mnWI, if you want to debate stats, have at it!!

[Wolves2011]

Considering you have replied to about 10% of the stuff I have posted, and most of that 10% with somebody else's words, I find it funny that you would ask me if I read part of a post.

For the record, I have read it. I read it long before you posted it, because I wanted to educate myself on the topic in order properly defend my position. It would be nice if I could get the same in return from you, but that obviously isn't going to be the case.

For example, the formula at the bottom of the post simply takes 0.3 of the rebound credit away from his original formula. It does very little about the real issue of why rebounds are overvalued vs efficiency, which is taking away the vast majority of the scoring credit by subtracting on all FGA"s, not just misses. Note that he does not compare this new formula to the NBA Eff formula. He compares it to his formula.

Not only did I point you in this direction in my original post, it also should have been readily apparent from reading the post.

Did you bother to read the post?

let me say again, Win Score and NBA efficiency rating TREAT REBOUNDS IDENTICALLY. THERE IS NO DIFFERENCE, as in ZERO DIFFERENCE. JUST EXAMINE THE FORMULAS. THEY TREAT REBOUNDS IDENTICALLY.

PER and NBA Efficiency rating are virtually identical

[Wolves2011]

Here is a post from the thread I referenced a moment ago that clears up a little misconception about whether or not Berri actually likes scoring efficiency -

http://sonicscentral.com/apbrmetrics/vi ... &start=150

Dave Berri has said the main difference between Win Score and metrics like PER or NBA Efficiency is that WS places more value on shooting efficiency. His critics seem to concede this point, while stressing that WP goes too far in valuing efficiency over shot creation. What’s been missed, I think, is that WS actually rewards shooting efficiency less than the other metrics, not more.

To see this, we need to separate the efficiency of shooting from the volume of shooting. If you hold attempts constant, PER punishes inefficient shooting more than Win Score . For example, shooting 7-for-10 on 2pt attempts gets you a 4.0 WS and 9.39 PER, while shooting 4-for-10 yields –2.0 WS and 2.28 PER. So PER credits the efficient scoring as 7.11 points more valuable than the inefficient scoring, vs. a 6 point spread for WS.

You can see this clearly when comparing players with similar shooting frequency. For example, if we look only at the portion of the metrics measuring 2PA/2PM, the gap between Ming (.52 2P%) and Iverson (.47) is larger in PER than in WS (2006-07). If two players take about the same number of shots, PER will reward efficiency (or punish inefficiency) more than WS. This makes sense when you think about it, since PER deducts points for missed shots while WS deducts points for all shots.

Dan Rosenbaum also reports that the overall correlation between PER and TS% is .57, while the correlation between WP and TS% is just .53.

The result is that while high-volume/low-efficiency shooters are indeed rated very low under WS, high-volume/high-efficiency shooters are also rated lower by WS. For example, someone like Boozer (.56 FG%, 18.1 FGA40) gets far more credit for his shooting from PER than from WS. So what makes Win Score/Wins Produced different is not its love of efficient shooting -- it rewards this less than the other metrics -- but the cold shoulder it gives to shot taking, regardless of efficiency.

Please respond...

I will respond to this, but I want to think about it first.

[casey]

Please don't totally talk down to me, I'm trying to be respectful. I have 9 years of higher education from some of the best schools in the country. I might not be as smart as you, but I'm not a dummy either....

As I said in my posting I was going to wait for a couple of more weeks for the data for adjusted plus minus to settle down. Someone here "pressed" me to answer now, so I presented what data I had.

I'm willing to wait till the end of November and see how adjusted plus minus looks then.

Don't forget there is only 2 seasons of data available on the web site you cited it. This would be the 3rd season. It will be interesting to see how much consistency is there and if, things that "look funny" are explainable logically.

I don't mean to talk down to you, but you should understand why the numbers vary so widely right now. Look at the standard error for this season so far, and look at it for previous seasons. Look at how high and low some players' +/- is so far compared to the norm from previous years. You should understand why the numbers look funny right now, and you should expect them to look that way this early into the season.

fyi, there are actually 5 seasons of data available on the downloads page. You should be able to state your opinion of the stat based on the previous seasons' data. And I'm not sure why you complain about the data being available, when there is far less data available for your wins produced stat.

And I want to hear your thoughts about how the data is collected and then put together. The logic behind the statistic. What flaws, if any, do you see with it? That's the most important question, and you haven't stated any flaw with it yet.

[Wolves2011]

Please don't totally talk down to me, I'm trying to be respectful. I have 9 years of higher education from some of the best schools in the country. I might not be as smart as you, but I'm not a dummy either....

As I said in my posting I was going to wait for a couple of more weeks for the data for adjusted plus minus to settle down. Someone here "pressed" me to answer now, so I presented what data I had.

I'm willing to wait till the end of November and see how adjusted plus minus looks then.

Don't forget there is only 2 seasons of data available on the web site you cited it. This would be the 3rd season. It will be interesting to see how much consistency is there and if, things that "look funny" are explainable logically.

I don't mean to talk down to you, but you should understand why the numbers vary so widely right now. Look at the standard error for this season so far, and look at it for previous seasons. Look at how high and low some players' +/- is so far compared to the norm from previous years. You should understand why the numbers look funny right now, and you should expect them to look that way this early into the season.

fyi, there are actually 5 seasons of data available on the downloads page. You should be able to state your opinion of the stat based on the previous seasons' data. And I'm not sure why you complain about the data being available, when there is far less data available for your wins produced stat.

And I want to hear your thoughts about how the data is collected and then put together. The logic behind the statistic. What flaws, if any, do you see with it? That's the most important question, and you haven't stated any flaw with it yet.

If you go back and read my posts on adjusted plus minus, you will see I initially conceded the point that it was early in the season and the data hadn't settle yet. You then proceeded to call me names because I brought up the data and didn't recognize the the data hadn't settle yet. I then admitted again that it was early. I readily admit and have admitted right along thats its early to look at 2009-10 data.

Wins produced goes back to 2005-6 on the WOW web site. Adjusted plus minus goes back to 2007-8 on the web site. I'm not downloading data and the like. I'm interested in these topics, but its not my life.

As for critiquing the adjusted plus minus data and formula, I haven't seen them yet, so I can't. I don't know how the data is collected or what is analyzed.

I know in general plus/minus data is adjusted for quality of teammates and opponents. But not the methodology for doing so or what data is used.

[casey]

You then proceeded to call me names

When was this?

As for critiquing the adjusted plus minus data and formula, I haven't seen them yet, so I can't. I don't know how the data is collected or what is analyzed.

I know in general plus/minus data is adjusted for quality of teammates and opponents. But not the methodology for doing so or what data is used.

http://www.82games.com/ilardi1.htm

I'm pretty sure that's already posted, but apparently you didn't read it.

[panth181]

You also totally skipped by, the most straightforward portion of berri's argument; compare the two formulas. "NBA efficiency" is almost identical to PER.

I totally skipped by the first part? The part that I DIRECTLY responded to, to which your only response was that I ought to read the second part? The part that I backed up by leaving a SECOND EXPLANATION via link that plays it out with a statistical example?

If your 9 years of higher education didn't provide you with the capacity to think through how a weighting works and how taking away the biggest portion makes the 2nd biggest portion worth significantly more in percentage terms, you wasted your freaking money!

Of course, the truth is that now just playing dumb because you are too stubborn to admit you are dead wrong on this.

[panth181]

Considering you have replied to about 10% of the stuff I have posted, and most of that 10% with somebody else's words, I find it funny that you would ask me if I read part of a post.

For the record, I have read it. I read it long before you posted it, because I wanted to educate myself on the topic in order properly defend my position. It would be nice if I could get the same in return from you, but that obviously isn't going to be the case.

For example, the formula at the bottom of the post simply takes 0.3 of the rebound credit away from his original formula. It does very little about the real issue of why rebounds are overvalued vs efficiency, which is taking away the vast majority of the scoring credit by subtracting on all FGA"s, not just misses. Note that he does not compare this new formula to the NBA Eff formula. He compares it to his formula.

Not only did I point you in this direction in my original post, it also should have been readily apparent from reading the post.

Did you bother to read the post?

let me say again, Win Score and NBA efficiency rating TREAT REBOUNDS IDENTICALLY. THERE IS NO DIFFERENCE, as in ZERO DIFFERENCE. JUST EXAMINE THE FORMULAS. THEY TREAT REBOUNDS IDENTICALLY.

PER and NBA Efficiency rating are virtually identical

You need to examine the formulas. Try putting one together on a game line, and tell me that the % of rebounds in the total score is the same.

You say you want to debate, and you make me lay it out for you, bread crumb by bread crumb. Really pathetic on your part.

[Wolves2011]

Let me emphasize again, that NBA Efficiency Rating and PER give the identical value to rebounds as Wins Score.

The difference between the systems is relative. "NBA efficiency rating" (and PER) gives greater relative weight [vs. win score] to all forms of scoring, both efficient and inefficient. If you score you benefit.

Wins produced gives less relative weight [vs NBA ER or PER] to shooting, but greater relative weight [within the wins score system] to shooting efficiently (most important of all parts of formula) vs shooting inefficiently (keep in mind its penalizing all the inefficient shooters).

Lets compare NBA Efficiency Rating for shooters to Win Score:

WIN SCORE= Points - ALL FGA

shoot 6 for 10 -- win score = 2 points

shoot 4 for 10 -- win score = minus 2 points.

NBA EFFICIENCY RATING = Points - missed FG

shoot 6 for 10 -- NBA ER = 8 points

shoot 4 for 10 -- NBA ER = 2 points

As you can see in the simple example, NBA efficiency rating (and PER) gives great weight to all forms of shooting, both good and bad. You greatly benefit if you are a good shooter. You still benefit if you are a bad shooter.

For Win Score, you benefit much less from being a good shooter. You are penalized for being a bad shooter.

Lets define those two models again.

Wins Score = Points + Rebounds + Steals + ½*Assists + ½*Blocked Shots - Field Goal Attempts – ½*Free Throw Attempts – Turnovers – ½*Personal Fouls

NBA Efficiency = Points + Rebounds + Steals + Assists + Blocked Shots – All Missed Shots – Turnovers

So what are the other differences besides FG shooting? Win Score (minus) NBA ER =

So besides the shooting differences, NBA Efficiency Rating credits full assists rather than half, credits full blocked shots rather than half, and has no penalty for missed free throws, and no penalty for committing fouls. Other than that, the systems are identical.

To take on example, blocked shots don't always become a possession for the team that blocks the shot, but they are treated as such by NBA ER.

When Berri analyzed NBA Efficiency Rating and PER he found they were almost totally driven by scoring. You can see why with low penalty for missed shots.

Getting possession -- steals, rebounds and turnovers, is valued the same way in all three systems.

[panth181]

I'm pretty sure that's already posted, but apparently you didn't read it.

Don't get your hopes up on that front.

[Wolves2011]

Considering you have replied to about 10% of the stuff I have posted, and most of that 10% with somebody else's words, I find it funny that you would ask me if I read part of a post.

For the record, I have read it. I read it long before you posted it, because I wanted to educate myself on the topic in order properly defend my position. It would be nice if I could get the same in return from you, but that obviously isn't going to be the case.

For example, the formula at the bottom of the post simply takes 0.3 of the rebound credit away from his original formula. It does very little about the real issue of why rebounds are overvalued vs efficiency, which is taking away the vast majority of the scoring credit by subtracting on all FGA"s, not just misses. Note that he does not compare this new formula to the NBA Eff formula. He compares it to his formula.

Not only did I point you in this direction in my original post, it also should have been readily apparent from reading the post.

Did you bother to read the post?

let me say again, Win Score and NBA efficiency rating TREAT REBOUNDS IDENTICALLY. THERE IS NO DIFFERENCE, as in ZERO DIFFERENCE. JUST EXAMINE THE FORMULAS. THEY TREAT REBOUNDS IDENTICALLY.

PER and NBA Efficiency rating are virtually identical

You need to examine the formulas. Try putting one together on a game line, and tell me that the % of rebounds in the total score is the same.

You say you want to debate, and you make me lay it out for you, bread crumb by bread crumb. Really pathetic on your part.

The actual absolute weights for possessions are treated identically. The relative weights are treated differently because of the lower penalty for missed shots and because of NBA Efficiency Rating credits full assists rather than half, credits full blocked shots rather than half, and has no penalty for missed free throws, and no penalty for committing fouls. Other than that, the systems are identical.

If you are saying that the relative weights differ because of these differences, thats clear. But you imply that the relative weights of the NBA ER - PER formula is better with no evidence to support your view.

Why is the lower penalty for missed shots better. Why is it better not to penalize fouls, missed free throws, a full point to block shots when half the time the offensive team gets back the possession etc.....

An assertion isn't evidence.

[Wolves2011]

You then proceeded to call me names

When was this?

As for critiquing the adjusted plus minus data and formula, I haven't seen them yet, so I can't. I don't know how the data is collected or what is analyzed.

I know in general plus/minus data is adjusted for quality of teammates and opponents. But not the methodology for doing so or what data is used.

http://www.82games.com/ilardi1.htm

I'm pretty sure that's already posted, but apparently you didn't read it.

From the article... its just descriptive........

Again, the gory mathematical details of the adjusted plus-minus model have been described elsewhere (and they are beyond the scope of this article) –

[Wolves2011]

From the end of the article you sent me to, about adjusted plus minus.....

[NOTE: Isn't this what Berri and the win share guys are already doing? They predict wins produced by player based on their weightings of player stats, and compare that to actual team wins.]

Prediction. The ultimate validation of any scientific model is derived from its ability to make useful predictions of future events. I have claimed herein that the adjusted plus-minus model provides a valid estimate of each player’s ultimate effectiveness. But, of course, there are myriad other "comprehensive statistics" out there, about which similar claims have been made (John Hollinger’s PER rating and David Berri’s wages of wins metric come quickly to mind). How can I actually prove that the adjusted plus-minus rating is superior? The best way, of course, is simply to pit it in a head-to-head contest with other rating systems in predicting team outcomes in future seasons. Presumably, one could come up with an aggregate (weighted-average) rating for each team for any given season, based upon each player’s expected minutes played and his most recent set of ratings (perhaps tweaked a bit to reflect any anticipated improvement or decline as a function of player age, cumulative games played in career [i.e., mileage], recent injuries, etc.). I have not made the requisite calculations yet for the upcoming season, but I’m confident that such an analysis should prove the adjusted plus-minus statistic to yield better prediction than that afforded by rival measures. Of course, a sufficient sample size for testing this will probably require at least a few upcoming seasons’ worth of data, but it’s an analysis just crying out to be done!




http://www.82games.com/ilardi1.htm see last paragraph...

[panth181]

Let me emphasize again, that NBA Efficiency Rating and PER give the identical value to rebounds as Wins Score.

Only if you believe 1 in 100 is the same rate as 1 in a million. Since you are basically admitting that you do, it is much easier to see how you could worship the WOW methodology.

The difference between the systems is relative. "NBA efficiency rating" (and PER) gives greater relative weight [vs. win score] to all forms of scoring, both efficient and inefficient. If you score you benefit.

Wins produced gives less relative weight [vs NBA ER or PER] to shooting, but greater relative weight [within the wins score system] to shooting efficiently (most important of all parts of formula) vs shooting inefficiently (keep in mind its penalizing all the inefficient shooters).

You said you were going to think about the link I sent you that refutes this. I guess you decided to hold off on the thinking for a while?

Lets compare NBA Efficiency Rating for shooters to Win Score:

WIN SCORE= Points - ALL FGA

shoot 6 for 10 -- win score = 2 points

shoot 4 for 10 -- win score = minus 2 points.

NBA EFFICIENCY RATING = Points - missed FG

shoot 6 for 10 -- NBA ER = 8 points

shoot 4 for 10 -- NBA ER = 2 points

As you can see in the simple example, NBA efficiency rating (and PER) gives great weight to all forms of shooting, both good and bad. You greatly benefit if you are a good shooter. You still benefit if you are a bad shooter.

Yep, and Win Score does the same thing with rebounds. It starts from zero, rather than starting from the league average. League average scoring gives you nothing (hence the higher weight in rebounds), league average rebound gives you a major positive. This is illogical.

For Win Score, you benefit much less from being a good shooter. You are penalized for being a bad shooter.

Yep. Instead, you benefit from being a good rebounder. Hence the ridiculousness of Andrew Bynum being better than Kobe Bryant and Dennis Rodman being better than Michael Jordan.

Lets define those two models again.

Wins Score = Points + Rebounds + Steals + ½*Assists + ½*Blocked Shots - Field Goal Attempts – ½*Free Throw Attempts – Turnovers – ½*Personal Fouls

NBA Efficiency = Points + Rebounds + Steals + Assists + Blocked Shots – All Missed Shots – Turnovers

So what are the other differences besides FG shooting? Win Score (minus) NBA ER =

So besides the shooting differences, NBA Efficiency Rating credits full assists rather than half, credits full blocked shots rather than half, and has no penalty for missed free throws, and no penalty for committing fouls. Other than that, the systems are identical.

Otherthan that? That's kind of like saying Charlie Manson was a great guy, other than all those people he killed!

Getting possession -- steals, rebounds and turnovers, is valued the same way in all three systems.

Yes, getting a defensive rebound that very likely has nothing to do with anything you specifically did other than be under the basket, has the same value as making a 2 point shot.

And that makes sense to you, right?

Right?

[panth181]

The actual absolute weights for possessions are treated identically. The relative weights are treated differently because of the lower penalty for missed shots and because of NBA Efficiency Rating credits full assists rather than half, credits full blocked shots rather than half, and has no penalty for missed free throws, and no penalty for committing fouls. Other than that, the systems are identical.

If you are saying that the relative weights differ because of these differences, thats clear. But you imply that the relative weights of the NBA ER - PER formula is better with no evidence to support your view.

Why in the world would I be talking about absolute weights? Do we talk about the total number of points somebody scored and call them the best scorer? Heck no we don't! Quit playing dumb!

And I really couldn't care less about PER or NBA ER right now. I'm saying that Wins Produced is a ridiculously flawed system. It doesn't work logically or pass the smell test. It is useless. PER isn't perfect, but at least it accomplishes what it set out to do.

Why is the lower penalty for missed shots better. Why is it better not to penalize fouls, missed free throws, a full point to block shots when half the time the offensive team gets back the possession etc.....

Those things really don't matter. There aren't enough of them recorded to dramatically influence the outcome of the rating. If you strip away scoring to nothing, which is basically what win score does, you end up with a rebounding-dominated metric. Not a scoring efficiency dominated metric like Berri would like you to think, but a rebounding-dominated metric.

And would I prefer to have a metric that weights the point of the game, which is to score, or one that weights rebounds, which in large part is a function of where you are in relation to the basket when a shot goes up? The answer is pretty obvious on that front...

(And please don't tell me I haven't provided any evidence on the "not a scoring efficiency dominated metric" when I left you a link above that you are supposedly thinking about. Don't bring it up again until you are done thinking.)

[casey]

From the article... its just descriptive........

Again, the gory mathematical details of the adjusted plus-minus model have been described elsewhere (and they are beyond the scope of this article) –

So...you have no thoughts on the stat or what?

From the end of the article you sent me to, about adjusted plus minus.....

[NOTE: Isn't this what Berri and the win share guys are already doing? They predict wins produced by player based on their weightings of player stats, and compare that to actual team wins.]

Prediction. The ultimate validation of any scientific model is derived from its ability to make useful predictions of future events.

I'm glad that you posted that, because I was going to point out that exact same part. What Berri does is "predict" the PAST. It sounds good when he says that the stat accurately predicts team wins. But that's not really what it's doing. And it doesn't matter how the wins produced are allocated between individual players, because he just adds them up in the end to prove his "predictions".

Shoot, I can come up with a stat that is even more accurate than his in "predicting" wins. The formula is individual minutes / total team minutes * team wins. Here are my predictions for the 08-09 season:

Cleveland Cavaliers 66
Los Angeles Lakers 65
Boston Celtics 62
Orlando Magic 59
San Antonio Spurs 54
Denver Nuggets 54
Portland Trail Blazers 54
Houston Rockets 53
Dallas Mavericks 50
New Orleans Hornets 49
Utah Jazz 48
Atlanta Hawks 47
Phoenix Suns 46
Miami Heat 43
Chicago Bulls 41
Philadelphia 76ers 41
Detroit Pistons 39
Indiana Pacers 36
Charlotte Bobcats 35
Milwaukee Bucks 34
New Jersey Nets 34
Toronto Raptors 33
New York Knickerbockers 32
Golden State Warriors 29
Memphis Grizzlies 24
Minnesota Timberwolves 24
Oklahoma City Thunder 23
Los Angeles Clippers 19
Washington Wizards 19
Sacramento Kings 17

A quick glance at the team totals shows that I am 100% accurate with all of my predictions. So obviously my stat is far superior to wins produced. Go me!