Statistical Impact of Players

[fluoxetine]

The results here will not be super popular, but I did it and I'm going to present it. I didn't create it to make any sort of point, and I didn't know what I would find when I started. A little bit about what this is:

The Raw Net Pts Created estimates the impact a player had on a game based on statistics. It is then normalized so that the impacts add up to the actual score differential in the game. This is NOT my work, I just collected this from Torvik's website for each game.

Avg Player Expected is the net points created that an average D1 player would expect in the minutes they played against that opponent (+/- home court). It is essentially a SOS, where a more negative value means stronger opponents. This I created/calculated myself.

Pts Above Avg is the difference between the Raw Net Pts Created and the expectation of an Avg D1 player.

Minutes is minutes.

Per 40 minutes is just Pts Above Avg normalized to 40 minutes.

Player	Raw Net Pts Created	Avg Player Expected	Pts Above Avg	Minutes	Per 40 Minutes
Cam Spencer	125.2​	-16.5​	141.7​	1071​	5.29​
Clifford Omoruyi	34.4​	-17.6​	52.0​	1031​	2.02​
Caleb McConnell	18.3​	-20.4​	38.7​	948​	1.63​
Paul Mulcahy	18​	-19.3​	37.3​	973​	1.53​
Mawot Mag	9.2​	-6.5​	15.7​	571​	1.10​
Aundre Hyatt	-9​	-12.2​	3.2​	786​	0.17​
Oskar Palmquist	-10.9​	-3.9​	-7.0​	191​	-1.47​
Antonio Chol	-6.6​	0.6​	-7.2​	19​	-15.17​
Antwone Woolfolk	-9.6​	-1.1​	-8.5​	235​	-1.44​
Dean Reiber	-13.1​	-2.2​	-10.9​	190​	-2.29​
Jalen Miller	-13.2​	0.9​	-14.1​	141​	-4.01​
Derek Simpson	-22.9​	-8.5​	-14.4​	684​	-0.84​

 

[fluoxetine]

Here is for conference games only (including conf tourney):

Player	Raw Net Pts Created	Avg Player Expected	Pts Above Avg	Minutes	Per 40 Minutes
Cam Spencer	46.5​	-19.9​	66.4​	685​	3.88​
Clifford Omoruyi	8​	-20.0​	28.0​	680​	1.65​
Paul Mulcahy	5.8​	-21.7​	27.5​	745​	1.48​
Caleb McConnell	0.8​	-20.8​	21.6​	721​	1.20​
Mawot Mag	5.8​	-8.6​	14.4​	302​	1.90​
Aundre Hyatt	-5.2​	-14.1​	8.9​	488​	0.73​
Oskar Palmquist	-7​	-4.5​	-2.5​	163​	-0.62​
Derek Simpson	-16.5​	-11.6​	-4.9​	392​	-0.50​
Jalen Miller	-6.6​	-0.9​	-5.7​	40​	-5.74​
Antwone Woolfolk	-10.1​	-3.1​	-7.0​	97​	-2.90​
Dean Reiber	-13.1​	-3.3​	-9.8​	113​	-3.47​

 

[NickRU714]

JK. Interesting. I'm sure many will be turned off at just the title.

But the eye test and injuries and........

 

[PSAL_Hoops]

Here is for conference games only (including conf tourney):

Player	Raw Net Pts Created	Avg Player Expected	Pts Above Avg	Minutes	Per 40 Minutes
Cam Spencer	46.5​	-19.9​	66.4​	685​	3.88​
Clifford Omoruyi	8​	-20.0​	28.0​	680​	1.65​
Paul Mulcahy	5.8​	-21.7​	27.5​	745​	1.48​
Caleb McConnell	0.8​	-20.8​	21.6​	721​	1.20​
Mawot Mag	5.8​	-8.6​	14.4​	302​	1.90​
Aundre Hyatt	-5.2​	-14.1​	8.9​	488​	0.73​
Oskar Palmquist	-7​	-4.5​	-2.5​	163​	-0.62​
Derek Simpson	-16.5​	-11.6​	-4.9​	392​	-0.50​
Jalen Miller	-6.6​	-0.9​	-5.7​	40​	-5.74​
Antwone Woolfolk	-10.1​	-3.1​	-7.0​	97​	-2.90​
Dean Reiber	-13.1​	-3.3​	-9.8​	113​	-3.47​

Hate to do this to you because it’s probably a lot of work, but you kind of would need to normalize by including only the conference games, Temple, Hofstra, Wake, SHU, and Miami (the 5 “real” games played outside the conference). At least with respect to Caleb’s data - No?

This analysis is not fair to him otherwise because what you presented provides a false narrative. His “drop off” vs conference opponents is mostly random differentiation since he barely played any cupcakes.

I guess what I’m trying to say is the swap in output Mag vs Caleb is likely just the difference between Caleb having a good game vs. Miami (which you stripped out), but Mag’s 15 point game vs OSU counts.

 

[fluoxetine]

Hate to do this to you because it’s probably a lot of work, but you kind of would need to normalize by including only the conference games, Temple, Hofstra, Wake, SHU, and Miami (the 5 “real” games played outside the conference). At least with respect to Caleb’s data - No?

This analysis is not fair to him otherwise because what you presented provides a false narrative. His “drop off” vs conference opponents is mostly random differentiation since he barely played any cupcakes.

I guess what I’m trying to say is the swap in output Mag vs Caleb is likely just the difference between Caleb having a good game vs. Miami (which you stripped out), but Mag’s 15 point game vs OSU counts.

Player	Raw Net Pts Created	Avg Player Expected	Pts Above Avg	Minutes	Per 40 Minutes
Cam Spencer	65.2​	-22.7​	87.9​	851​	4.13​
Mawot Mag	7.6​	-10.9​	18.5​	413​	1.79​
Clifford Omoruyi	11.3​	-22.7​	34.0​	837​	1.63​
Paul Mulcahy	7.7​	-23.0​	30.7​	855​	1.44​
Caleb McConnell	5.6​	-23.3​	28.9​	860​	1.34​
Aundre Hyatt	-6.9​	-16.6​	9.7​	624​	0.62​
Derek Simpson	-29.6​	-13.1​	-16.5​	510​	-1.30​
Oskar Palmquist	-10.6​	-4.5​	-6.1​	170​	-1.43​
Dean Reiber	-13.3​	-3.8​	-9.5​	133​	-2.87​
Antwone Woolfolk	-18​	-3.7​	-14.3​	141​	-4.06​
Jalen Miller	-7.5​	-1.2​	-6.3​	59​	-4.25​

 

[Greene Rice FIG]

What type of defense is factored in to these numbers? Is defense represented by blocks and steals or is there some type of PPP when on the floor used?

 

[fluoxetine]

What type of defense is factored in to these numbers? Is defense represented by blocks and steals or is there some type of PPP when on the floor used?

Basically blocks and steals will count for something and then any residual defense will get spread across the players that are playing.

 

[Greene Rice FIG]

Basically blocks and steals will count for something and then any residual defense will get spread across the players that are playing.

So defense is just represented by blocks and steals. So it overrates Spencer and underrates Mag SIGNIFICANTLY on the defensive end.

If this is the case 50% of the output is clutter.

There is very little to no or even a negative correlation to steals and defense. I think there is a positive correlation with blocks. I don't have any data to support this though.

 

[ImBadRU]

Avg points per ncaa player by years of experience?

 

[fluoxetine]

So defense is just represented by blocks and steals.

Not really. It's normalized to the actual score differential of the game. So defense will come through even if there are no blocks or steals. But it will have trouble differentiating between the players on the floor in this case.

So it overrates Spencer and underrates Mag SIGNIFICANTLY on the defensive end.

Probably.

If this is the case 50% of the output is clutter.

Don't agree.

There is very little to no or even a negative correlation to steals and defense. I think there is a positive correlation with blocks. I don't have any data to support this though.

I doubt this. And if it were true then steals wouldn't count for anything (or would count for a negative amount) in these kinds of regression based models anyway.

 

[Greene Rice FIG]

If there is positive correlation between steals and a player's defense it wouldn't be large.

Obviously the steal is effective as it terminates the defensive possession and probably increases the expected offensive point per possession of the next possession.

The question is how and why did the player get the steal. Does the player gamble too much and get beat often? How many steals occur when the defensive player is behind the offensive player? How many steals come as a result of plays made by others or is there a player like Cliff that allows perimeter players to gamble

I assume your numbers include foul rate so the offset of extra fouls from getting steals is represented

 

[fluoxetine]

If there is positive correlation between steals and a player's defense it wouldn't be large.

Obviously the steal is effective as it terminates the defensive possession and probably increases the expected offensive point per possession of the next possession.

The question is how and why did the player get the steal. Does the player gamble too much and get beat often? How many steals occur when the defensive player is behind the offensive player? How many steals come as a result of plays made by others or is there a player like Cliff that allows perimeter players to gamble

I assume your numbers include foul rate so the offset of extra fouls from getting steals is represented

Yes, fouls are included. The (average) impact of a steal should be correctly represented. Does the model miss the non-steal/block/foul things that players individually contribute on defense? Yes (although they still get some credit for them if it stops the opponent from scoring. But that credit may be unfairly split with other players). But it shouldn't be overvaluing steals/blocks/fouls.

 

[Greene Rice FIG]

How is “if it stops defense from scoring” represented?

 

[Scangg]

Caleb and Audige were top 2 in steals and the 2 best defenders in the league

You can say steals are overrated, but implying steals mean you're somehow a worse defender is weird. It's usually players who are locked in defensively and do a good job of reading the game and anticipating the offense

 

[Greene Rice FIG]

YOU ARE NOT IN A POSITION TO CALL ANYTHING WEIRD! I HAVE PHOTO EVIDENCE.

Saying steals is somewhere between slightly positive and slightly negative correlated with defense is not saying players with more steals are worse.

Are Caleb and Chase great defenders? YES
Was Scottie and MJ great defenders? YES
Is Spencer a very good defender? ABSOLUTELY NOT
Was Myles Mack a good defensive player? NO
Is Mawot Mag an average defensive player? ABSOLUTELY NOT
How about Allen Iverson?

 

[Greene Rice FIG]

The greek God of Stats weigh in...

Pamela Poker | Your One Stop Gaming Site

Pamela Poker | Top Tier 24/7 Entertainment, Best of its Class

greekgodofstats.com

I realize i have used something that references the NBA. I will skip dessert tonight.

 

[Scangg]

YOU ARE NOT IN A POSITION TO CALL ANYTHING WEIRD! I HAVE PHOTO EVIDENCE.

Saying steals is somewhere between slightly positive and slightly negative correlated with defense is not saying players with more steals are worse.

Are Caleb and Chase great defenders? YES
Was Scottie and MJ great defenders? YES
Is Spencer a very good defender? ABSOLUTELY NOT
Was Myles Mack a good defensive player? NO
Is Mawot Mag an average defensive player? ABSOLUTELY NOT
How about Allen Iverson?

Is it possible to be good at steals without being a good all around defender in all aspects? Yes

Spencer is an example. I would say his ability to read the game, get into passing lanes, and anticipation is pretty elite. His on ball man to man defense is not good due to lack of footspeed and explosiveness for lateral movement

 

[ImBadRU]

Spot on re cam

 

[RUJMM78]

Here is for conference games only (including conf tourney):

Player	Raw Net Pts Created	Avg Player Expected	Pts Above Avg	Minutes	Per 40 Minutes
Cam Spencer	46.5​	-19.9​	66.4​	685​	3.88​
Clifford Omoruyi	8​	-20.0​	28.0​	680​	1.65​
Paul Mulcahy	5.8​	-21.7​	27.5​	745​	1.48​
Caleb McConnell	0.8​	-20.8​	21.6​	721​	1.20​
Mawot Mag	5.8​	-8.6​	14.4​	302​	1.90​
Aundre Hyatt	-5.2​	-14.1​	8.9​	488​	0.73​
Oskar Palmquist	-7​	-4.5​	-2.5​	163​	-0.62​
Derek Simpson	-16.5​	-11.6​	-4.9​	392​	-0.50​
Jalen Miller	-6.6​	-0.9​	-5.7​	40​	-5.74​
Antwone Woolfolk	-10.1​	-3.1​	-7.0​	97​	-2.90​
Dean Reiber	-13.1​	-3.3​	-9.8​	113​	-3.47​

Spencer was the most valuable player because he was the best shooter and got the most attention from opponents.Rutgers was woeful on offense and would have been even worse if Spencer wasn't on the court.The debate between offense and defense impact will always have pro/con points of view.

 

[Greene Rice FIG]

Spencer was the most valuable player because he was the best shooter and got the most attention from opponents.Rutgers was woeful on offense and would have been even worse if Spencer wasn't on the court.The debate between offense and defense impact will always have pro/con points of view.

1. Cliff
2. Mag

I almost think this is not debatable.

 

[BillyC80]

I would love to see each player’s stats in two more of these charts: one for just games we won, and one for just games we lost..

 

[fluoxetine]

The greek God of Stats weigh in...

Pamela Poker | Your One Stop Gaming Site

Pamela Poker | Top Tier 24/7 Entertainment, Best of its Class

greekgodofstats.com

I realize i have used something that references the NBA. I will skip dessert tonight.

I mean this guy is arguing that steals are meaningless by saying that steals ONLY have a .543 correlation with points allowed.

.543 is pretty high

 

[Greene Rice FIG]

I mean this guy is arguing that steals are meaningless by saying that steals ONLY have a .543 correlation with points allowed.

.543 is pretty high

Are you doubting the GGOS

 

[Greene Rice FIG]

I mean this guy is arguing that steals are meaningless by saying that steals ONLY have a .543 correlation with points allowed.

.543 is pretty high

Reading the article more. Definitely wouldnt recommend taking any statistics courses (even 101) in Greece

 

[PSAL_Hoops]

Steals are one indicator. But there are others. What’s not being factored in is that certain skills can only be compensated in certain ways.

For example, prior to Mag’s injury, Pike utilized the press with Mag on the floor for pretty much every possession that Caleb was on the bench for. Mag is a master at that press and we were able to run it effectively with or without Caleb on the floor. So when you look at points allowed with Caleb out of the game, I’d speculate those numbers are very deceiving.

Why? Because we didn’t have remotely close to the depth that would be needed to deploy a full time pressing scheme. So you can’t just assume those minutes are interchangeable because we had to revert to halfcourt schemes at some point. Just as Mag made the press what it was - the same could be said of Caleb in the halfcourt sets. Having Mag as a complement made us near unstoppable, but no set of statistics can do justice to the importance of Caleb in defending the perimeter last season. The eliteness of Mag’s halfcourt defense will be put to the test next season. He and Caleb are not interchangeable. They excel at different things.