Advanced Metrics Nonsense

#77
#77
PKT_VOL said:
As to the other two articles, rebounding, defense, and passing/court vision/unselfishness are all more important than shooting/scoring abilities.

However, the NBA is more entertainment than a real game.
Click to expand...

Winning is how you put butts in the hometown seats. The data is clear. Tim Duncan proved you can be a very boring winner in a small market and still fill up the arena.
 
#80
#80
milohimself said:
Baker how do you feel about the Hollinger PER rankings
Click to expand...

It's a joke. It's not based on any sort of mathematical relationship between variables. It's just arbitrary. You can improve your PER with every bucket scored as long as you shoot 33% (it may be 30%, I just know it's ridiculously low). This means you can shoot so poorly that you destroy your team's chances of being competitive, while Hollinger's formula sings your praises (Montae Ellis). It's nonsensical.
 
#82
#82
n_huffhines said:
Winning is how you put butts in the hometown seats. The data is clear. Tim Duncan proved you can be a very boring winner in a small market and still fill up the arena.
Click to expand...

I am the biggest Wake basketball fan you will ever meet. Tim Duncan is my favorite basketball player ever. He is the reason why I started to play basketball at an early age.

Although he has had a lot of team and individual success (MVP's and is considered the best power forward to play the game), most GM's would pass over him if they were to build a team. If you can get an equally "good" player that is "flashy" instead of "Mr. Fundamental", they are going to build their team around the "flashy" guy every time.
 
#83
#83
PKT_VOL said:
How do you defend them counting those wins and loses per player? That is beyond suspect to me.
Click to expand...

It's the Wins Produced stat, and it's based on 7 statistical categories: scoring efficiency, rebounding, assists, steals, blocks, fouls, and turnovers. They have a strong relationship with winning. The model explains 97% of winning. Based on the team stats accumulated wins produced will tell you with 97% accuracy how many games Team X won. It's rarely exact, but it's usually within 2 games. You can take the individual stats of the players and derive how much they contributed to their team's total wins.
 
#84
#84
PKT_VOL said:
I am the biggest Wake basketball fan you will ever meet. Tim Duncan is my favorite basketball player ever. He is the reason why I started to play basketball at an early age.

Although he has had a lot of team and individual success (MVP's and is considered the best power forward to play the game), most GM's would pass over him if they were to build a team. If you can get an equally "good" player that is "flashy" instead of "Mr. Fundamental", they are going to build their team around the "flashy" guy every time.
Click to expand...

.
 
#85
#85
n_huffhines said:
It's the Wins Produced stat, and it's based on 7 statistical categories: scoring efficiency, rebounding, assists, steals, blocks, fouls, and turnovers. They have a strong relationship with winning. The model explains 97% of winning. Based on the team stats accumulated wins produced will tell you with 97% accuracy how many games Team X won. It's rarely exact, but it's usually within 2 games. You can take the individual stats of the players and derive how much they contributed to their team's total wins.
Click to expand...

See, the individual player part is etch a sketch (to me). I can see it in a sport like baseball. Basketball is too much of a team sport for that to fly with me.
 
#86
#86
FWIW, I believe the wins produced metric shows more correlation with winning than any model that has been developed for baseball.
 
#87
#87
I can buy statistical models for baseball. Basketball, not so much. Especially when we are talking about tying individual players to a specific number of wins or loses.
 
#88
#88
It's not for everybody, but it proves to be hella accurate.

A summation of all the individual wins produced will demonstrate a predicted number of wins for a team. Usually very accurate:

Houston
Predicted wins: 33.4
Actual: 34

The NBA Geek - Stats for Houston Rockets

Dallas
Predicted wins: 36.3
Actual: 36

The NBA Geek - Stats for Dallas Mavericks

Charlotte
Predicted wins: 25.5
Actual: 21

The NBA Geek - Stats for New Orleans Hornets

Atlanta
Predicted wins: 40.6
Actual: 40

The NBA Geek - Stats for Atlanta Hawks

* I chose these 4 examples at random.
 
#90
#90
Well, there is randomness no matter what*. A shortened season results in a little more randomness. Even still, do you think there's any scout that can look at their roster and accurately tell you within 4.5 games how many Charlotte will win next year?

*This is usually the result of a team winning a large amount of their close games, or losing a large amount of their close games. When we deal with averages, we have to predict the close games will average out evenly. If you only have 12 close games in a season it's somewhat likely you win a lot more or a lot less than half of those. If you have 40 close games in a season, it's a lot less likely to be skewed one way or the other.

Elaborating...relating this all to football, there are some interesting stats that I think are a testament to Peyton Manning's quality as a QB. Based on point differential, you can somewhat accurately predict how many games a team wins. For like 9 (?) straight seasons Manning's Colts outperformed their predicted record. What this means is for 9 straight seasons Manning won most of the close ones. He is an incredible game manager. Elway was the only other QB who had done it 7 seasons in their career. Montana only did it like 4 times.
 
#91
#91
I was just kidding about charlotte. I'll probably go see them play some this year, they just suck something awful.

That football stat sounds interesting.
 
#93
#93
Shaq says Brooke Lopez is better than Dwight Howard. Almost nobody would agree, but the fact that Shaq thinks he can get away with saying that demonstrates basketball establishment is stupid. Lopez is a bad player. I don't understand it. There is nothing good about his game. He's athletic and huge, so he just "looks" right. He can't rebound. He can't bang. And he scores efficiently inside, but refuses to do that, shooting 15' jumpers all day. That guy got a max contract.
 
#94
#94
n_huffhines said:
Shaq says Brooke Lopez is better than Dwight Howard. Almost nobody would agree, but the fact that Shaq thinks he can get away with saying that demonstrates basketball establishment is stupid. Lopez is a bad player. I don't understand it. There is nothing good about his game. He's athletic and huge, so he just "looks" right. He can't rebound. He can't bang. And he scores efficiently inside, but refuses to do that, shooting 15' jumpers all day. That guy got a max contract.
Click to expand...

I'm sure it has nothing to do with Shaq egging on Howard because he stole his gimmick.
 
#95
#95
n_huffhines said:
It's not for everybody, but it proves to be hella accurate.

A summation of all the individual wins produced will demonstrate a predicted number of wins for a team. Usually very accurate:

Houston
Predicted wins: 33.4
Actual: 34

The NBA Geek - Stats for Houston Rockets

Dallas
Predicted wins: 36.3
Actual: 36

The NBA Geek - Stats for Dallas Mavericks

Charlotte
Predicted wins: 25.5
Actual: 21

The NBA Geek - Stats for New Orleans Hornets

Atlanta
Predicted wins: 40.6
Actual: 40

The NBA Geek - Stats for Atlanta Hawks

* I chose these 4 examples at random.
Click to expand...

n_huffhines said:
Well, there is randomness no matter what*. A shortened season results in a little more randomness. Even still, do you think there's any scout that can look at their roster and accurately tell you within 4.5 games how many Charlotte will win next year?

*This is usually the result of a team winning a large amount of their close games, or losing a large amount of their close games. When we deal with averages, we have to predict the close games will average out evenly. If you only have 12 close games in a season it's somewhat likely you win a lot more or a lot less than half of those. If you have 40 close games in a season, it's a lot less likely to be skewed one way or the other.

Elaborating...relating this all to football, there are some interesting stats that I think are a testament to Peyton Manning's quality as a QB. Based on point differential, you can somewhat accurately predict how many games a team wins. For like 9 (?) straight seasons Manning's Colts outperformed their predicted record. What this means is for 9 straight seasons Manning won most of the close ones. He is an incredible game manager. Elway was the only other QB who had done it 7 seasons in their career. Montana only did it like 4 times.
Click to expand...

Interesting. You should post the new predictions for the upcoming season so we can track it as the season progresses.

I will say, team statistics make more sense than individual player wins/loses.
 
#96
#96
n_huffhines said:
Shaq says Brooke Lopez is better than Dwight Howard. Almost nobody would agree, but the fact that Shaq thinks he can get away with saying that demonstrates basketball establishment is stupid. Lopez is a bad player. I don't understand it. There is nothing good about his game. He's athletic and huge, so he just "looks" right. He can't rebound. He can't bang. And he scores efficiently inside, but refuses to do that, shooting 15' jumpers all day. That guy got a max contract.
Click to expand...

Yeah. Howard and Lopez should not even be in the same sentence. As a cynical Atlanta resident, I laughed my ass off at JJ getting a max contract. I guess max contracts don't mean what they use to.

To the bolded part, why do you have to get sexual? I don't even want to know how you know that about Lopez. :)
 
  • Like
Reactions: 1 person
#97
#97
PKT_VOL said:
Interesting. You should post the new predictions for the upcoming season so we can track it as the season progresses.

I will say, team statistics make more sense than individual player wins/loses.
Click to expand...

That would take a lot of work. If Wages of Wins does it, I'll post their predictions. Injuries and trades make things difficult to predict.

The model works best in hindsight. It tells you who is responsible for the wins. It can predict accurately for the future if it knows the parameters. For instance, when Melo got traded they assumed he would stay healthy, play about 35 MPG, and that the Knicks would improve their record by about 1-2 wins. All of that turned out to be accurate, but if Melo had been hurt the prediction would've been bunk.
 
#98
#98
n_huffhines said:
That would take a lot of work. If Wages of Wins does it, I'll post their predictions. Injuries and trades make things difficult to predict.

The model works best in hindsight. It tells you who is responsible for the wins. It can predict accurately for the future if it knows the parameters. For instance, when Melo got traded they assumed he would stay healthy, play about 35 MPG, and that the Knicks would improve their record by about 1-2 wins. All of that turned out to be accurate, but if Melo had been hurt the prediction would've been bunk.
Click to expand...

I didn't mean for you to do it by hand lol. If Wages of Wins or some other place does it, you should post it.

Injuries and trades go without saying. I don't think any rational person would hold that against the predictions. Although, if they do it by individual statistics or projected statistics, I don't see why they couldn't make a quick revision of their predictions.
 
#99
#99
PKT_VOL said:
I didn't mean for you to do it by hand lol. If Wages of Wins or some other place does it, you should post it.

Injuries and trades go without saying. I don't think any rational person would hold that against the predictions. Although, if they do it by individual statistics or projected statistics, I don't see why they couldn't make a quick revision of their predictions.
Click to expand...

It's hard because you don't know how the minutes will be allocated. If Melo gets hurt, who is picking up his minutes? We know like clockwork Melo will play 35, but with him out, 10 minutes will go to player X, 8 minutes to player Y, etc. and all of that is subject to change on a nightly basis.
 
#100
#100
Speak of the devil. They don't know what to predict for rookies:

Early-Projection-table-1.png
 
New posts
You must log in or register to reply here.

VN Store



Back
Top