Reviewing Every Player on Every Team was a post offered last month. This column reported every player’s WP48 [Wins Produced per 48 minutes] and Wins Produced somewhere around the quarter-pole of the season.
On Tuesday night, the Atlanta Hawks finally played their 41st game. So now every team has reached the midpoint of the season.
Over the next few days I will offer some thought on the first half of the 2007-08 campaign. For now, though, here is every player in the NBA after 41 games.
Table One: All NBA Players after 41 games of the 2007-08 season
A few notes to consider in looking at this table:
1. Players are ranked in terms of Wins Produced. So this might make finding your favorite player a bit difficult. Soon I will post the results organized by team.
2. Projected Wins Produced is simply Wins Produced *2. This assumes that what you did in the first half – including minutes played – stays the same in the second half. Obviously for a player like Andrew Bynum – who is injured – this is not true. So keep that in mind in looking at the projections.
3. If a player played for multiple teams he is listed twice.
4. The results don’t appear much different from what I posted in December. Seven of the top ten players in December are still in the Top Ten at the mid-point of the season.
5. Here is what the All-Star game rosters would look like if fans and coaches only considered Wins Produced. I will comment more on the All-Star game in a future post. For now, though, I just wanted to note that WordPress does allow me to put tables directly into the posts. These are not pretty, but I think they can be read.
Rank |
Western Conference Stars |
WP48 |
Wins Produced |
Starters | |||
4 |
Chris Paul |
0.377 |
11.4 |
10 |
Steve Nash |
0.321 |
9.1 |
7 |
Shawn Marion |
0.314 |
9.6 |
16 |
Carlos Boozer |
0.281 |
8.1 |
2 |
Marcus Camby |
0.437 |
12.8 |
Reserves | |||
17 |
Baron Davis |
0.237 |
7.9 |
18 |
Kobe Bryant |
0.240 |
7.6 |
27 |
Dirk Nowitzki |
0.204 |
6.4 |
31 |
Josh Howard |
0.202 |
5.8 |
8 |
Tyson Chandler |
0.323 |
9.4 |
13 |
Tim Duncan |
0.322 |
8.4 |
15 |
Chris Kaman |
0.256 |
8.2 |
Rank |
Eastern Conference Stars |
WP48 |
Wins Produced |
Starters | |||
5 |
Jason Kidd |
0.348 |
10.8 |
9 |
Chauncey Billups |
0.335 |
9.2 |
3 |
Kevin Garnett |
0.398 |
11.9 |
6 |
LeBron James |
0.331 |
9.9 |
1 |
Dwight Howard |
0.440 |
14.6 |
Reserves | |||
12 |
Jose Calderon |
0.319 |
8.5 |
28 |
Mike Dunleavy |
0.215 |
6.3 |
11 |
Caron Butler |
0.253 |
8.5 |
25 |
Paul Pierce |
0.205 |
6.7 |
30 |
Samuel Dalembert |
0.207 |
5.9 |
29 |
Antawn Jamison |
0.175 |
5.9 |
32 |
Emeka Okafor |
0.203 |
5.7 |
In closing this post I wanted to repeat something I said back in December. Specifically, I wanted to repost the following observation I offered on baseball stats – an observation that I think applies to those looking at statistical analysis of the NBA.
A Comment on Statheads in Baseball
My first love growing up in Detroit was baseball. A good part of my youth was spent collecting baseball cards and I spent many hours looking at the numbers on the back of each card. Such numbers are – by Sabermetric standards – quite simple. Hits, runs, RBIs, and of course batting average are the primary stats you tended to see back in the 1970s. At that time there was no mention of OPS or any other “advanced” metric.
And of course we didn’t need such stuff. Baseball fans knew who the best players were. Although we looked at the numbers, all we had to do was watch the players and we could tell who was “good” and who was “bad.”
Now, thanks to Bill James and others of his ilk, we have all these new numbers. And of course people look at these numbers as if they “prove” something. But anyone who knows baseball knows that these numbers don’t prove anything.
For example, consider a number like Runs Created. Runs Created supposedly considers everything a player does offensively and tells us how many runs a player “creates.” And since creating runs is the purpose of offense in baseball, Runs Created should tell us who is “better” or “worse.”
But all you have to do is look at the numbers and you can see that these Sabermetric numbers don’t tell us anything. Consider the rankings posted by ESPN of each hitter in terms of Runs Created per 27 outs. Fourth on the list is Carlos Pena. As a Tigers fan I am quite familiar with Mr. Pena. Pena played more than three seasons in Detroit and never saw his batting average go above 0.250. He was so talented he couldn’t even make the Tigers roster in 2006 and consequently spent most of that season in the minors.
Meanwhile, Albert Pujols spent 2005 “proving” that he was the Most Valuable Player in the National League. In 2006 Pujols finished second in voting for the MVP award. And then this past season Pujols hit 0.327 while Pena only hit 0.282. For those non-math majors out there, that’s a 45 point difference.
But the Runs Created stat ranks Pujols as only the 13th best player in baseball. Yes, Pujols – the 2005 MVP – is ranked nine spots below a player who spent 2006 in the minors with a batting average in 2007 that was 45 points lower.
When you see stuff like that you have to say, “these Sabermetric stat-heads need to get their head out of their computers and go watch a game. Pena better than Pujols? Yea, I think the Devil Rays would make that trade in a second.”
A Note for the Non-Satirical
Of course, no Sabermetrician would look at Runs Created in 2007 and declare that Pena is “better” than Pujols for all time. One would look at these numbers, though, and say that Pena was pretty good this past season.
Well, we would say that if we looked past batting average. In terms of this archaic 19th century stat, Pena was only the 83rd best hitter in baseball last year. There were only 162 hitters who qualified for the rankings last year, so batting average lists Pena in the bottom half of all hitters. The more advanced stats, though, place him in the top five. Given the flaws in batting average, we tend to believe the more advanced stats.
But even though we believe the advanced stats, we don’t look at results for one year and declare that this trumps the entire history we have on two players. Again, numbers help us think. Numbers do not do our thinking for us.
– DJ
Our research on the NBA was summarized HERE.
The Technical Notes at wagesofwins.com provides substantially more information on the published research behind Wins Produced and Win Score
Wins Produced, Win Score, and PAWSmin are also discussed in the following posts:
Simple Models of Player Performance
What Wins Produced Says and What It Does Not Say
Introducing PAWSmin — and a Defense of Box Score Statistics
Finally, A Guide to Evaluating Models contains useful hints on how to interpret and evaluate statistical models.
Paulo
January 31, 2008
Three of the bottom 10 play for the Nets (Krstic is just outside the bottom 10). No wonder he wants to be traded. It probably used to be somewhat tolerable when it was just Jason Collins.
Paulo
January 31, 2008
*PS. By “he” in my previous post, I meant Jason Kidd.
Tom
January 31, 2008
I just noticed that for Emeka Okafor, his projected wins produced is much more than twice his wins produced. What’s up?
William
January 31, 2008
Because Okafor is about to go on a RAMPAGE.
TheMC5
January 31, 2008
Hahahaha! Dunleavy in an All-Star game. Too funny.
dberri
January 31, 2008
Tom,
All the Bobcats are off just a bit. I will try and get this fixed later today.
dberri
January 31, 2008
Okay, the Bobcats projected Wins Produced have been fixed. Thanks for letting me know.
TG
January 31, 2008
“Again, numbers help us think. Numbers do not do our thinking for us.”
Howard IS great, but Camby and Kidd are overvalued in your system (rebounds).
JPF yesterday brought up an interesting point or two, which have been mentioned before.
1) Roles: Team constructs and strategy often determine individual productivity that doesn’t necessarily correlate 100% to individual ability (or potential individual productivity). Famous example: Bob Cousy surely had fewer defensive rebounds because he was always heading up the floor for a fast break outlet pass from Bill Russell. Most teams on the offensive end designate one guard to hang back to prevent opponent fast breaks. Surely this designated guard gets punished with fewer offensive rebounds… and further punished with increased fouls on opponent fast breaks.
It is interesting that Deron Williams does get such a low number of rebounds, considering he has almost a Barkley-type body for a guard. Allen Iverson also has always had a low rebound total for a guard… but that could just be ‘body preservation’ because he takes repeated beatings in the lane when he’s trying to score. Again, this might come down to strategy in a way: “Allen, you can only take so much abuse… so take it when you have the ball and trying to score… and let the big guys try to get the team misses.”
2) Playoffs: Deron Williams was electric in the playoffs last year… so he certainly showed what he COULD do… when the games were more important and the 82-game team strategies weren’t going to get the job done. Again, this is a case a ‘Roles’… and Deron’s productivity now correlated better with his true ability.
It would be interesting to see a comparison of ‘Deron 2006/07 82-games’ vs. ‘Deron 2007 playoffs’.
3) I think the problem with many player debates is that one side is arguing that one player is more ‘productive’… while the other side is arguing that the other person is ‘better’. Meaning… one side is arguing about productivity… and the other side is arguing about ability. They aren’t always the same due to the reasoning above.
4) I think that what NBA coaches are trying to do is maximize the synergy of the team, thus maximize TEAM productivity. This may have the effect of certain individuals individual productivity being less than ‘true’ individual productivity (ability)… and this will naturally occur when you have five players but only one ball.
PS: Your bad trade analysis idea would be interesting. Wilt vs. Russell would be best.
Jason
January 31, 2008
Wilt vs. Russell would be a horrible trade. While Russell is far, far, far past his prime at this point, Wilt hasn’t been alive for a few year.
dustin
January 31, 2008
T.G., DB has said before that wins produced doesn’t say WHY a player is productive. It may be the case that ai/dw are slightly less productive because of their team’s strategy in employing them.
Also, I’d be interested in knowing if the team adjustment would help offset a strategy that was better for the team but perhaps led to an individual being less productive.
Westy
January 31, 2008
Thanks for posting this, DB! Interesting to look through the rankings.
Per usual, I’m sure certain players’ rankings will spur discussion.
I’m wondering, did you have the chance to read this recent post on Basketball Prospectus about the Rookie of the Year race? It actually mostly focuses on the balance of weighing statistically usage and efficiency. I thought he had some pretty good thoughts. I’d be curious as to yours and this forum’s reaction to some of it. Here are some tidbits:
Usage rate measures the percentage of a team’s possessions a player “uses up” while he is on the floor. The skill being measured is a player’s ability to create his own offense for his team. It’s one of the most underrated metrics in basketball…
…Usage rate is an essential tool for rating perimeter players. Any successful team must have players that can create offense. A team of the five most efficient players in the game (call them the Fabricio Obertos) would have a tough time scoring because they rely on these high-usage types to create opportunities for them.
This explains statistically the difference between a guy like Durant and someone like J.J. Redick. We can watch those two play and the difference is obvious. Redick is undersized and slow and even though the Better Basketball folks consider him perhaps the greatest shooter of all time, it doesn’t really matter because he can’t create his own offense. Meanwhile, the long, athletic Durant can pretty much pull up and take a jump shot any time he wants, which is something he chooses to do a little too often at this early juncture of his career.
dberri
January 31, 2008
Westy,
Quick thought on the Redick-Durant comparison:
JJ Redick’s WP48 = -0.043
Kevin Durant’s WP48 = -0.037
As bad as Durant is, he is still better than Redick.
So people are right, you have to be able to get your shot off. But if you can’t, that does show up in the numbers.
Big Al
January 31, 2008
I was wondering if you would consider breaking down Big Al Jefferson’s game. He’s got a high winshares, and nice raw stats…yet, there seems to be something missing in his game. Mainly, defense.
John Woo
January 31, 2008
Andrea Bargnani was the worst player in the entire league for the first half? o_O
Did Bryan Colangelo blow that pick up or what?
joe
January 31, 2008
If the wins-produced stats were available every day, then I’d refer to them when deciding who to vote into the ASG (but I’d still be leaving Dunleavy and Dalembert off my ballot!)
Daniel
January 31, 2008
WP48 is better to evaluate All-Stars. Is Josh Howard with a .202 WP48 a better choice than Manu at .310 if minutes are limited? It seems to me that if you only have 48 minutes to allot in a game, you want the players who will be most productive each minute, rather than those who have simply played more minutes for their respective teams. Can Howard (or really anyone else on either team, for that matter) be more productive in 20 minutes off the bench than Manu?
You worked very hard to make an incredible tool for statistical analysis, and when it comes time to implement the tool, you muck it up completely.
For shame, Mr. Berri… For shame.
TG
January 31, 2008
Jason,
You’re right, Russell couldn’t make a playground team right now. But Wilt’s first string in heaven (and JC plays point).
Animal
January 31, 2008
Dr. Berri,
Do you think this year will be the first year ever that the rookies beat the sophomores in the All-Star game? Wasn’t last year’s rookie class the worst ever? How does this year’s rookie class compare?
Animal
January 31, 2008
Also, I agree with your general point about stats but I’m not sure I think runs created is misrepresentative for Pena’s very good year last year. Carlos Pena hit 46 home runs, had 121 RBIs, and an OBP of .411. Ignoring defense I would take his season last year over that of Pujols (32 home runs, 103 RBIs, and .429 OBP). I’ll bet on Pujols, though. Regression to the mean. Pujols is much, much better for career. (And again defense excluded b/c I think Pujols ranks highest in 1B range stats)
Animal
January 31, 2008
Oops, I meant to write I’d bet on Pujols having higher runs created for 2008. Which I guess is what your point partly was.
Sam Cohen
February 1, 2008
I noticed that the top and bottom two spots belonged to centers. Out of curiosity, do the same positions tend to have greater variability in WP48 year-to-year, or does it change all the time?
If the same positions have greater variability every year, then I’d think in comparing the best players at each position I’d be more interested in comparing their WP48 to the average WP48 of starters at their position instead of to the overall average for the position.
Tim
February 1, 2008
It seems to me that if you only have 48 minutes to allot in a game, you want the players who will be most productive each minute
The Franchise
February 2, 2008
Sam-
Due to the “short supply of tall people,” there is a greater deviation in the productivity of big men.