NBA Now Rankings as of 1/29/2011

Predictions are a tricky business. The more specific you are, the more likely you are to be wrong. But the public loves them and so writers and analysts constantly put themselves out there to feed the beast. As with everything, there are some interesting tricks of the trade.

Cool website of the day informationisbeautiful puts together some fantastic visualizations. They recently did a great project where they mined 22,000 horoscopes for commonly used words (Go here for a full look). The relevant quote:

“How do you gather 22,000 horoscopes? Obviously you could manually cut and paste them from one of the many online Zodiac pages. But that, we calculated, would take about a week of solid work (84.44 hours). So we engaged the services of arch-coder Thomas Winnigham to do a bit of hacking.

Yahoo Shine kindly archive their daily predictions in a simple and very hackable format (example). Thank you! So Thomas wrote a Python script to screen-scrape 22,186 horoscopes into a single massive spreadsheet. Screen-scraping is pulling the text off a website after it’s displayed. Python is a programming language. You can use it to write scripts that only gather the specific text you want. Then you run it multiple times so it mines an entire website.”

The result of this experiment is the following wordcloud:

And they used the results to produce the following metahoroscope of the most commonly used words:

The point? All horoscopes look the same and use non-specific words to convey a feeling rather than concrete action. With that in mind, here’s my touchy-feely version of the power rankings for the NBA so far (games are current as of 01/29/2011 and I excluded all Cavaliers games as I felt they were warping the numbers, NBA teams only ). Remember these are what I feel is the situation at the moment, keep that in mind and be sure that it could all change :

San Antonio feels like the the best team in the league right now. Miami,LA, Boston keep having the expected issues with injuries but they may recover promptly. Orlando and New Orleans are hot at the moment but that energy may not last forever. They all need to keep feeling the love within them to rise to the occasion and possibly contend. Maybe.

Feel the Love?

Memphis and Philly keep charging up the rankings to the point that if they keep doing so my feelings tell me that they could even win a playoff series. Then again they might not.

 

Cleveland is maybe hard. Changing their mind and a better mood might come along (but I’m not holding on to hope).

 

45 Comments

  1. Chicago Tim
    1/30/2011
    Reply

    “Miami, LA, Boston keep having the expected issues with injuries but they may recover promptly.”

    Don’t forget, the Bulls are getting Noah back.

  2. 1/30/2011
    Reply

    Sure? Enjoy the secret moment a lot and remember the world is energy.
    :-)

  3. E. J.
    1/31/2011
    Reply

    SAS at the top sure feels right to me. I’d love them to keep it up.

  4. Leon
    1/31/2011
    Reply

    Information is beautiful is a beautiful book.

    The recurring theme from your rankings Arturo is that it is wide-open as to who will reach the conference finals, let alone who will win the championship. Playoff time will be very interesting and exicting one must feel.

  5. entityabyss
    1/31/2011
    Reply

    Cleveland’s really starting to bother me because they shouldn’t be this bad. They really shouldn’t.

    • 1/31/2011
      Reply

      EA,
      Two things going on I think.
      1. All the defensive analysis I’ve done suggest that Lebron has a huge impact on the defensive end (much like Howard and Garnett) . All the offensive analysis suggests that he has a further impact on offense (think Melo effect on steroids). So we completely and totally underrated him (and we had him as a 30 win guy last year). Lebron and a d-league roster gets you in the playoffs in the East.
      2. With the injuries they’ve had they’re playing people out of position and that is a recipe for total and utter disaster.

      At least that’s my feel of the situation.

      • EntityAbyss
        1/31/2011
        Reply

        They weren’t playing that bad before they played Lebron, but more importantly, a lot of these guys played on teams without Lebron and they played waaay better than they are now. It’s hard for me to believe Lebron can do all that. Mo Williams, and Antawn Jamison played on different teams before Cleveland. I think they’ll turn it around. This is ridiculous.

  6. some dude
    1/31/2011
    Reply

    Cleveland is playing D-leagues out there legit minutes. This isn’t a normal team. More than just lebron left, too. And the ones that haven’t aren’t playing much (Parker, Moon, Varajeo) for various reasons.

    Obviously Lebron’s departure had a major impact. But this team isn’t the same just minus Lebron. It’s a completely different team. I think something that’s also missing is that the Jamison trade made the team worse and they weren’t really a 60 win team anymore, so their ceiling was not adjusted properly.

  7. EntityAbyss
    1/31/2011
    Reply

    Hey Arturo, I know I make lots of suggestions for future posts, but I have a great idea.

    I read a comment by dberri in a post and brought an idea to me.

    “One issue one should have with line-up studies is how little time any one line-up seems to spend together across a season. Looking at the 82games.com data, it appears most of these line-ups are together for less than 500 minutes.

    There is also the issue Dean Oliver raised some time ago. Players are not trying to win 3 minutes in the second quarter. They are trying to win the entire game. Players will vary their effort level across that game, which means looking at specific segments is going to run into difficulties. This is one reason why plus-minus data has such problems.

    That is why I prefer to look at aggregate data. In Stumbling on Wins we talk about how a player’s shooting efficiency varies with shot attempts. And we talk about how one player’s shot attempts are impacted by their teammates’ attempts. Each of these studies control for a variety of factors. And neither seem to support the notion that shooting efficiency is impacted tremendously by shot attempts or that getting shot attempts in the NBA is difficult (for NBA players).” – dberri on line-up data.

    the second paragraph was the most important to me. Players don’t play consistent throughout a game or even throughout a season, but the full season numbers tend to be consistent. This is probably why line-up data is inconsistent. Players aren’t consistent all the time, but as a whole are consistent.

    For instance, I’ve seen posts on other websites that show diminishing returns in respects to certain statistics with respect to line-ups. For instance, rebounding. Although the R^2 are low, the studies show what seem to be high diminishing returns with respect to rebounds. However, when I checked full season data on high rebounders, the results showed that the diminishing returns were very small.

    (examples-
    Marcus Camby played 29 games and 619 minutes in the 02-03 season for the denver nuggets. The next season he played 72 games and 2162 minutes. For each player on the 03-04 nuggets that played 1000+ minutes, and Rodney White who played 985 minutes, using their 02-03 rebounding percentages, in the minutes they played in 03-04, they should have expected 2574 rebounds, but grabbed 2525 rebounds, which is 49 rebounds less than expected. If Marcus Camby gets all the blame, then in (2162 – 619) 1543 minutes he cost the team 49 rebounds, although, he himself got 27 less rebounds than expected. That would mean his teammates that played 1000+ minutes (and rodney white) lost (49-27) 22 rebounds if marcus camby is to blame. if he gets the blame, he caused a loss of (27 / 1543 * 36) 0.63 rebounds per 36 minutes in diminishing returns.

    Doing the same thing For Ben Wallace (another big rebounder) for when he moved to Detroit in 00-01, he caused a loss of 46 rebounds in 2760 minutes (once again, if it’s all his fault) which is (46 / 2760 * 36) .6 rebounds less, that the rest of his team got. )

    I only showed 2 examples. I could see more, but the point is, while these sites shows posts showing large diminishing returns with line-up data (not just rebounding, but all stats, especially assists), it seems that the end of the season, the numbers correlate closely, and the aggregate data shows small diminishing returns with respect to the stats. As a matter of fact, I don’t see any diminishing returns with offensive rebounds with aggregate data, but the blogs and sites that have shown line-up data suggest that players steal 0.3 offensive rebounds from teammates from each one, they get.

    I tried it in a few examples, however, so I was wondering if there could be a post on line-up data vs aggregate data. What dberri said, seems to be true. Players aren’t consistent throughout a game or throughout a season, but at the end of the year (besides the effects of injury), the stats are consistent.

    Long comment, but my point is, someone should do a post on line-up data vs full season data. :-)

    • some dude
      2/1/2011
      Reply

      Your Camby example is hogwash. You use Carmelo in your sample of “expected” when it was his rookie year. Second, Andre Miller had the worst year of his career in ’02-03 and it was his only year below 6.4 TR% of his career at 6.1. His career average is 7%. So yeah, when you predict based on an outlier year, of course you’ll get better results.

      How about we take Melo out, since he had no 02-03, and use Miller’s career average of 7.0 which is still biased since the 3 years before the clipper season he was 7.5, 7.3, 7.4.

      Then Earl Boykins also had the worst rebounding season of his career the year prior. Why? because he played on one of the best rebounding teams in the entire league, the ’02-’03 GSW. That he jumped up is not evidence for Camby but the opposite. More evidence of the diminishing factor. In fact, the 4.4 TR% Boykins put up in ’03-04 was close to his career average at the time (his 10th season). Again, I’ll be generous and use the total career average of 3.9% even though that’s also biased in your favor.

      That brings us to projected total of 2645 rebounds (non-Melo) versus an actual total of 2525.

      If we re-adjust to be fair the Miller and Boykins and give them their normal projection (say exactly what they got that year) and I’ll even reduce Barry’s projections (he had a career high the previous season), it would be more like 2660 projected teammate rebounds.

      135 rebounds is a LOT of rebounds to be off.

      Denver was out-rebounded in 02-03 3521-3470 for a team TRB% of 49.6%

      Switch those 130 rebounds and Denver outrebounds opponents 3600-3391 for a team TRB% of 52.5%

      That is HUGE. Denver’s efficiency splits were 103.9 to 102.7. Switch 130 rebounds and what does the projected efficiency splits become? Is it 105.5 to 101.2? I forgot if I did that part right. If that’s right, they were projected to be a 53 win projected team and not a 44 win projected team. Yeah, 9 wins sounds a lot to me.

      It’s obvious looking at the TRB% of Camby’s teammates over time…

      Even if you take Camby’s numbers out, it’s still a big gap in rebounds. Remember, each rebound you don’t get the opponent does get. Those 100 rebounds matter.

      Yeah, when you mischaracterize the rebounding numbers like with Andre Miller, Melo, and Boykins, you can get closer to your desired result. But those numbers were misleading. Not sure if it was intentional, and I will give you the benefit of the doubt, but you have to look at the context of your numbers to see if you are correct.

    • some dude
      2/1/2011
      Reply

      i gotta expand on what i said.

      If we look at Camby’s teammates, roughly 100 rebounds short of projection, we have to blame Camby for this. These are rebounds the team would have had anyway and he just reallocated them.

      He had 627 rebounds when you subtract the 100 from his list. Roughly 14% of his rebounds were “stolen” from teammates. 14% of his rebounding numbers are inflated in his WP numbers.

      Also forgot, Nene played games with Camby in ’02-03, which probably depresses his projection some. Camby also rebounded less than usual, so his affect on his teammates were less. My guess is that 14% number was his lowest as a Nugget.

      here’s a question for you EA. Why is it that the ’04-05 Nuggets were a better rebounding team (by a lot) than the ’03-’04 team. Melo, Camby, Boykins, Miller, were all the same or worse at rebounding. Kenyon Martin replaced Birdman but rebounded at a worse %. Why, is that Nene’s rebounding % going up? Why, according to 82games.com it seems the Nene-Martin frontline rebounded the heck out of the ball better than Camby frontlines. Martin is a good defender, so could….gulp, defensive play have to do with this!?

      Also interesting to note that the new additions (besides Russell) to the 05 squad had a sudden drop in TRB%. :D

    • Guy
      2/1/2011
      Reply

      EA, this is just sad. There is no serious dispute any more about whether there are large diminishing returns on rebounds, at least on defense. If you don’t like lineup studies, look at Phil Birnbaum’s recent posts on rebounds at Sabermetric Research. He looks at full season data, and finds that the more rebounds a team gets from any given position the fewer rebounds it gets from the other four — and the effect is enormous.

      Or let’s use bigger data sets with even less noise, by examining player careers. Camby’s teammates, over his entire career, have secured hundreds of fewer rebounds than we’d expect from average players. Perhaps Camby has just had bad luck in teammates, you might wonder. But this has also been true for Ben Wallace. And guess what — it’s true for every other “great rebounder” as well.

      Read Berri’s FAQ. Even he now admits that 50% of defensive rebounds come from teammates. In another 5 or 6 years, he’ll probably acknowledge diminishing returns on offensive rebounds too. WOW is kind of like a time capsule: it puts you on the cutting edge of basketball analytics, circa 1999.

      • bduran
        2/3/2011
        Reply

        Guy,

        You like to use misleading statements. If I just read your statement I would think, “This Berri guy has been denying the effect of diminishing returns on rebounds and now admits that its exists and is significant.”

        This is not really what his FAQ says though. First off, he says he uses the effect of diminishing returns from his own book and research. So this is not a new thing for him. Secondly, he says that it does little to change his player rankings when using the recalculated WP48. This of course has been pointed out to you over and over again.

        • Guy
          2/3/2011
          Reply

          I said nothing that was “misleading.” Berri appears to report in his FAQ that teams realize only 50% of the defensive rebounds you would expect looking at a player’s individual total (and I acknowledge Berri’s writing here is a bit unclear; if he says that is not what his research shows, I’ll happily retract the statement). As far as I know, he has not reported that result previously, or if he has it’s certainly not a figure he has used for most of the time he has been writing about basketball. So my saying that he “now” reports a 50% rate is accurate.

          I did NOT claim that Berri admits this 50% rate “is significant” (your words). He certainly does not acknowledge that, as I assume you know. Indeed, he claims that his results show “diminishing returns doesn’t make any real difference.” And despite the fact that Wallace, Camby, Howard, and Boozer all lose about 4 wins each using his re-estimated WP, he says that the results when altering WP accordingly “are not very different.” To me, these are not recognizable uses of the English language. If I had believed Berri when he said on multiple occasions over the paast few years that changing the coefficient for rebounds in WP “does not change your results,” I might feel deceived upon seeing the actual impact in his FAQ table. Perhaps that is your situation, bduran — but if so, please don’t take your frustration out on me.

          It also seems curious that Berri does not use this new version of WP, even though his own research seems to suggest it is more accurate. Why deliberately keep using an inferior version of the metric?

          • bduran
            2/3/2011
            Reply

            I said misleading, not lying. The way you state things implies something that isn’t true. Saying, “Even he now” implies that he has finally come around. Yet his FAQ repeats something he has stated elsewhere, which has been pointed out to you.

            He also reports no significnat difference in player rankings. That is the point he is making. He did said nothing about whether or not 4 wins gained or lost by a player is significant.

            • Guy
              2/3/2011
              Reply

              This is just ludicrous. Saying that Berri “now” reports 50% diminishing returns on drebs means only this is a relatively new finding for him, certainly post-WOW and I think post-Stumbling. You say he has “stated it elsewhere” previously. Can you show us evidence of that? If so, then I missed it and will gladly correct my statement. If not, then please quit making false charges. (And the phrase “even Berri” was simply in contrast to EA, who appears to believe the rate of diminishing returns on drebs is much lower than 50%.”)

              I have criticized Berri on this very blog for refusing to acknowledge how substantial diminishing returns are, even according to his own data. You have commented in those threads, so you must be aware of this. Why in the world would I try to suggest Berri recognizes something that I clearly think he doesn’t? Obviously, you read something into my statement that wasn’t there. Why don’t you just admit you misinterpreted it, and move on?

              And if you are patrolling the internet for “misleading statements,” a good place for you to start would be Berri’s repeated claims that it “makes no difference” if you change the rebound coefficient to .7 or .3/.7. Surely you recognize that his prior practice of reporting only the correlation between WP and adjusted-WP would mislead many people into believing that using new coefficients would make, um, little difference. Since Berri developed and promotes a metric which rates players’ per game productivity to the third decimal place (1/1000 of a win), I find it hard to imagine he really believes that 4 wins “don’t matter much” or is a “very small” difference. And even if one agrees these differences are “small,” why not improve the metric this small amount? If indignation is your thing, it would be far better directed at Professor Berri….

              • bduran
                2/3/2011

                From the FAQ

                “This adjustment follows the diminishing returns results reported in Stumbling on Wins for defensive rebounds”

                I’ve bought his book, but leant it to a friend so I can’t verify what he reported.

                I don’t patrol the internet. I go to blogs and read posts and find that at many of these blogs you post long diatribes with inflammatory statements such as

                “This is just ludicrous.”

                and misleading statements such as

                “Even he now admits”.

                You’ve often said Berri doesn’t admit diminishing returns exist, when he has all along but has only argued about it’s significance. This is very different. I’m fine with you taking exception to it’s lack of significance, but it’s very misleading to say he doesn’t believe in it. Things like this are a consistent theme in your very long and prolific posts.

                You, among others, have led me to re-evaluate how I think of rebounding numbers. However, it annoys me to give me you any credit because of the way you write your posts.

              • Guy
                2/3/2011

                You show up here and begin your comment “Guy, you like to use misleading statements,” and now you say I make “inflammatory statements.” LOL. You claimed that I implied Berri had acknowledged something which I have said publicly many times he does NOT acknowledge, even though you knew all that. And when I pointed out I had said no such thing, you stuck to your claim. So I’d say “ludicrous” is a kind description.

                Still you have produced zero evidence that Berri had reported the 50% rate prior to SOW, or even prior to his FAQ. So I can’t see a problem with my statement that Berri is “now” reporting that result.

                And now you add a new slander that you also know is false. I have said many times that Berri believes diminishing returns are “very small,” or “don’t matter,” not that he denies they exist. We’ve even had this exact same discussion before. If I have at some point in a blog comment used a shorthand description that implies Berri believes DR are zero rather than “very small,” I happily retract it. But this distinction can’t be very important, since Berri’s own position is that the actual extent of DR is not significantly different from zero. The distinction between Berri’s view and saying that DR don’t exist at all must be trivial — by Berri’s own standards! So this is much ado about nothing.

                I don’t know what your problem is, bduran, but I hope you’ll have the good manners to apologize….

              • Gabe
                2/3/2011

                Guy:

                Your comments in response to other peoples generally innocuous posts/comments often are shot through with an unappealing mixture of hyperbole and smug condescension. This makes it extremely difficult (at least for me) to actually give a crap whether or not you are correct.

                The 50% part of Berri’s FAQ was to show that WP isn’t wholly tied to rebounds. From my understanding, Berri’s argument is that the effect of DR on rebounds has a “small” effect on WP. This is why he used the 50% example. His point was that players who are productive according to WP are still productive even if you divide their rebound totals in half.

              • Guy
                2/3/2011

                Hi Gabe: I’m guessing you’ve mainly read my comments here, or at other WOW-affiliated sites. True?

                And can you give me a couple examples of “hyperbole?”

              • Gabe
                2/3/2011

                Guy: True.

                Hyperbole: “ludicrous,” “slander.”

                Condescension (with a dash of hyperbole): “if you want to be the last person in the world denying the reality of diminishing returns on rebounds.” “WOW is kind of like a time capsule: it puts you on the cutting edge of basketball analytics, circa 1999.”

                It’s not enough for you to simply make your point, you like to make your point while implicitly calling the person you are debating an idiot (at best).

              • Guy
                2/3/2011

                Gabe: Ah, that explains it. You have to consider the context.
                If you peruse other sites (abprmetrics, The Book blog, Sabermetric Research), I think you might get a different impression.

                Still, it’s not a totally unfair criticism. But if I’m recalling your past comments correctly (which I may not be), I think you are a pretty big fan of David Berri’s. If so, I’d say you have forfeited your right to complain about “condescencion.” :>)

              • bduran
                2/4/2011

                Here’s let try something.

                I can only assume for your comment that you lack even basic reading comprehension.

                Now, was that a nice way to phrase what i’m thinking?

  8. entityabyss
    2/1/2011
    Reply

    Guy, david berri never said that 50% of defensive rebounds are taken from teammates. At least, I didn’t see that.

    Also, from the research phil birnbaum did, apparently, point guards steal every assist from their teammates, and some. So that would suggest that steve nash, cp3, or rajon rondo are hurting their teams with all those assists, so… I’m not buying it. I’m pretty sure the R^2 for those graphs were very low. Do you want to believe that derrick rose hurts his team every time he gets an assist? As a matter of fact, every thing you’ve posted that showed diminishing returns effects, and included assists in it, showed the greatest diminishing returns with regards to assists (not any rebounding). That would suggest that bulls could switch d rose for monta ellis (more efficient scorer, turns over at a lower rate and better with regards to steals) and maybe be the favorite in the east. I don’t find these results to be very true.

    As for some dude, I’ll check and see.

    • Guy
      2/1/2011
      Reply

      Read the FAQ, EA.

      Interesting that you have to switch to talking assists. I assume that means you have no answer for the findings on rebounds, nor the career statistics I cited. Which makes sense, since there is no answer. Can we agree it’s case closed on rebounds, before moving on to assists?

  9. EntityAbyss
    2/1/2011
    Reply

    Ok Guy, I read the faq. Wins Produced was re-estimated with defensive rebounds worth half as much. It was to show that there wasn’t much of an effect. I don’t know if I misread it, but it seems that he did not that there’s 50% in diminishing returns on rebounds. So, that’s not true.

    Secondly, I’m not moving away from the argument of diminishing returns on rebounding. I’m trying to say that those studies done to show it, that you show on this blog, are done badly. I’m pretty sure that the R^2 in those graphs to show the diminishing returns are very low. Therefore, the I’m not buying the results of the study. I was saying that if you do accept the results, you would have to accept that the biggest diminishing returns in basketball are in respect to assists, and that point guards hurt their teams with each assist that they get. I absolutely don’t believe that Steve Nash’s assists hurt the suns. I find it interesting that in those studies, people ignored the graph that showed the biggest diminishing returns effect – assists.

    As for some dude, I checked my files, and I belive you’re mistaken. I’ll check the rebound percentages of the teams of the players for the year prior later. I got class in a few minutes. What I will say is this. I did NOT include Carmelo Anthony as he was a rookie, I included the players that 1000+ minutes and rodney white who played 985 minutes. I did not include any rookies. That list includes Andre Miller, Voshon Lenard, Nene Hilario, Earl Boykins, Marcus Camby, Rodney White, Jon Barry, and Chris Andersen. These players combined for 2525 rebounds. Had they kept their rebound percentages from the season prior, they would have collected 2574 rebounds. that’s a -49. Not including Marcus Camby, that’s a -22. I did not adjust for factors like age, and percentage of rebounds collected year prior. I’ll look at the year prior later, after class. Anyways, I also did the same thing for Ben Wallace and saw little diminishing returns.

    The studies that show huge diminishing returns tend to look at line-up data (also tend to not adjust for many thing and end up with low R^2, so low explanatory power).

    • some dude
      2/1/2011
      Reply

      I got 2525 too, so we used the same players then and my bad on thinking you used Melo.

      However, your projection is bullocks because of this: you use the previous year even though it doesn’t represent the player’s rebounding.

      A. Andre Miller’s 2003 rebounding number was a career low by a LOT (and his worst statistical season in the NBA period). The previous 3 seasons his rebounding % was higher by a lot. You’re using Andre Miller’s worst season to project his future season. That is absurd. At the very least, one should use all relevant historical data to make projections. You cherry-picking one season is ridiculous.

      B. Same for Boykins. He came from an amazing rebounding team in ’03 to join a weak one in the Warriors in ’04. Of course his rebounding went up. It couldn’t go any lower. His ’03 numbers were way below his career norm and the worst of his 1st 10 years in the NBA.

      To me, it seems you cherry-picked a season where players had their worst seasons or went from better rebounding teams to lesser ones to mask the effect.

      Here’s what we know. Nearly all the players were below their career norm (or recent norm) playing with Camby. If you do an HONEST pojection, accounting for more than just the previous season, we see that the projection is off by about 130 rebounds, or 100 (or was it 105) without Camby.

      In ’05 nearly every single player besides Byron Russell and Nene were below their career norms or projected norms for the season. And looking at 82games we can find that a lot of that resulted from the Nene-Martin line-up.

      And this scenario plays out over and over again.

      Hey, what about David Lee in Golden State? Save the elbow injury, we’re talking rebounds. Why is is that Golden State rebounds better when Lee is on the bench by a significant margin even though he’s their best rebounder? What happened to all those projected rebounds for his teammates?

      • some dude
        2/1/2011
        Reply

        And I scaled Jon Barry’s numbers down because he had a career high the previous season, so I didn’t do it one-sided.

        Andre Millers 1st four season’s rebounding:

        4.5
        4.3
        4.4
        6.1

        And he never got near 6.1 again. But you use 6.1 as you projection point. So absurd.

        • EntityAbyss
          2/1/2011
          Reply

          I also did a Ben Wallace example, but uh, let’s work with this. If I remove Miller and Barry out the picture, that’s an expected rebound total of 2139 and an actual rebound total of 2036. That’s a difference of -103. Marcus Camby dropped 27 rebounds for the expected, so his teammates were (-103 + 27) -76 from the expected. Adjusting for pace, (denver had a pace of 93.3 that year so 100/93.3 *76) 81.46 rebounds lost in a 100 possession per game season. that’s a loss of about .99 rebounds per game and about 2.63 wins lost by diminishing returns.

          So, doing it your way (and still giving Marcus Camby all the blame), Marcus camby loss the nuggets 2.63 games. Once again, that’s all him (I personally don’t believe he caused all of that, but if you do, ok).

    • Guy
      2/1/2011
      Reply

      From Berri’s FAQ: “Recently WP48 was re-estimated, but this time a defensive rebound would only be worth half as much as points, field goal attempts, offensive rebounds… , turnovers, and steals. This adjustment follows the diminishing returns results reported in Stumbling on Wins for defensive rebounds.”

      Now I admit this isn’t the most clearly written statement in the world, but it sure sounds like this latest re-estimation of WP — which uses a coefficient of .5 for defensive rebounds — is based on Berri’s own research on diminishing returns. Unless you think he chose a coefficient that was inconsistent with the finding of that research, we can assume he found something like a 50% rate of diminishing returns.

      But if you want to be the last person in the world denying the reality of diminishing returns on rebounds, knock yourself out…..

  10. EvanZ
    2/1/2011
    Reply

    EA, what exactly does this mean:

    “Players aren’t consistent throughout a game or throughout a season, but at the end of the year (besides the effects of injury), the stats are consistent.”

    I’m scratching my head trying to understand this statement.

    • some dude
      2/1/2011
      Reply

      What he means to say is that a player’s numbers fluctuate throughout the season and games. If you average 25 ppg, some nights you get 35, other 18. You don’t get 23-27 points every night. And you might score 20 in 1 quarter and 4 the next 2.

      But when you look at the end of season averages each year, he’ll be right around 25ppg.

      I think what he is saying is end of season averages are consistent over time despite the large amount of variance game to game.

      He’s trying to rationalize that it doesn’t matter if certain 5 man line-ups don’t produce as expected. How this does that, I have no idea.

      • EntityAbyss
        2/1/2011
        Reply

        How this does that is simple. When the line-ups play together and the minutes played by each line-up. Some guys score more in the 4th quarter than the 2nd, but that might be dependant on how aggressive or whatever they are at certain times. They might also play with the same line-up in the 2nd quarter. That is what I’m trying to say.

        • some dude
          2/1/2011
          Reply

          I understand that, but you’re running off an incorrect assumption that the end of season stats are consistent. The rebounding stats for non-elite rebounders are not that consistent.

          Look at the ’07 Pistons, the example I give below, for evidence of it. Sometimes the numbers appear to be consistent because something else is masking it. If an elite rebounder misses 20 games in a season, the other players will have their rebounding numbers inflated (someone has to get those boards when he’s hurt, right?). Etc.

  11. entityabyss
    2/1/2011
    Reply

    For some dude, I also gave a ben wallace example, and did more (I’m on my cell phone now). My whole point is that line-up data doesn’t seem to work because of several reasons. Players don’t play the same at all times of the game, so line-up data can get skewed (it’s not consistent). I’ve noticed that these studies on diminishing that keep getting posted here have low R^2s. I’m not buying them. Full season data (at least in the few examples I did, and in the one arturo did a few weeks back) show small diminishing returns.

    I want a study that explains a high amount of the variance. I doubt that can be done with line-up data (if it can, do it). It seems aggregate data would do a better job at that. Also, if that’s done, a lot of things should be controlled for.

    • some dude
      2/1/2011
      Reply

      No, they really don’t show small diminishing returns. I just re-did you Camby example because i knew it was wrong (we went over it before) and it’s about 130 rebounds off, which is an ENORMOUS difference.

      Line-up data works too because it shows that you cannot just put 5 ben wallaces together and get 100% rebounds or even 60% of rebounds. It doesn’t work that way.

      You claim this is countered by the fact that players still get their numbers. But this is not true at all. Your Camby example demonstrates pretty clearly that nearly every player he was with dropped from what we’d expect.

      The same happened in ’05. And I’m sure in every year after.

      I didn’t bother with your Ben Wallace example because if you’re math on Camby is wrong, why bother?

      I do know that the year Ben Wallace left he was replaced by 3 worse rebounders and they rebounded the same!

      Ben Wallace in ’01 is a dumb example because that team had major turnover. Not to mention, Ben took up most of Jerome Williams’ minutes, who was also a really good rebounder. It was also one of Ben’s best rebounding years because of so many injuries to other rebounders like Jerome, Joe Smith, Etc. And guys like Stackhouse did have career lows.

      Look at the ’07 Pistons. In Ben’s last year (’06) he had TRB% of 19.0, the lowest of all his Detroit years. Who replaced him? Webber, Maxiel, Nazr. ’06 TRB% were 15, 10.5, 17.8, respectively. All lower than Wallace taking up his minutes.

      Det ’06 Rebounding = 49.8%
      Det ’07 rebounding = 49.6%

      Big drop without Wallace! How could this be when his replacements were much worse at rebounding? Well, let’s look at it more closely.

      Webber, Maxiel, and Nazr rebounded at these rates respectively: 13.4, 11.7, 17.6. So Nazr was the same, Maxiel picked it up, and Webber playing on one leg was showing badly. Now to everyone else who played with Ben the previous season and 1000+ minutes.

      Prince: had career high in ’07 at the time, would continue to increase post Wallace
      Hamilton: 2nd highest % of career
      Billups: jumped from 5.0 to 5.6, highest of the last 7 years of his career.
      Sheed: Jumped from 11.7 to 13.3, 3rd highest of his career (2 of his 3 highest post Ben)
      Delfino: Career high of 11.5, up from 9.3

      So all 6 (including) Maxiel who played with Wallace suddenly became better rebounders, often posted career best numbers, when Wallace’s minutes were replaced by Webber, Nazr, and Maxiel (all worse rebounders). Amazing, huh?

      The team only lost .2 percentage points from their TRB%!

      • EntityAbyss
        2/1/2011
        Reply

        Cool Story. I called my friend who loves cool stories, but he didn’t like this one. I asked him why and he said.

        Jason Maxiell – played 159 in the 05-06 season
        Chris Webber – actually declined from the season prior (age maybe)
        Nazr Mohammed – came from the spurs and rebounding % slightly dropped
        Ronald Murray – went to the team

        These 4 new additions (jason maxiell additional minutes) came to the team when Ben Wallace left. Later, I’ll check the effects of his leavin on the other players. 4 players though, and they still fall short.

        • some dude
          2/1/2011
          Reply

          Why would you compare players from OTHER teams? Are you isolating the variables?

          Webber, Maxiell, and Muhommad in prior seasons and the ’07 were all WORSE than Wallace in rebounding on Detroit in ’06. That’s the only part that matters.

          Once we know that, we have to ask ourselves this. If Wallace’s replacements were all WORSE than Wallace at rebounding, how is it that Detroit saw NO SIGNIFICANT CHANGE in rebounding?

          I forgot to throw in McDyess. He also went up from 15 to 17, highest in 5 seasons for him!

          Also, Murray replaced Mo Evans and was half the rebounder Evans was. More evidence we should have seen a drop in rebounding, yet it didn’t come!

  12. entityabyss
    2/1/2011
    Reply

    They were all worse, but it’s 4 players doing the replacing. Also, I didn’t mean to nit-pick, but my point is (and hasn’t been proven otherwise) that the line-up data presented have had low R^2s so do not explain much. I don’t believe they could be taken as proof for either argument. If line-up data was consistent, then maybe there’d be consistency with +/- stats. What seems to be the case is that aggregate data is consistent and +/- data isn’t.

    Me or you picking examples doesn’t show anything, which is why I suggested arturo (because he has the tools) do a post on line-up data vs aggregate data, and for all stats.

    Or, I can believe the line-up data presented when it suggests that steve nash hurts the suns with each additional assist. I’m not too sure that’s true.

    The stuff I had was on rebounding and that’s why rebound became the topic, but I can use assists. I also believe that there aren’t big diminishing returns with regards to assists (I actually believe chris paul’s passing makes the hornets better). I also find it interesting how rebounding is the such a big debate (probably because guys don’t believe high rebounders over scorers) when all the data you should suggest even bigger diminishing returns with assists. Why doesn’t someone say WP is wrong because it over-values assists. CP3, steve nash, and deron williams are overrated.

    Lastly, if you’re gonna prove large diminishing returns, show a study that has many controls and thus a higher R^2. Please, for me.

  13. […] The Toronto Raptors — who are stuck in Indianapolis due to a winter storm — fall to 27th in the latest edition of SB Nation’s NBA Power Rankings and 29th in Arturo Galletti’s NBA Now Rankings. […]

  14. some dude
    2/2/2011
    Reply

    The R^2 is poor because there are too many omitted variables changing things year to year (line-ups, player movement, systems, etc). Goes back to Phil’s post linked earlier.

    I’m not sure what you mean by the Nash comment. I think there is a large problem with the way assists are tabulated in the NBA. I think a good percentage of them are not right.

    Exhibit A: http://www.youtube.com/watch?v=WGqn3oyGDb0 (and it’s not that uncommon!)

    Guys like CP3 and Deron and Rondo get assists they have no business getting credit for at all. I have this issue with turnovers too. If you sit through a game and tabulate assists on your own, you might be surprised what you’d find.

    Plus, assists are not like rebounds. Rebounds often just come to you or are a result of height/position on the floor. To get an assist are you doing something with the ball, not going to receive them. Doesn’t make any sense to base an argument against diminishing returns of rebounds because of assists. The dynamics are completely different.

    The point about rebounds is that someone is actively making a rebound someone else on the team surely would have gotten. This is not true with assists. You can statistically link cheese to the moon if you tried hard enough, I’m sure. Doesn’t make it right (correlation vs causation).

    The notion that there are DR in rebounds is borne out of watching the game. Looking at stats seem to confirm this (rebounding percentages). Running regressions supports the claim. Because assists come up with similar data doesn’t refute anything or mean assists have DR. That’s not how arguments work. The statistics cannot MAKE the argument, they can only support the argument.

    Though I will say, the way assists are tabulated make them overrated (note, I am not arguing CP3 isn’t the best distributer in the NBA or that his impact is any less) and the DR numbers probably exist because of how they are misappropriated.

    As for a study with more controls, it’s impossible. Too many variables in basketball to ever do that for any statistic at the individual level. Well, unless we get a lot more stats to use than we got right now. As for a higher R^2, I think you’re overrating the impact that has.

    I don’t get how anyone could still have doubts. If there were small DR, why is difference in all team rebounding rates so small?

    Since 2002 not a single team in the NBA has posted a better than 53.5 rebound % and only like 2 teams bested 53%. Are we to believe that not a single team happened to employ high enough rebounders so that they’d rebound more than 53.5% by random chance!?

    Occam would say it’s because of diminishing returns.

    Look, in most seasons, except when he hve an outlier team (GSW) teams are 47-53 range. Knowing that, there are roughly 200 total rebounds difference in a season (varies based on pace and shooting percentages and stuff). That’s about 2-3 per game that could go either way. So when you bring in an elite rebounder, at best he can only affect 2-3 rebounds total per game.

    And yet, we don’t even see 2-3 change hands because of one elite rebounder, let alone what Reb% projections tell us. Go research the data yourself. Enough has been posted on the topic. It should be a fairly obvious concept (I knew this well before I ever knew rebound % was tabulated) just by being familiar with the game.

    Golden State is still a terrible rebounding team. And they’re at their worst with David Lee, their best rebounder statistically, in the game. Ask yourself, from a basketball perspective, how this could be possible?

  15. 2/2/2011
    Reply

    Chicago Tim:

    Miller hasn’t really impacted LeBron playing point. He’s always brought the ball up the floor in certain situations this season & when he’s on the floor w/ Miller & Wade, they bring the ball up sometimes, too. LeBron’s role in the offense doesn’t really change if it’s Miller/Arroyo/Chalmers on the floor.

Leave a Reply

Your email address will not be published. Required fields are marked *