That's like saying the world is flat because it looks like it's flat (common sense, right?). It's a good thing Galileo doesn't post here. And it's not a skill ranking, it's a ranking of how big a difference each player makes to his team. Thus, the other 49 players aren't "better" than Yao, their absence in their respective teams makes a larger difference in the final outcome.
Also, the only way to compare players across teams in this list is to remove that player from his team, and think about which team would suffer more. The rankings are specific to the system each player plays in. Saying stuff like "well according to this list playing Ostertag instead of Yao should help our team" is completely ridiculous. It's like if i said that KG Shaq Duncan Elton Brand and Nowitzki would be a good team (they're all in the top 10). I'm sure you're all happy you got a reading certificate from Kumon for graduating, but apparently the meaning behind the words is completely lost for some of you.
The main reason some of you think it's flawed seems to be that you don't understand that if a player's back up is good, their ranking will be lower. If their backup is bad, their ranking will be higher. So, Yao's #50 ranking isn't that bad because he had Cato for a backup who played pretty good defense last season. If the season started tomorrow and we still didn't have a legit backup center, Yao would probably be in the top 5 in this ranking system. It takes less time to read the methodology than it does to knock it because you don't like or understand the results.
this thread has just turned into a slapping contest. I AM a homer. I guess his ratings put into account the fact that Yao gets double and tripled every time he gets the ball? IMHO stats are meaningless. Unless the stat u r talking about is 4-0 in the finals.
Anyone watched "I Robot"? Will Smith asked Dr.: What do you do? Dr. answered: !@#$... (something sounds very scientific) Will Smith: OK, now, tell me what do you do? Doctor answered: make robots more loke human Now, can anyone who read the Dr.'s analysis tell me what this ranking is really about?
Except Sam hasn't explained the list much at all. Or actually, the author himself is certainly overstating the goodness of his list. Because a list can be "factual" yet still useless. If I ranked all the NBA players based on just their speed, vertical, height, and strength, I can get a list that is factual and can be helpful to teams when looking at players. Yet at the same time, it's really quite pointless. As it exist right now, I really don't see the +/- system used by Rosenbaum to be that useful. Now, it doesn't mean I don't like the research. On the contrary, I'm thrilled that people are developing new, more objective ways to gauge player value. Because there's way too much talk of "clutchness", "championship caliber", "team leader", and other media-nonsense in the NBA right now. But there needs to be a lot more refinement before any ranking using it can be taken seriously. As of right now, I feel the system is most helpful in determining the values of stars. Because stars are (A)ireplacable pretty much by definition, so the argument of a good backup usually doesn't apply, and (B)they are stars and the system therefore tends to be built around them, so there's less chance of "not fitting the system". Therefore, I tend to put more weight on his ranking of all-star caliber players, separating those who help their teams greatly, and those who only get stats. On the other hand, I think it can't deal with role players. For the same two reasons above. Role players who brings an irreplacable talent to the position(meaning bad backup) would shine in such a system. As those who fits his system very well(like Posey and the fast paced/trapping Memphis team). But bad fits and capable replacements can make a good role player look bad.
The problem with the simple explanation is that it would be a bit misleading. But here's a hypothetical example for evaluating player X. Suppose when player X is on the floor, his team outscore the opponents by 4 points per 100 minutes. When he's off the floor, the opponents ouscore his team by 2 points per 100 minutes. Therefore, it means his appearance on the court accounts for +6 points per 100 minutes. And you do the same for every other player. Then rank them. I'm sure you can see a lot of problems with this, which is where the "adjustments", or as you call them, "@%#$%" comes in to help remedy some problems. But this is the basic premise. By far the greatest positive of the system is that it's not biased against bad teams.
I don't know what you people are arguing about, Trade Yao(50) for Nene(27) or Divac(38) for all I care.
Two things: (1) Role and bench players are tough to evaluate based on this system. A role player is usually playing off the ball and will experience more instances of random variation. You can almost divide out the guys like Ostertag and figure that some of this is simply an anomoly. A poor back-up could affect your numbers positively. Conversely, being a back-up for a good player may negate a player's positives. (2) The stats cannot factor for coaching and team chemistry. The makeup of the 5 gives on the court at a given time is the pervue of a coach, not a player. Therefore, their numbers (in vs. out) may not be reflective at all of their individual value, but rather how the coach determines to shift his lineup (and how effective they are at doing it). For example, I believe Tony Parker would not be in the Below Average category if you could factor in Pop's great coaching ability somehow.
In defense of snowmnt, I think he has a legit criticism of Roesnbaum. Snowmnt doesn't have a problem with the methodology, or the resulting data, he has a problem with what Rosenbaum is saying his data is saying. I have to agree that I really don't understand what the guy is trying to prove with his data.
Ok, I think this ranking shows one thing: how great a couch Sloan is. He knows how to use his players to their maximum strength.
No, he didn't say that impact level was tied to minutes. He just pointed out the correlation he observed. In other words, playing time is not a factor in his formula. BUT the result of his calculation agrees in general (not absolutely) with coaches' decisions on playing some players more than others. He does factor in bench strength, team strength, and opponent strength. Those are exactly the things that separate his method from the simple +/- method which every team uses to some extent. As I said above, he did factor in teammates. But it is true that coaching is not in the formula, as far as I understand it. That is one imperfection in his method. In fact, I believe the biggest flaw in his method is the neglect of a thing called "chemistry" in team sport. He adjusts the +/- scores by strength of teammates, but doesn't (or can't) adjust it in terms of how a player "fits" within a certain type of team and how a player relates to his teammates. You see, there are flaws in the method. But what method doesn't? Even the so-called "human" methods (which basically are people's subjective opinions) are full of flaws. Yet so many posters in this thread are relying on those subjective methods to refute this ratings, saying that "If x is lower than y, than this method must be worthless," as if their subjective methods of evaluating players are never wrong. Even if there are wrong ratings for some players in this methodology, does that mean the whole thing is a sham? If that is true, then everyone of our opinions here is a sham. No? No, this method is not only useful for the MVP award. In fact, I believe that this method is not perfect for evaluating any individual player because there is always the possibility of error due to the deficiency mentioned above. IMO, it is good for understanding what kind of players are more effective than others.
Again, your post shows that you don't understand the method at all. Being double-teamed presumably helps his teammate work better. Shot intimidation helps preventing the opponent from scoring. So does taking charges. You are still in the mindset of conventional stats. He's not measuring how well a player rebounds or shoots or blocks shots etc. He is measuring how much a player impact the result of his TEAM.
Agreed. One of the more intriguing aspects is that first and second year players come out very low across the board -- maybe there is something to be said for veteran experience; another trend that is notable is that multi-category guys (Kirilenko, Tracy, etc) do really well.
I hate to be a "statistics guru" again But his final list has too many outliers. It won't be convincing until he can cover his ass with so many anomalies. Statistics dummies won't bother looking into his methodology. They will discard it in a minute.
I thought you were intentionally trick me and confuse me. Looks like you are not. Again, you present logic fallacies. Because contracts don't come into play. It's all irrelevant. Let's go back and look at the whole thing. This thread is posted in the GARM. Apparently we should concern with rockets players on the list. T-Mac is 4th and Yao is 50th, it's apparent what the reaction would be on a rockets board. However it's all twisted. Why don't you go back and take some of your own medicine and click on the links. There is a final list and an article which should be what we really discuss. Don't know if you didn't read those or forgot it. The "methodology" is outdated and is not what the author is trying to present to the public. There Yao ranks 50th and Amare and Randolf etc ranks 100+. Like I quoted and explained numerous times. The list is about "how a player help their team win". The author made claims that Eddie Curry is the 6th worst player, proved Michael Redd is the 71th contributor in the NBA. You samfisher also was trying to show how people would want Cardinal than players lower on the list. this list IS about how good a player is It doesn't even pass the laugh test. So people said it's useless then somehow other people jump in and say it's "interesting" or whatever and discredit posters. That's why people start to waste time and look into the methodology and whatever. There should be no need to go that far. But there are just people creating drama and twisting simple things. The methodology made another claim that you can have a replace-palyer effect. Apparently it SHOULD not only work for the top five players but can be done roughly with rest of the list. But it doesn't. I won't look into the fomulas but everything the author is presenting for "mere mortals" to see will be "discarded in a minute." @Panda, just realised my last post could be taken the wrong way, I was in a hurry. But you probably wouldn't. I probably share the same reason why you were here, in a thread not about how to improve Badbain and Boki.
That's not saying anything at all IMO, if I use NBA efficiency rating the results will agree with the amount of playing time a player receives, but the results will be very different from the Rosenbaum method. If basketball is thug fight, and my scrubs beat up the opposing gangs scrubs more than I beat up their starters. That only means THAT my scrubs are better than their scrubs more than I am to their starters. It doesn't indicate that my gang will win more without me, nor that I have less impact than my scrubs. To use the bench performance as a gauge to measure the starter performance and the other way around is tautology. It's like saying oh the bench is good because the starters are bad and vice versa. I think you missed my point. I know every method is flawed. My point was the whole impact concept hinges on the specific structure of a team too much that makes it useless except for MVP award ratings, and it's not a very good one at that given the abundance of anomalies. I think you can't use this method and tell me with a straight face that you would rather have AK47 than Kobe Bryant if a straight swap is possible, nor can you tell me that you would rather have Brian Cardinal instead of Amare Stoudamire. This method doesn't measure a lot of impact areas because it's using the wrong gauge, the bench performance to interpret the impact of starters. There is like 100 such no brainers that's ridiculously argued against by this Rosenbaum guy. So what is the use of this method if roster moves will be a disaster if it's followed? The ones that shoots higher percentage, rebounds more, blocks more, assists more?
I knew he's measuring the impact of an individual player, but his method doesn't really reflect all the impacts, as I mentioned above. There is no way to objectively and comprehensively measure the impact of an individual player on a team. If Yao is injured and Cato starts at the C, and the Rockets performance stays the same over a long term, does that mean Cato has the same impact as Yao? No! Because if the Rockets depth gets thinner yet still perform the same, that means Cato is actually a more impact player than Yao, either he's more able or has more chemistry. But how much better can you say? There is no way to know because the team performance is the same! If the Rockets performance get better with Cato, that means Cato is more of an impact player, but how much more exactly? Logically it follows that if the impact of Cato can't be known when the team performance remains the same then Cato's impact can't be known or measured when his team's performance is higher than Yao's team! To use another lineup as a gauge to measure an individual player's impact is therefore LOGICALLY flawed in its NATURE.
I remember in the early 90s when Hakeem was out for 15 games because of eye injury, the team won most of them. By Resenbaum's criteria, Hakeem would have been one of the least impact players then.
