My latest article for RealGM

[Wiz99]

-= original quote snipped =-



It's only shocking to those folks on this board who have b**tched and moaned all season long (and seasons before that) about how AJ is too "soft" and how the Zards will never win anything with him as a starter.

All the guy does is put up a double-double most nights. In fact, with the way he's carried the Zards this season we all should be kissing his butt.

Given the loss of GA and the missed games by CB, the Zards would have been lucky to win 25 games without AJ, imo.

Look. I like AJ personally. He seems like a nice fellow, upstanding character. And without a doubt he has played his ass off this year and been the rock we needed when Gilbert went down. And what's more, I am not surprised he ranks so highly among the league's elite.

Hats off to ya, AJ!

But I still maintain there is NO WAY we win a championship with Jamison as the most valuable player on our team (which is what these stats say he is). AJ could very well be his generation's Dominique Wilkins. A forward who does nothing but put up big, BIG numbers year after year after year... and never got his team within sniffing difference of the Finals.

'Nique was one of the most prolific scorers of all time (averaged 30 ppg 3x, 24 for his career). He had Moses Malone, Kevin Willis, Doc Rivers, Dan Roundfield as running mates (so don't tell me Nique never had anyone to work with). Mike Fratello as a coach [no slouch, no Phil Jackson either, but I'm just saying he wasn't hampered by coaching].

Some players just don't have that streak of greatness in them. To me, Jamison walks, talks and smells like 21st century Dominque.

Nice dude, big producer his whole career, just not great enough to get his team to the Finals.

[Wiz99]

BTW, great article, Kev.

Very readable. When it comes to discussing stats, achieving that is an art. You're a Picasso!

[MJG]

By the way... did anyone else read Hollinger's predictions this morning on ESPN and wonder how in the world he could muster...

(a) a comment like this: "Meanwhile, the Boston juggernaut will dismiss Washington in four." (((4...? is he serious? we wont even win ONE? When we spanked them thrice in a row? Whatever dude.)

Doesn't he also think Boston will beat the Pistons in five and the western champ (Utah, if I remember correctly) in six? I think he just really likes Boston, more so than he's making a dig at us.

(Not that I agree with him - I'd be rather surprised if we didn't beat the Celtics once).

[wermolwermol777]

I really love the article and the system you used to come up with the list.
Advanced stats are great.

I do have one small question about your method.

While I understand your reasons for using the individual PER stats, do you not agree that it somewhat duplicates the effect from the on/off ratings?
Perhaps using some sort of a coefficient for the PER stats would help?

Nevertheless, this is a great method for MVP rankings and after seeing this I will sleep a lot better than I would have, should Queen James get it.

[Duiz]

Offensive PER, but how about Opponents PER, and why don't you rather use Roland Ratings which are much more efficient.

[TheSecretWeapon]

I really love the article and the system you used to come up with the list.
Advanced stats are great.

I do have one small question about your method.

While I understand your reasons for using the individual PER stats, do you not agree that it somewhat duplicates the effect from the on/off ratings?
Perhaps using some sort of a coefficient for the PER stats would help?

Nevertheless, this is a great method for MVP rankings and after seeing this I will sleep a lot better than I would have, should Queen James get it.

PER doesn't really duplicate the on/off numbers. The on/off stuff is what the team does while that player is on the floor vs. what it does when he's off the floor. PER is a summary of the player's box score stats. I do deflate PER (which is a per minute stat) by minutes played. I've been thinking about rescaling the two scores (the on/off part and the PER part) so that they're on exactly the same scale. Right now, PER accounts for about 65% of a player's final MVP score on average. Rescaling could bring that down a bit.

[Wizards2Lottery]

-= original quote snipped =-



It's only shocking to those folks on this board who have b**tched and moaned all season long (and seasons before that) about how AJ is too "soft" and how the Zards will never win anything with him as a starter.

All the guy does is put up a double-double most nights. In fact, with the way he's carried the Zards this season we all should be kissing his butt.

hah funny how in our first post season game AJ played like a total softy on offense by going 1-8 from the three point line. Cap this up with him missing 3 wide open shots in the most crucial stretch of the game.

this is why no one will ever kiss his ass.

[Duiz]

TheSecretWeapon... Did you just ignore me? Do you know what the Roland Ratings are?

http://82games.com/ROLRTG8.HTM

Get yourself acquainted. So far to me the Roland Ratings are the far most efficient way to give a true pure statisical number of the player's impact on the game.

[BruceO]

that roland rating seems about right. Alot of those players with similar ratings get compared alot with each other with arguments about who's better.
what I'm surprised about is how low deron williams and carmelo are on the scale. Jamario moon is actually higher on the scale. Which makes me think theres a flaw somewhere. Things that may tinker with ON/off numbers might be who comes in to sub you. Also the difficulty level the opponent sets for you is a factor never considered by these numbers. LIke e.g a lebron james has good off/on numbers despite the opponents attempt to make it hard for him. It's almost like a college strength of schedule addition to calculations. Also I am comparing positional players. E.g I see Billups, Nash and Iverson side by side. And I see Paul higher up.
In the TSW's rankings I'd take Howard over Jamison, and garnetts team has a much better record than the wiz. So what are the numbers calculating? Who's most invested in by a team? who is integral in driving the team? or who has more impact.
Also I think theres some offcourt things that some players do that improve their team. The way the practice, their leadership that they show that gives their team confidence to compete.

The thing about on/off numbers that I question is this, I've seen scenarios where a star player sits for a while and his team doesnt suck enough to crumble while he's on the bench. Maybe his sub is decent enough not to let up. Or is better than the opposing bench player. Doesn't that affect your on/off numbers? I'm just wondering what these numbers actually tell you and the flaw that they possess.

I'm editing this to say I know TSW has nothing to do with the roland rating but both of you guys are using the principle of on/off numbers. I noticed also the roland rating has blatche at -6.1 and has someone like jannero pargo at -4.1. Somethings wrong here.
I think something that might be more interesting is creating some sort of strength of schedule based on who the player faces in their matchups and how they perform against those players. Include in that the teams performance as well as the players performance. You can have tiers of players to grade against e.g how do they play against the chris pauls, the williams, and other starters.. then how effective they are against second unit. Possibly rank according to position, because they have to do similar things and face similar people so its easier to grade.

[nate33]

There are reasons to think the Wiz have a chance to go deep into the playoffs. But "improved defense" ain't one of them.

While I agree totally with your statistical breakdown, my gut tells me that this Wizards team is playing better defense than in years past. I think part of the reason why our defensive stats for the entire season showed little improvement is because we suffered so many injuries and were forced to play a bunch of inexperienced 21-year-olds extensive minutes.

I note than when either Butler or Jamison are on the floor the team yielded 109.6 points per 100 possessions. EJ tries to keep either Butler or Jamison on the floor at all times. When they're both off, it was only because it was garbage time, or Butler was hurt (i.e. we were injury depleted).

Likewise, when DMac or Pecherov were off the floor, the team again yielded 109.6 points per 100 possessions. It was when those guys were on the floor that our defense suffered. With DMac on the floor (17% of the minutes), we yielded 114.4 points/100. When Pecherov was on the floor (8% of the minutes), we yielded 118.6 points/100.

In a normal season without all the injuries, DMac and OPec would never have played except in pure garbage time against total scrubs. The way I look at it, when reasonably healthy, this team allowed 109.6 points per 100 possessions. That's not great, but it's a solid improvement over last year's 112.5 points per 100 possessions.

(Note, I used 82games.com definition of a possession with the 109.6 figure. I used Knickerblogger's definition on the 112.5 figure. As I understand it, they use slightly different tabulation methods so these numbers aren't directly comparable. But they should still be close.)

[jmrosenth]

TheSecretWeapon... Did you just ignore me? Do you know what the Roland Ratings are?

http://82games.com/ROLRTG8.HTM

Get yourself acquainted. So far to me the Roland Ratings are the far most efficient way to give a true pure statisical number of the player's impact on the game.

Can you explain further? I'm not sure if TheSecretWeapon is familiar with them.

[TheSecretWeapon]

TheSecretWeapon... Did you just ignore me? Do you know what the Roland Ratings are?

http://82games.com/ROLRTG8.HTM

Get yourself acquainted. So far to me the Roland Ratings are the far most efficient way to give a true pure statisical number of the player's impact on the game.

Sorry, I overlooked your post.

Yes, I'm familiar with the Roland Ratings. I don't use them for a few reasons:

1. Roland doesn't fully explain how they're calculated.
2. They seem to be using much of the same data I'm using.
3. They include counterpart data, which I think is very flawed data.

[miller31time]

Kev, do you still have that article you wrote about defense not necessarily winning championships stored? It would be nice to post that on the General Board where people are convinced you need an elite defense to win the title, but I remember your article refuting that.