a team of average D1 players.

	2021-22	Off 3.1	Def 5.7	2022-23	Off 4.6	Def 8.9	2023-24	Off 6.6	Def 5.9	2024-25	Off 7.2	Def 7.9
	Rating	Poss	+/-	Rating	Poss	+/-	Rating	Poss	+/-	Rating	Poss	+/-
Nance	3.48	1398	+87	---	---	---	---	---	---	---	---	---
Buie	3.26	1558	+84	3.37	1923	+141	5.17	2050	+171	---	---	---
E. Williams	2.36	606	+75	---	---	---	---	---	---	---	---	---
Berry	1.93	1160	+79	3.64	1574	+165	3.71	1130	+101	2.89	1612	+34
Greer	1.83	986	+67	---	---	---	---	---	---	---	---	---
Audige	1.63	1090	-53	3.69	1893	+165	---	---	---	---	---	---
Young	1.42	872	-11	---	---	---	---	---	---	---	---	---
Roper	0.92	982	+6	1.52	548	+24	---	---	---	---	---	---
Simmons	0.78	525	+14	---	---	---	---	---	---	---	---	---
Beran	0.39	1161	-28	2.53	1428	+129	---	---	---	---	---	---
Nicholson	0.2	102	+2	2.96	1172	+134	3.58	1026	+140	5.25	1354	+151
Barnhizer	-0.36	122	-19	2.31	1334	+95	3.60	2046	+132	4.27	1056	+58
Verhoeven				1.55	839	+41	---	---	---	---	---	---
Martinelli				-0.53	300	-14	1.23	1421	+47	2.69	2062	+113
Langborg							3.42	1745	+150	---	---	---
Preston							2.50	462	+29	---	---	---
B. Smith							1.67	238	+20	-0.68	90	-28
Hunger							0.23	670	-14	-0.01	499	+1
Mullins							-0.15	156	+5	1.68	1162	+41
Clayton							-2.71	324	-52	-0.39	342	-18
Windham										3.55	802	+101
Leach										2.87	1212	+86
Ciaravino										1.66	552	+40
Fitzmorris										0.61	310	-27

One fact that jumps out is that the 2024-2025 team is rated HIGHER than any other NU team.
Another is that Matt Nicholson had the single highest individual rating of any NU player, ahead of Boo Buie in 2023-24.
Again, this is normalized for actual minutes played.
KJ Windham was good last season, as a freshman, playing to the level of Ty Berry in his two best seasons or Ryan Langborg as a grad student, but better than Jalen Leach. This is largely because of his defensive rating of 2.08.

 

[EagerFan]

Looks like Clayton made a huge jump in his second year, +2.32 movement in ratings, right? Mullins had a nice jump as well, good to know.

 

[Purple Pile Driver]

The following table was re-created from statistics in Evan Miyakawa's excellent website.
It tell us which players were most important to each of the most recent Northwestern teams.
It also tells us which players improved or flatlined - and when.
Poss (Possessions) tells us who was getting minutes..
It is important to understand that Miyakawa measures productivity while each player is in the game. He is normalizing for minutes played.
The ratings represent Points per Game the player would add to a team of average D1 players.

	2021-22	Off 3.1	Def 5.7	2022-23	Off 4.6	Def 8.9	2023-24	Off 6.6	Def 5.9	2024-25	Off 7.2	Def 7.9
	Rating	Poss	+/-	Rating	Poss	+/-	Rating	Poss	+/-	Rating	Poss	+/-
Nance	3.48	1398	+87	---	---	---	---	---	---	---	---	---
Buie	3.26	1558	+84	3.37	1923	+141	5.17	2050	+171	---	---	---
E. Williams	2.36	606	+75	---	---	---	---	---	---	---	---	---
Berry	1.93	1160	+79	3.64	1574	+165	3.71	1130	+101	2.89	1612	+34
Greer	1.83	986	+67	---	---	---	---	---	---	---	---	---
Audige	1.63	1090	-53	3.69	1893	+165	---	---	---	---	---	---
Young	1.42	872	-11	---	---	---	---	---	---	---	---	---
Roper	0.92	982	+6	1.52	548	+24	---	---	---	---	---	---
Simmons	0.78	525	+14	---	---	---	---	---	---	---	---	---
Beran	0.39	1161	-28	2.53	1428	+129	---	---	---	---	---	---
Nicholson	0.2	102	+2	2.96	1172	+134	3.58	1026	+140	5.25	1354	+151
Barnhizer	-0.36	122	-19	2.31	1334	+95	3.60	2046	+132	4.27	1056	+58
Verhoeven				1.55	839	+41	---	---	---	---	---	---
Martinelli				-0.53	300	-14	1.23	1421	+47	2.69	2062	+113
Langborg							3.42	1745	+150	---	---	---
Preston							2.50	462	+29	---	---	---
B. Smith							1.67	238	+20	-0.68	90	-28
Hunger							0.23	670	-14	-0.01	499	+1
Mullins							-0.15	156	+5	1.68	1162	+41
Clayton							-2.71	324	-52	-0.39	342	-18
Windham										3.55	802	+101
Leach										2.87	1212	+86
Ciaravino										1.66	552	+40
Fitzmorris										0.61	310	-27

One fact that jumps out is that the 2024-2025 team is rated HIGHER than any other NU team.
Another is that Matt Nicholson had the single highest individual rating of any NU player, ahead of Boo Buie in 2023-24.
Again, this is normalized for actual minutes played.
KJ Windham was good last season, as a freshman, playing to the level of Ty Berry in his two best seasons or Ryan Langborg as a grad student, but better than Jalen Leach. This is largely because of his defensive rating of 2.08.

I am not buying into any rating system that claims Big Matt’s 24-25 season contributed more to the team than Boo’s 23-24 season. In fact, how on Earth is KJ’s rating higher than Nicky buckets last year.

 

[ThatkidfromHolland]

I am not buying into any rating system that claims Big Matt’s 24-25 season contributed more to the team than Boo’s 23-24 season. In fact, how on Earth is KJ’s rating higher than Nicky buckets last year.

It has Buie as the 3rd best player in 22-23. Any rating that comes to that conclusion has some major flaws.

 

[PurpleWhiteBoy]

I am not buying into any rating system that claims Big Matt’s 24-25 season contributed more to the team than Boo’s 23-24 season. In fact, how on Earth is KJ’s rating higher than Nicky buckets last year.

There is so much useful information in those numbers, it makes me sad that you don't see it.
You are free to ignore the information.
However, we all know that the guy who generated it is
a) not biased in any way
b) very intelligent
and
c) uses a ton of information that none of us has the ability to process - across all of D1.

Some people just don't seem to comprehend the positive impact that Windham had on the team, when he got to play.
Same goes for Nicholson.

It has Buie as the 3rd best player in 22-23. Any rating that comes to that conclusion has some major flaws.

Major Flaws? Hardly. It accounts for defense. Ty Berry and Chase Audige improved dramatically after Lowery took over the defensive schemes and taught some new techniques. Buie did not.

	2021-22 Off	2021-22 Def	2022-23 Off	2022-23 Def	2023-24 Off	2023-24 Def
Buie	2.33	0.93	2.21	1.16	4.08	1.09
Audige	0.54	1.10	1.32	2.37	---	---
Berry	0.80	1.13	0.87	2.77	2.54	1.17

The fact is that Miyakawa's numbers are the best publicly available player ratings. Torvik has numbers that rate players as well, but they're totally dependent on how many minutes the player gets - and therefore not nearly as useful as Miyakawa's.

There are some MINOR tweaks that might be made to try to account for things like "Buie was the focus of the opposing defense" or "Barnhizer was playing almost 40 minutes a game, so he was tired at times" and I don't know if Miyakawa does that or not, but his analysis is so good, its a shame you'd dismiss it. Its been a goldmine for me.

Edit - Miyakawa's website offers no way to eliminate garbage time or filter out the games against the really bad teams. Thats a shame and I'm sure it affects the player ratings to some degree.

 

[ThatkidfromHolland]

There is so much useful information in those numbers, it makes me sad that you don't see it.
You are free to ignore the information.
However, we all know that the guy who generated it is
a) not biased in any way
b) very intelligent
and
c) uses a ton of information that none of us has the ability to process - across all of D1.

Some people just don't seem to comprehend the positive impact that Windham had on the team, when he got to play.
Same goes for Nicholson.



Major Flaws? Hardly. It accounts for defense. Ty Berry and Chase Audige improved dramatically after Lowery took over the defensive schemes and taught some new techniques. Buie did not.

	2021-22 Off	2021-22 Def	2022-23 Off	2022-23 Def	2023-24 Off	2023-24 Def
Buie	2.33	0.93	2.21	1.16	4.08	1.09
Audige	0.54	1.10	1.32	2.37	---	---
Berry	0.80	1.13	0.87	2.77	2.54	1.17

The fact is that Miyakawa's numbers are the best publicly available player ratings. Torvik has numbers that rate players as well, but they're totally dependent on how many minutes the player gets - and therefore not nearly as useful as Miyakawa's.

There are some MINOR tweaks that might be made to try to account for things like "Buie was the focus of the opposing defense" or "Barnhizer was playing almost 40 minutes a game, so he was tired at times" and I don't know if Miyakawa does that or not, but his analysis is so good, its a shame you'd dismiss it. Its been a goldmine for me.

Edit - Miyakawa's website offers no way to eliminate garbage time or filter out the games against the really bad teams. Thats a shame and I'm sure it affects the player ratings to some degree.

If it ranks the clear best player on a team who was All Big Ten first team and a Cousy award finalist as the third best player on said team, it has major flaws. Rating Berry and Chase above Buie during that season is a glaring indicator there’s some major flaws.

Martinelli and Barnhizer were easily the best two players last season for NU. Martinelli being the 4th best player is a crime.

 

[PurpleWhiteBoy]

If it ranks the clear best player on a team who was All Big Ten first team and a Cousy award finalist as the third best player on said team, it has major flaws. Rating Berry and Chase above Buie during that season is a glaring indicator there’s some major flaws.

Martinelli and Barnhizer were easily the best two players last season for NU. Martinelli being the 4th best player is a crime.

Thats a terrible argument regarding Audige and Buie.
The coaches in the Big Ten put both Audige and Buie as 2nd team All Conference.
So you say Buie was clearly the best player on the team. Well not according to the coaches.
Or Miyakawa's numbers.
I thought Buie was better, but the evidence challenges my opinion. And yours.

Ignoring Barnhizer's play when injured (which hurt the team) is foolish and obviously the result of personal bias.
The stats guys don't have that bias and their numbers include how badly Barnhizer played at the end.
If you want to say "When Barnhizer was healthy, he was the best player on the team" sure, go for it.

As for Martinelli, he was not an efficient scorer, not a good passer, did not stand out defensively and he played a ton of minutes.
It shouldn't really surprise anyone that other players would be more effective carrying less of the load.

These great metrics give us all a way to compare players across different teams and seasons.
Thats why I presented all that information above - to cast the bias aside.

Thankfully Chris Collins has started using the metrics in his decision-making.

 

[ThatkidfromHolland]

Thats a terrible argument regarding Audige and Buie.
The coaches in the Big Ten put both Audige and Buie as 2nd team All Conference.
So you say Buie was clearly the best player on the team. Well not according to the coaches.
Or Miyakawa's numbers.
I thought Buie was better, but the evidence challenges my opinion. And yours.

Ignoring Barnhizer's play when injured (which hurt the team) is foolish and obviously the result of personal bias.
The stats guys don't have that bias and their numbers include how badly Barnhizer played at the end.
If you want to say "When Barnhizer was healthy, he was the best player on the team" sure, go for it.

As for Martinelli, he was not an efficient scorer, not a good passer, did not stand out defensively and he played a ton of minutes.
It shouldn't really surprise anyone that other players would be more effective carrying less of the load.

These great metrics give us all a way to compare players across different teams and seasons.
Thats why I presented all that information above - to cast the bias aside.

Thankfully Chris Collins has started using the metrics in his decision-making.

Do you think there were 3 better players than Martinelli on last years team?

If this changed your opinion on Audige how’d he score on your adjusted +\- that season and what’s that say about that metric?

Audige was 2nd best imo. Boo Buie was 1st team All big ten by the media and considered by almost everyone as the best player on that team. I would still take hobbled Barnhizer over a lot of players. He was still stuffing the stat sheet and being a lockdown defender.

 

[TheC]

Thus, the limitations of focusing solely on individual stats is made glaringly obvious.

 

[Purple Pile Driver]

Do you think there were 3 better players than Martinelli on last years team?

If this changed your opinion on Audige how’d he score on your adjusted +\- that season and what’s that say about that metric?

Audige was 2nd best imo. Boo Buie was 1st team All big ten by the media and considered by almost everyone as the best player on that team. I would still take hobbled Barnhizer over a lot of players. He was still stuffing the stat sheet and being a lockdown defender.

Forget even trying. It’s hilarious how people can mine the internet to find something to support a ridiculous opinion, I have heard dumb arguments in my day but this one may take the cake. Boo was our best player in 23-24. End of story. Barney was the best player last year until he got hurt and then Nick took over. You can argue Nick was better the entirely of the year and I can accept that. However, when you publish rankings saying Big Matt and KJ were stronger contributors last year than either of them then you are on some serious mind altering drugs. I could care less which so called stats guru is quoted. If anyone truly believes this I would highly question their basketball acumen or motivation for making than comment. Then you frame the argument that we don’t acknowledge Big Matt or KJ’s contribution because we don’t bow to some “unbiased” metrics. I can’t wait for three more years of being told by PWB how he was the first dude to recognize KJ’s talent and how everyone else thought he stunk. Of course, that is as nonsensical as this whole conversation.

Thats a terrible argument regarding Audige and Buie.
The coaches in the Big Ten put both Audige and Buie as 2nd team All Conference.
So you say Buie was clearly the best player on the team. Well not according to the coaches.
Or Miyakawa's numbers.
I thought Buie was better, but the evidence challenges my opinion. And yours.

Ignoring Barnhizer's play when injured (which hurt the team) is foolish and obviously the result of personal bias.
The stats guys don't have that bias and their numbers include how badly Barnhizer played at the end.
If you want to say "When Barnhizer was healthy, he was the best player on the team" sure, go for it.

As for Martinelli, he was not an efficient scorer, not a good passer, did not stand out defensively and he played a ton of minutes.
It shouldn't really surprise anyone that other players would be more effective carrying less of the load.

These great metrics give us all a way to compare players across different teams and seasons.
Thats why I presented all that information above - to cast the bias aside.

Thankfully Chris Collins has started using the metrics in his decision-making.

I can’t believe you insist on finding ways to denigrate Boo, Barney and Nicky in your feeble attempt to be the smartest “analyst” on the board with your unending praise of your two boys Matt and KJ. Oh, I personally don’t give a flying **** what your other favorite Miyakawa’s numbers say! He rating system clearly sucks if Nick was rated as our third best player. Ripping him for not being an efficient scorer is comical. Well, he was doubled most of the season and had many teammates that were either reluctant to shot of lacked the skill set to score. I won’t mention the names because it’s so obvious which players I am talking about. He’s not a good passer. The best chance of NU scoring was Nick Shooting. Being doubled or tripled each trip may have something to do with that. Then you nit pick his defense. Shows how stupid this rating system is where the player gets hammered and dropped below players because he is forced to carry the load. Big Matt and KJ played well for the most part. They both played poorly at times too. This rating system clearly favors support players because they are less of a focal point!

Thank God Collins starting using these Miyakawa metrics in his decision making and was smart enough to hire the Wizard as his defensive guru! I not sure how he ever functioned without either.

 

[hdhntr1]

I am not buying into any rating system that claims Big Matt’s 24-25 season contributed more to the team than Boo’s 23-24 season. In fact, how on Earth is KJ’s rating higher than Nicky buckets last year.

There definitely seems to be some flaws in this.

 

[CoralSpringsCat]

Forget even trying. It’s hilarious how people can mine the internet to find something to support a ridiculous opinion, I have heard dumb arguments in my day but this one may take the cake. Boo was our best player in 23-24. End of story. Barney was the best player last year until he got hurt and then Nick took over. You can argue Nick was better the entirely of the year and I can accept that. However, when you publish rankings saying Big Matt and KJ were stronger contributors last year than either of them then you are on some serious mind altering drugs. I could care less which so called stats guru is quoted. If anyone truly believes this I would highly question their basketball acumen or motivation for making than comment. Then you frame the argument that we don’t acknowledge Big Matt or KJ’s contribution because we don’t bow to some “unbiased” metrics. I can’t wait for three more years of being told by PWB how he was the first dude to recognize KJ’s talent and how everyone else thought he stunk. Of course, that is as nonsensical as this whole conversation.



I can’t believe you insist on finding ways to denigrate Boo, Barney and Nicky in your feeble attempt to be the smartest “analyst” on the board with your unending praise of your two boys Matt and KJ. Oh, I personally don’t give a flying **** what your other favorite Miyakawa’s numbers say! He rating system clearly sucks if Nick was rated as our third best player. Ripping him for not being an efficient scorer is comical. Well, he was doubled most of the season and had many teammates that were either reluctant to shot of lacked the skill set to score. I won’t mention the names because it’s so obvious which players I am talking about. He’s not a good passer. The best chance of NU scoring was Nick Shooting. Being doubled or tripled each trip may have something to do with that. Then you nit pick his defense. Shows how stupid this rating system is where the player gets hammered and dropped below players because he is forced to carry the load. Big Matt and KJ played well for the most part. They both played poorly at times too. This rating system clearly favors support players because they are less of a focal point!

Thank God Collins starting using these Miyakawa metrics in his decision making and was smart enough to hire the Wizard as his defensive guru! I not sure how he ever functioned without either.

Put this post in the WR HOF!

 

[AdamOnFirst]

Forget even trying. It’s hilarious how people can mine the internet to find something to support a ridiculous opinion, I have heard dumb arguments in my day but this one may take the cake. Boo was our best player in 23-24. End of story. Barney was the best player last year until he got hurt and then Nick took over. You can argue Nick was better the entirely of the year and I can accept that. However, when you publish rankings saying Big Matt and KJ were stronger contributors last year than either of them then you are on some serious mind altering drugs. I could care less which so called stats guru is quoted. If anyone truly believes this I would highly question their basketball acumen or motivation for making than comment. Then you frame the argument that we don’t acknowledge Big Matt or KJ’s contribution because we don’t bow to some “unbiased” metrics. I can’t wait for three more years of being told by PWB how he was the first dude to recognize KJ’s talent and how everyone else thought he stunk. Of course, that is as nonsensical as this whole conversation.



I can’t believe you insist on finding ways to denigrate Boo, Barney and Nicky in your feeble attempt to be the smartest “analyst” on the board with your unending praise of your two boys Matt and KJ. Oh, I personally don’t give a flying **** what your other favorite Miyakawa’s numbers say! He rating system clearly sucks if Nick was rated as our third best player. Ripping him for not being an efficient scorer is comical. Well, he was doubled most of the season and had many teammates that were either reluctant to shot of lacked the skill set to score. I won’t mention the names because it’s so obvious which players I am talking about. He’s not a good passer. The best chance of NU scoring was Nick Shooting. Being doubled or tripled each trip may have something to do with that. Then you nit pick his defense. Shows how stupid this rating system is where the player gets hammered and dropped below players because he is forced to carry the load. Big Matt and KJ played well for the most part. They both played poorly at times too. This rating system clearly favors support players because they are less of a focal point!

Thank God Collins starting using these Miyakawa metrics in his decision making and was smart enough to hire the Wizard as his defensive guru! I not sure how he ever functioned without either.

Police, I'd like to report a murder

(these numbers are dumb and useless)

 

[prez77]

I don't know his methodology but from the figures and players, I'm guessing his method gives defense the same value as offense and allocates the players' individual scores within each, defense and offense separately. Interpreting these stats goes to whether we think defensive ability is as important as offensive ability. As fans, we tend to focus on offense, in part because making a basket for our team is more pleasing than the other team missing a shot, which we sort of take for granted. But the opponent's missed shot helps us win just as much as our made shot helps us win.

In a football game, we tend to credit the defense a lot. In 1995, our offense was very so-so overall. The defense was All World and the reason we got to the Rose Bowl. Period.

If offense fails just because the offensive player fails, and the defender didn't have much to do with it, then Miyakawa's method is flawed.
If the defender causes shots to be missed as much as our player causes his own shots to be made, then Miyakawa may not be flawed.

We know a great pitcher completely shuts down an offense sometimes and we credit him. We know the '85 Bears defense and the '95 Cats defenses carried those teams. What's up with basketball?

 

[ThatkidfromHolland]

I don't know his methodology but from the figures and players, I'm guessing his method gives defense the same value as offense and allocates the players' individual scores within each, defense and offense separately. Interpreting these stats goes to whether we think defensive ability is as important as offensive ability. As fans, we tend to focus on offense, in part because making a basket for our team is more pleasing than the other team missing a shot, which we sort of take for granted. But the opponent's missed shot helps us win just as much as our made shot helps us win.

In a football game, we tend to credit the defense a lot. In 1995, our offense was very so-so overall. The defense was All World and the reason we got to the Rose Bowl. Period.

If offense fails just because the offensive player fails, and the defender didn't have much to do with it, then Miyakawa's method is flawed.
If the defender causes shots to be missed as much as our player causes his own shots to be made, then Miyakawa may not be flawed.

We know a great pitcher completely shuts down an offense sometimes and we credit him. We know the '85 Bears defense and the '95 Cats defenses carried those teams. What's up with basketball?

Audige is a better defender than Berry. This stat ranked Berry noticeably higher despite Audige winning DPOY and being one of the best perimeter defenders I’ve ever seen at NU.

Any stat that ranks Martinelli as the 5th best player last year is wrong. Ranking Berry above Buie at any point is wrong. This stat is very flawed. I watch defense. I value defense. It’s why I’m patient with calling for minutes for the freshmen and why I didn’t think Young was the best thing since sliced bread. This is just a bad stat if it’s meant to measure player impact.

 

[prez77]

We were 13th in conference play in defense and 16th in scoring last year.
A point not scored against you is as valuable as a point you score.
If defense has about the same role in whether a point is scored in basketball as it does in football or baseball, then
we had more value for the rating system to allocate on the defensive end of the court than we did on the offensive in valuing all the players.
This leaves room for some non-scorers to actually be more valuable than the conventional box score would suggest.

I won't defend Miyakawa's system or what it produced per se because I didn't look up what it is. When you play your minutes and who you played them against matters and I don't know how he handled that.
But I will say sometimes looking at the whole system as it works in combination, reveals things that looking at a particular piece would not necessarily reveal.

 

[hdhntr1]

Police, I'd like to report a murder

(these numbers are dumb and useless)

It was justifiable homicide

 

[hdhntr1]

Audige is a better defender than Berry. This stat ranked Berry noticeably higher despite Audige winning DPOY and being one of the best perimeter defenders I’ve ever seen at NU.

Any stat that ranks Martinelli as the 5th best player last year is wrong. Ranking Berry above Buie at any point is wrong. This stat is very flawed. I watch defense. I value defense. It’s why I’m patient with calling for minutes for the freshmen and why I didn’t think Young was the best thing since sliced bread. This is just a bad stat if it’s meant to measure player impact.

Over all in Audige's last year, he and Berry were rated almost identical. Audige was ahead by a narrow margin. Audige was more at the point guarding PGs and shooting guards and more visible with blocked passes and steals but both were very important to the team vut Audige was more visible and Berry was somewhat inconsistent offensively.

Last year Berry was rated higher than Audige(in his final year the year before) and was more consistent before he went down vs Nebraska I believe. He was having a breakout season before the injury

 

[ThatkidfromHolland]

Over all in Audige's last year, he and Berry were rated almost identical. Audige was ahead by a narrow margin. Audige was more at the point guarding PGs and shooting guards and more visible with blocked passes and steals but both were very important to the team vut Audige was more visible and Berry was somewhat inconsistent offensively.

Last year Berry was rated higher than Audige(in his final year the year before) and was more consistent before he went down vs Nebraska I believe. He was having a breakout season before the injury

Take a look at the post showing the defensive ratings for Audige and Berry from the 2022–23 season, Berry’s was 0.4 better. Defense is very visible, you just have to see it on film not on a box score. Audige was flat-out one of the best perimeter defenders in the country that year. He had real next-level ability on that end. The instincts, the length, the foot speed, the footwork, the strength, the hands, it was all elite for a guard. It genuinely felt like the basketball gods put together the perfect defensive guard. He was special. We can complain about inefficiency on the offensive end and lack of court vision but on defense, dude was just different.

Audige is exhibit A of the adjusted +\- not measuring defense. Should be for this stat too.

 

[PurpleWhiteBoy]

Do you think there were 3 better players than Martinelli on last years team?

If this changed your opinion on Audige how’d he score on your adjusted +\- that season and what’s that say about that metric?

Audige was 2nd best imo. Boo Buie was 1st team All big ten by the media and considered by almost everyone as the best player on that team. I would still take hobbled Barnhizer over a lot of players. He was still stuffing the stat sheet and being a lockdown defender.

On offense, Martinelli was our best scorer after Barnhizer got hurt.
On defense, Martinelli was our 4th or 5th best defender behind Nicholson, Barnhizer, Berry and Windham.

Arguably Ciaravino was also a better defender than Martinelli in the limited minutes he played.
Leach was not a good defender.

Anything else you want to know?

As for my metrics, I like them and they're good. Your question about Audige is weird. I'm open-minded about this stuff.
I don't think you are at all. You criticize without understanding and nothing you write is constructive, so you become a bore.
If Miyakawa or Torvik produces numbers that don't generally align with mine, I try to figure out why, rather than rant about how useless they are or whatever it is that you're doing.

You have to get past this arrogant, misguided notion that you know more than the numbers.

 

[ThatkidfromHolland]

On offense, Martinelli was our best scorer after Barnhizer got hurt.
On defense, Martinelli was our 4th or 5th best defender behind Nicholson, Barnhizer, Berry and Windham.

Arguably Ciaravino was also a better defender than Martinelli in the limited minutes he played.
Leach was not a good defender.

Anything else you want to know?

As for my metrics, I like them and they're good. Your question about Audige is weird. I'm open-minded about this stuff.
I don't think you are at all. You criticize without understanding and nothing you write is constructive, so you become a bore.
If Miyakawa or Torvik produces numbers that don't generally align with mine, I try to figure out why, rather than rant about how useless they are or whatever it is that you're doing.

You have to get past this arrogant, misguided notion that you know more than the numbers.

KJ wasn’t a better defender early on, but by the end of the season he was comparable, which is really impressive considering we are comparing him to a guy in his third year in a very tough defensive system to master. I’m with you on KJ’s potential; it just wasn’t producing early, and he still needs to find more consistency as a scorer and playmaker.

I’ll agree with you that the rest of the group were better defenders overall. But Martinelli is the best pure scorer among them, and the second-best rebounder behind a healthy Barnhizer. Martinelli had the best season of anyone last year, and it honestly wasn’t close. Not ranking him 1–2 overall, even when factoring in defense, is a miss. He was 2nd team all big ten for crying out loud.

On Gelo: no, not close right now. If we’re comparing freshman years, then yes, Gelo has the tools athletically. He’s clearly coachable defensively; he was the only player I saw actually implementing the specific defensive strategy Collins wanted against Ace according to the press conference. Gelo just didn’t have the length to really bother him.

Also, I’d love to know what your +/- system ranked Audige at in 2022–23 compared to the rest of that team. You’ve been a big supporter of that stat for years. If this current model is producing drastically different results, how can both supposedly measure player impact so well?

I’m not saying I “know more” than the numbers. I’m saying these numbers don’t capture player impact, especially on defense, particularly well. Anytime you scale stats to minutes played, you introduce error. There are simply too many unquantifiable variables in basketball. Stats are best used to support your opinions, or help you figure out why reality and the numbers don’t line up.

Like with Matt’s rebounding his individual rebounding numbers were lower because he was often boxing out to let Martinelli or Barnhizer grab the board. If you don't understand what the numbers are actually measuring, you’re going to misinterpret them. The key is critical thinking: numbers are a tool, not the final answer. Context and understanding how the numbers are built is everything.

 

[prez77]

I use statistics for a living so I will weigh in here. There are many problems in the world where multiple causes, some readily observable and some not, create an outcome - scoring a basketball point, or winning a basketball game for instance. Economists, medical researchers, educators etc. use complex statistics to identify relationships that are not readily observable. With many common medicines, we don't know exactly why they work, but they are associated with the outcome of interest and so if safe, the FDS approves them. The whole point of statistics is to identify an interaction not easy, or perhaps impossible, to see. So the numbers are often valuable and sometimes tell us surprising things because we can't see them or because human beings are easily biased because of what is easy to see or what we would like to see. Sometimes placebos make people better - even though they aren't physically affecting the underlying condition.

On the other hand, there are causes of things data sets don't pick up very well. There isn't any data analysis in any sport I know of that accounts for a player's injury, or whether they got a proper night's sleep before a game. Football line play and basketball defense have been very difficult to measure statistically because their play is so interdependent and the measurable actions hard to isolate without watching very intensively - as their coaches and analysts do. So some deference to the simple eye test becomes necessary as a practical matter.

We also try to validate our findings by comparing them to something that we think we know - does an unusual statistical finding run impossibly counter to what we think we know? Validation can be hard when we learn something completely new - what to compare it with?

I have come to think many inferential statistics should be used generally and directionally, rather than relying exactly on what the equation says the relationship is - because there are so many sources of error and happenstance in almost any data set. I would hesitate to use data analysis to absolutely order how effectively each of our players plays defense, but I would use it to say this guy seems to be pretty good, or maybe even really good, and it looks like this guy is not so good. And those results may be a bit different from the eye test sometimes. The numbers I've seen here have altered my view of Big Matt - I think he was making a much bigger difference on the defensive side than the eye test may reveal, and hence was a much more valuable asset than the conventional stats would indicate.

 

[GatoLouco]

I don't do statistics, but I do numbers for a living.

I've never seen a model that accounts for every little nuance of a business. I believe that is the same in sports. And that's why people attack them: "what about XYZ", "what happens when ABC?".

It's not like finding a flaw invalidates a model. And frankly I've lost count on how many times I've seen people dismiss it just because they pointed some instance the model does not do well. In the end, in most cases, all they are doing is trying to validate their own belief, or some variation of that, like doing things like they've always been done, avoiding learning something new...

A flaw in a model is often just an opportunity to improve it, or a caution to validate its findings. It is not necessarily a reason to toss it.

I'm sure all the Pom's, Torvik's, MiyaKawa's models have a lot of blind spots. I am also sure they know that and have worked to minimize it. Those are not useless models, they are very good sources to question our (perhaps lying) eyes. Proof of that is how widely they are used.

 

[ThatkidfromHolland]

I don't do statistics, but I do numbers for a living.

I've never seen a model that accounts for every little nuance of a business. I believe that is the same in sports. And that's why people attack them: "what about XYZ", "what happens when ABC?".

It's not like finding a flaw invalidates a model. And frankly I've lost count on how many times I've seen people dismiss it just because they pointed some instance the model does not do well. In the end, in most cases, all they are doing is trying to validate their own belief, or some variation of that, like doing things like they've always been done, avoiding learning something new...

A flaw in a model is often just an opportunity to improve it, or a caution to validate its findings. It is not necessarily a reason to toss it.

I'm sure all the Pom's, Torvik's, MiyaKawa's models have a lot of blind spots. I am also sure they know that and have worked to minimize it. Those are not useless models, they are very good sources to question our (perhaps lying) eyes. Proof of that is how widely they are used.

It’s not just one flaw in this model, unfortunately. If the goal is to “tell us which players were most important to each of the most recent Northwestern teams,” it doesn’t accomplish that very well. For example, the model ranks Nick Martinelli as the fifth-best player last season, suggesting his impact was comparable to Blake Preston’s, a conclusion that clearly doesn’t align with reality. It also ranks Boo Buie as only the third-best player in 2022–23 and places freshman KJ Windham ahead of that version of Buie another result that simply doesn't pass the eye test.

I work with analytics and numbers every day, and a huge part of my job is not just reporting numbers, but explaining the context behind them. I’m not saying these numbers are useless, there is definitely value here, but I don’t think they measure what PWB claims they are measuring particularly well. Without knowing exactly what inputs and weightings went into these calculations, it’s hard to say exactly what the model is capturing accurately. But the conclusions being drawn from it don’t make much sense.

I’ve said the same thing about adjusted plus-minus in the past: it can mislead if it’s used carelessly, especially if someone tries to present it as a full measure of a player’s value or total impact. However, it’s still excellent at summarizing box score production. The key is knowing the limits of what a number actually tells you. I don't think either this model or adjusted plus-minus are great tools for fully measuring "overall impact."

Maybe in this case it's especially a Northwestern-specific issue, given the minute load on stars and the lack of depth on recent teams. Maybe it doesn’t draw context of what is on each players plate entirely and adjust their efficiency off of that. There’s many potential issues that lead to some shockingly inaccurate conclusions. That being said the number could very easily have some solid use. I just don’t think it’s measuring “overall impact.”

 

[AdamOnFirst]

It’s not just one flaw in this model, unfortunately. If the goal is to “tell us which players were most important to each of the most recent Northwestern teams,” it doesn’t accomplish that very well. For example, the model ranks Nick Martinelli as the fifth-best player last season, suggesting his impact was comparable to Blake Preston’s, a conclusion that clearly doesn’t align with reality. It also ranks Boo Buie as only the third-best player in 2022–23 and places freshman KJ Windham ahead of that version of Buie another result that simply doesn't pass the eye test.

I work with analytics and numbers every day, and a huge part of my job is not just reporting numbers, but explaining the context behind them. I’m not saying these numbers are useless, there is definitely value here, but I don’t think they measure what PWB claims they are measuring particularly well. Without knowing exactly what inputs and weightings went into these calculations, it’s hard to say exactly what the model is capturing accurately. But the conclusions being drawn from it don’t make much sense.

I’ve said the same thing about adjusted plus-minus in the past: it can mislead if it’s used carelessly, especially if someone tries to present it as a full measure of a player’s value or total impact. However, it’s still excellent at summarizing box score production. The key is knowing the limits of what a number actually tells you. I don't think either this model or adjusted plus-minus are great tools for fully measuring "overall impact."

Maybe in this case it's especially a Northwestern-specific issue, given the minute load on stars and the lack of depth on recent teams. Maybe it doesn’t draw context of what is on each players plate entirely and adjust their efficiency off of that. There’s many potential issues that lead to some shockingly inaccurate conclusions. That being said the number could very easily have some solid use. I just don’t think it’s measuring “overall impact.”

I'll only add to this, as somebody who has the lumps from the statistics and sabermetrics takeover in baseball and argued with anti numbers people for years and years: you can't just dump out a big matrix of numbers and assume they have any value in and of themselves just because they're a bunch of numbers, and you CERTAINLY can't expect anybody else to accept them if you can't articulate them. You can't just say "these numbers are good" and have it carry any meaning whatsoever, or even convey that you yourself understand them AT ALL, frankly.

Pythagorean win expectation is valuable because it correlates with future W-L MORE than plain W-L does. FIP and xFIp were better because they correlated with future ERA BETTER than actual ERA does. X forecast model is better than Y because it is based on these specific theoretically sound assumptions and then has a better track record than Y. A big wall of numbers being a big wall of numbers does nothing to demonstrate it has any value whatsoever, any idiot with a keyboard can write a big wall of numbers and claim it's gold... as we see on this board.