I love twitter. It’s a window into the mind of some very creative/Talented/interesting people. It’s a constant source of amusement and inspiration. Now yesterday I was looking around for my muse (damn thing had gotten lost again) and I came upon the following tweet:
DannyLeroux @chadfordinsider I would LOVE to see a “strong argument” that Kevin Durant is “the best player in the NBA” right now. Haven’t yet or ever
I proceeded to reply:
@DannyLeroux Durant, best young player, yes.Best value for next ten years, probably. Best player now? Lebron is still #1, but KD can claim 2
And this led to the following exchange:
DannyLeroux@ArturoGalletti No arguments whatsoever with that. Just saying he’s #1 right now is pure insanity.
@DannyLeroux In terms of Physical Gifts Lebron has it for the next Ten. But KD has a drive to improve,win that makes a better centerpiece
DannyLeroux@ArturoGalletti Actually, I think I’d take LBJ over KD for the next 10 years- LeBron is only 25, after all. But it’s a close call.
@DannyLeroux I actually think KD’s intangibles are off the charts. Could you see LBJ leading Team USA like that at any age (let alone 22)?
DannyLeroux@ArturoGallettiHaving met him and seen him in action, KD’s intangibles are great. But LBJ’s passing ability is something that KD can’t get
@DannyLeroux You get the feeling with Lebron vs. Durant, we’re arguing Picasso v. Michelangelo?
The wonderful thing about this exchange was that once it was over inspiration was sitting on my lap. I decided that I was going to rank all the players in the NBA based on their for last year which led to me wanting to do a preliminary ranking of all the NBA teams going into 2010-2011 which led to me thinking that I might as well do a full preliminary version of the 2010-2011 roster (or as it will come to be known “The Build” ) which led to my wife wondering why I was up so late writing endless run-on sentences. I immediately headed to the excel cave to try and produce something awesome.
The goddamn excel man
The result is the first version of my NBA now rankings. Since I love my readers, I’ve put the data in a google table which you can download and play with.This table will form the basis of the model that I will be building to project the coming NBA season.
I've said this many times.(Image by http://xkcd.com/)
But before you all go off and do something silly with it, I’m going to walk you through it.
NBA Now Player Rankings take 1
For the Player rankings, I looked at the following:
- Players who played more than 400 Minutes last year (sorry Yao)
- Wins Produced (WP)
- WP48
I then calculated the percentile value (or the percent of players that could be expected to be below the player’s Wins Produced and WP48 value) for WP and WP48 and multiplied these two numbers to give the player a cumulative score on what for now shall be known as the Lebron Scale (not-so subtle hint, I know). It then used this scale to rank all the players based on their cumulative score. The NBA Now Top 50 looks like this:
Some thoughts:
- Lebron is obviously number 1 in the Lebron Scale (I’m working on getting off the list can you tell?).
- Howard is a close number two at almost full Lebron bio-equivalency and coming very close to turning this into the Superman Index.
- Some old favorites round out the top 5 in Camby, Kidd and Duncan.
- Lebron’s teammate D-Wade rounds out the top six at .985 Lebron units.
- The real interesting thing about that top six is that every player n it has been the best player on a Finals winner or Runner-up.
- Given that I’m using the full year for this first version, Durant is only number 9 but he is the youngest in the top ten by three years (and his second half numbers suggest the 2011 could be the Plastic Man list).
- Chris Paul is number 17 but this was done in 1712 minutes. You know the drill:
- Kobe comes in at 40 at about 3/4 of a Lebron unit. Carmelo misses the top fifty entirely (number 95- .425 of a Lebron unit) and his partners in the trade rumor, Noah and Deng, come in at 22 (.909 units) and 56 (.638 units).
Now we can use this data to rank the NBA Rosters for the coming season.
NBA Now Team Rankings take 1
As I said earlier, I want to make some things clear about this build before any of you go off and do something silly with it. This is a preliminary team rankings based on 2010-2011 performance only so I’ve not included:
- Detailed Minute allocations
- The Rookie Model (yes Devin I’m working on it
) - Injuries (rest easy Rocket fans)
- Age modeling ( Lakers & Celtics fans beware)
I’m also not using older player data (2005-2006 through 2008-2009) which I will include in the final build. For each team I’ve worked out:
- The current roster (thank you Wikipedia and Excel)
- Wins by Players currently on each Roster for the 2009-2010 Season
- Minutes played by players currently on each Roster for the 2009-2010 Season
- Cumulative Roster Minutes in 20009-2010 as a % of Required Minutes (82*5*48)
- Adjusted Wins based on Minutes Played
- If Minutes are above 100%. Divide Wins by Cumulative Roster Minutes
- If Minutes are below 100% Fill out remaining minutes at .050 WP48 (avg for rookies)
- Adjusted Wins based on Minutes and Available Wins. Basically wins have to add up to 1230 or 30 times 41.
- Roster Quality or the sum of Player Scores from Last Year (Lebron Units)
- Projected Playoff Win Generation per Game for Top 6 based on MP last Year. This takes the win generation of the top 6 and projects it to the playoffs based on the half-baked theory.
- Finally I rank each team based on Adjusted Wins (past performance), Roster Quality (depth) and playoff win generation (or how good they’ll be in a seven game series), I figure out the average rank and come up with a preliminary Team Rank
The rankings look like this:
Now featuring Nene Hilario
Preliminarily, the Heat are expected to win the most games in the regular season and romp through the playoff 96 Bulls style. The Bulls and Celtics look to be their primary obstacles. Surprisingly the numbers rank the Blazers roster coming out of the West (It really likes all the Tall guys). Before Blazers fans get too excited remember that injuries and improvement projections based on age will very probably change this significantly.
On a final note, our friends the wizards come out as last overall and very close to our previous projections.
Note: Questions? Here a guide to commonly used terms on this blog from Devin at the NBeh? blog . Here’s a Guide to Calculating Wins Produced from Wages of Wins. Anything else? Feel free to use the comments and I will try to answer your questions.
Note 2: Was missing Nene (the one name thing threw my algorithm). The good news for Denver is that their in the playoffs. The Bad news is that they should be favorites to lose in the first round against pretty much anybody.
Daniel Leroux
09/16/2010
I feel honored to have been part of the root of all this.
Great work, and it’ll be exciting to see where it goes with minute allocation and whatnot.
Two questions/ideas:
1. How did you deal with a Miami team that I believe has more than 15 guys on roster right now? Did you just take their 15 best guys?
2. Would it be possible to run some sort of simulation that projects regular season success based on taking out minutes from a random (or non-random) group of the high-end rotation guys and move them down the list? Seems like even though it would be flawed, it could help remedy the problem that the lack of injuries make every prediction a little sunnier than it likely should be.
Regardless, great read.
arturogalletti
09/16/2010
Daniel,
No, thank you for triggering my inspiration.
As to the Questions:
1.For the current version, I’m using a hack and slash approach to minutes. I just took the wins and adjusted by minutes as noted. For Miami I took the 14 guys that played in the NBA last season (who netted 75.6 Wins in about 112% of the required minutes) divided 75.6 by 1.12 and added back about 2/3 of win to make the win totals for all teams add up to 1230.
2. This is very similar to the injury luck concept that I’ve seen used by football outsiders. I’m going to play with this. I’m going to use player history, age and try to work out some sort of injury constant based on games lost over the last ten years. It should be an interesting exercise.
I’m playing with exactly ho
Chicago Tim
09/16/2010
Regarding the Anthony-to-the-Bulls rumor, another nail in the coffin:
“The Bulls are viewed by one source as ‘not a realistic candidate’ due to the team’s unwillingness at this point to include Joakim Noah in the deal. The Bulls and Noah are currently negotiating an extension.”
http://ken-berger.blogs.cbssports.com/mcc/blogs/entry/11838893/24605774?source=rss_blogs_NBA
Yay!
Note that the Dampier sweepstakes could alter your chart a little. If Miami get him they are just a better number one. If Houston gets him (they’ve offered more money for two years), they could improve significantly. Several other teams have expressed interest, but only a few will offer more money than Miami, and Dampier has expressed interest in playing for a championship.
Houston is also going after free agent Fesenko, who most assumed would return to Utah.
arturogalletti
09/16/2010
Tim,
, I’m being outpaced by events right now.
This is why I’m holding off on a Free agent Update
arturogalletti
09/16/2010
Although Dampier to Miami could possibly cause the Universe to collapse. Their top six then would give me 77.3 wins for last year.
Devin
09/16/2010
Again, wow – that’s a lot of work. You can bet I’ll be referring to this stuff a lot in the future. I’m going to be busy for the next couple of days though…this makes me want to write a bunch of posts. Once all the tweaks have been done, it’ll perfect for trade analysis. It’s like Bill Simmons list, only objective; if you can trade one player for another player higher on the list, you do it. If you can trade two players for one player who’s higher on the list and one player who’s lower on the list, you do it (and then find a way to upgrade the lower-ranked player later).
arturogalletti
09/16/2010
Devin,
Again this is why you’d have the job with me. As a rule of thumb, you always do the two for two if it leaves your top 10 about even in Lebron units (I get such a kick out of typing that) and makes your top 6 better. Practically you want to look at age and history but it’s a good reference tool.
Alex
09/16/2010
Not to nitpick, but aren’t all intangibles off the charts? I mean, since they can’t be measured and all
arturogalletti
09/16/2010
Can you spot the nitpick?
Ty Ahmad-Taylor
09/16/2010
Arturo,
I have been lurking since you started this blog, and I can say that, regardless of what was written in the post, that you have topped yourself with the Batman-lightsaber-Jaws montage. Still cleaning up bits of my brain after looking at that.
Secondly, this is a fantastic post. While fully taking into account your caveats, I believe that your preliminary model is fascinating in and of itself, especially with respect to what the Knicks did in the offseason (as well as my hometown Warriors.)
Are you going to make model updates in subsequent posts, or via post here?
Ty of FanFeedr
arturogalletti
09/16/2010
Ty,
I blame Drew Curtis for putting that image in my brain. It does just tick just about every check-box of awesome.
One of my principal goals with writing this blog is to build and refine this model. I will definitely keep updating and improving it. It will be one of my regular features: The Build, The Half-Baked Notion, 30 to 16 to 1, The Sunday kind of Piece, Build me a winner. I do need to redo the formatting to make things easier to find. The next thing I’m going to add is the rookie model (which was almost done but I had to go to sleep
)
Chicago Tim
09/16/2010
Okay, you’ve got me excited about the season already. The Bulls have a chance to be the second best team in the league, and an outside chance (granted, very outside) of an upset over the Heat. That’s barring injuries, of course, and they could be even better if their young players improve. It’s a lot of “ifs,” but it’s enough for me. Let’s start the season, already! Or even the preseason.
Based on your chart, these should be a good preseason Bulls’ games: October 5 at Milwaukee, October 7 at Dallas, October 15 Dallas at Chicago, October 16 at Orlando. Then they open October 27 in Oklahoma, Portland visits November 1, they visit Boston November 5, and then there’s the Circus Trip (while the circus visits Chicago) November 16-27, including games in L.A. (Lakers), Dallas, San Antonio, Phoenix, then back home to face Orlando, then off to Boston … man, what a roller coaster! And I’ve left out those games against lesser teams that they should win, but those are the “trap games.”
My only worry is that the Bulls may get off to a slow start while they learn to play with a new coach and mostly new teammates. They have a tough schedule at the beginning of the season, and face a number of teams who have not changed as much since last year. They could easily end up fourth in the East and face Miami in the second round. I would really like to see them finish third or higher so they have a chance to win a couple of rounds before facing Miami.
greyline
09/17/2010
Great work!
I realize not everything has been factored in, but wow, New Jersey in the playoffs?
Oh, wait. Acording to the chart, New York makes the playoffs because of a slightly better regular season. Still. Nice to know the eastern GM’s are earning their keep.
some dude
09/17/2010
Camby 3rd. Portland #2.
yeah, and I got a bridge to sell ya!
arturogalletti
09/18/2010
Camby has taken a team to the finals.
As for Portland, I did say preliminary (got to work in injuries) but that’s a tall roster.
evanz
09/18/2010
Nice to see the Warriors headed back to the playoffs. Even if it’s only a projection at this point.
some dude
09/18/2010
Camby has taken a team to the Finals?
he played 24mpg in the playoffs and it was the lockout year. No one can lead a team to do anything playing just 1/2 the game. But nice try.
In fact, he hasn’t been passed the 1st round in the last 10 years and failed to make the playoffs multiple times. Explain to me how a top 5 player like Camby can’t even bring his team to the playoffs or past the 1st round in TEN years?
He is one of the 3 most overrated players in the league by fans and stats. Stats don’t value his contributions properly. He’s like Battier, only from the other end of the spectrum. He steals rebounds from teammates, cheats for rebounds at the expense of giving open shots, over helps for blocks, doesn’t do anything in the post and when the outside shot isn’t falling he’s a liability on offense.
Too many weak defensive players that over-rebound (Lee, Murphy, Camby, Boozer) are too high on the list because the stats can’t tell the difference between a smart rebounder and one who over rebounds. It values all rebounds the same when that is not true.
The system needs to be tweaked. I don’t understand why you defend it when Camby being 3rd on this list should be about as big a red flag as can be, even if you disagree that he’s overrated.
If you’re a top 3 player, don’t tell me you can’t get out of the 1st round in 10 seasons with decent quality players (which he’s had).
arturogalletti
09/19/2010
sd,
He put up 3.9 wins as the player with the 4th most minutes in the playoffs for that Knicks team and an ungodly .327 WP48. That team was not great either. The reason they did well may be related to Ewing getting hurt and Van Gundy upping Camby’s minutes.
As for teams? Mark Jackson, Birdman and T-Mac are the only >.200 WP48 teammates up to this year in Portland (and that team was totally banged up). I expect that team to be clearly better this year (though not number 2 perhaps)
evanz
09/19/2010
It would be worthwhile to try using WS48, in addition to WP48. There are some important differences, including the fact that guys like Camby, Murphy, and Kidd tend to drop down quite a bit.
arturogalletti
09/19/2010
Evanz,
Take a look here. and here:
WS48 is good and has correlation year to year (which PER doesn’t) but it’s not close to as good as WP48 at predicting wins.
Evanz
09/19/2010
Thanks, Arturo. I was looking for a good explanation like that.
BTW, I noticed that you’re a fan of xkcd.com (as am I). You really should attribute his work, as per the instructions on the website:
http://creativecommons.org/licenses/by-nc/2.5/
arturogalletti
09/19/2010
Evanz,
Hmm. I thought I was complying by linking straight to his provided link. I’ll see what I can do to add the link to his page (perhaps as a note?).
Evanz
09/19/2010
Where is the link you are referring to? I’ve looked all over your blog, and don’t see it.
arturogalletti
09/19/2010
The image is the one provided from his site. It’s also on the about page now. I’m going to add as a note to the different articles.
evanz
09/19/2010
Yeah, I would do that, just to be safe. If you go to the About page on xkcd.com, it says:
Note: You are welcome to reprint occasional comics pretty much anywhere (presentations, papers, blogs with ads, etc). If you’re not outright merchandizing, you’re probably fine. Just be sure to attribute the comic to xkcd.com.
arturogalletti
09/19/2010
Done.
evanz
09/21/2010
Just wanted to clarify something here that I think is important. When I mentioned “WS48″, what I was referring to are “win shares/48 minutes“, as discussed at bball-reference here:
Calculating Win Shares
Comparing Win Shares (29) and Wins Produced (31) for the GSW of 2009-2010, it turns out that Win Shares is actually quite a bit closer to the actual wins (26). On the bball-ref website linked to above, it says that the RMS error since 62-63 is 3.41 wins. That seems to be very comparable to WP.
arturogalletti
09/21/2010
WP and WS are virtually identical when it comes to predicting wins. Over time WP is more consistent (Consistency is the correlation between per-minute performances in successive seasons). Prof. Berri has forthcoming piece (Berri, David J. “Measuring Performance in the National Basketball Association.” In The Handbook of Sports Economics, eds. Stephen Shmanske and Leo Kahane; Oxford University Press, forthcoming.) that covers this. I’m hoping to cover this in a review piece.
evanz
09/22/2010
Great. I’m looking forward to it. Thinking about this last night I made a list of the Top 12 players from 2009-10 by averaging the two metrics, and I like how it came out compared to either alone:
LeBron James
Dwight Howard
Kevin Durant
Pau Gasol
Dwyane Wade
Chris Paul
Tim Duncan
Manu Ginobili
Carlos Boozer
Steve Nash
Chris Bosh
Gerald Wallace
LunaticWolf
09/20/2010
Utah at 16? Below GSW, Sac-town, and Milwaukee?
Did you not include Jefferson in this?
Evanz
09/20/2010
If you go by last season’s WP48, I think this is how it works out by the numbers. Jefferson’s WP48 last season was 0.145 compared to 0.290 for Boozer. That’s a big loss, whether Jefferson, Okhur, or Millsap take those minutes.
Gray Jay
09/20/2010
I appreciate the great amount of work Arturo’s put into this first draft. I come up with different results for the Rockets, no doubt based on different minute allocations, but that’s a minor nitpick. And some dude‘s comment needs to be stickied at the top of every Rockets’ message board this year:
“[H]e played 24mpg in the playoffs and it was the lockout year. No one can lead a team to do anything playing just 1/2 the game. But nice try.”
All I can say is that if you have faith in this model, correlation to wins aside, then you should go get a bet down on Golden State getting to 45 wins and getting into the playoffs ahead of Utah and Houston. I remember reading the “Warriors project to 50 wins” post at WoW and frankly view it as the basketball statistics’ version of Poisson’s Spot. We’ll see.
Thanks again for sharing the product of the great amount of work you’ve obviously put into this.
arturogalletti
09/20/2010
Gray Jay,
Point sources of light indeed. Made me laugh out loud. Wait till you see my lockdown surprise team in the final model.
some dude
09/20/2010
“He put up 3.9 wins as the player with the 4th most minutes in the playoffs for that Knicks team and an ungodly .327 WP48. That team was not great either. The reason they did well may be related to Ewing getting hurt and Van Gundy upping Camby’s minutes.”
What do I care what Camby did last century? We’re talking about Camby now. Camby was a much less selfish player at the beginning of his career did he’s been the previous 6 seasons.
10 years, no 2nd round games. none. zilch. He played 24mpg for that Knicks team. He played only 16 minutes in one of their 2 Final wins.
As for Portland, they’re not #2 in the league. Lakers #1 in the West too. Don’t be fooled by LA’s releasing of the throttle the final 2 months of the season. Phil Jackson’s final hurrah and the Miami issue – LAL will play the full season this year unlike last.