Put it this way, If I say you are totally wrong, and you ask me why, I say you go read the article and the thread. Don't you feel I owe you an explanation?
Yeah; same problem as before; you are not able to discuss this intelligently or rationally; you didn't bother to read the article; you didn't bother to read my posts. Don't know why I tried again. See ya.
I read all your posts, just failed to see good logic there. In fact I believe you didn't read my post, so you did't know I said I already read the article.
I let it go yesterday, and didn't respond to Sam's name calling. But he keeps insulting me without valid argument today, that's when it started to get personal. Actually I knew what he was trying to say, but now he realized it's not valid, all he left is insulting.
Choujie: That's because they only have a mess that's not cohesive nor clearly illustrated at all to fall back on, Rosenbaum lives in a dream world in which he thinks that he can hold all other relevant variables constant by manipulating some numbers and arbitrarily weighted ratios. It's also due to that certain posters here developed a nasty habit of attacking position rather than points. It's the real world and it's normal to meet guys like that so just laugh it off.
From the article: "Jeff Sagarin has successfully applied this methodology to teams for years and to individuals in non-team sports, such as golf and tennis, but this is the first time that I know of that this methodology has been applied to a individuals in a team sport."
Picking the Difference-Makers for the All-NBA Teams By Dan T. Rosenbaum Being a Sportscenter star or a fantasy league stud does not make a player All-NBA. Difference-makers ¨C the guys who make their teams better when they are out on the floor ¨C that is who belongs on the All-NBA teams.
Rule to judge a player rating system: 1. see where Yao is 2. see where T-Mac is 3. see if the name of the author is cool enough 4. judgement time
LOL. But seriously, if a system says Ostertag at this moment is better than Yao or Yao at this moement is better than Hakeem ever was impactwise, every basketball fan is entitled to judge it immediately no matter how good the author claims it to be. Of course for people who never watched basketball, that's a different story.
sam, after saw your post in D&D get locked moments ago I came here and here you are again playing your old little trick of name calling on choujie just because he happens to disagree with you yao-bashing to satisfy your little ego that probablly kept get crushed in your real life you might be a contributing member but i still go and get a life you are a tumor of this board I do not care to get locked or booted for this post if that is it be it it's great to let my opinion out
You are not alone forebay. No poster deserves childishly abusive treatment like this. I think I'm witnessing a real freak in life on this board.
Still a MENSA meeting going on in SF's mind. The problem with any "impact" stat is that you cannot measure players who are not on the ball (OsterFAT) vs. players who play on the ball - "give me the rock" - (Yao, Steve or Amare Stoudemire). Also, rebounding does not have the impact it has for a team that wants to be great. It's obvious to me why. Lets see if SF understands why.
LOL. I'm feeling incredibly patient today. So let me try the last time. (choujie, you can ignore SamFisher's antics. When he gets frustrated by people, he tends to act like that. ) This is for those who are too lazy to actually try to understand the rating system. I am doing a great service for you to summarize the basic premises here. I charge $100 per hour for my consultation fee. You can pay that to the tipjar. 1. The system does NOT calculate any of the conventional stats such as scoring, rebounding, assists, etc. So it does not favor players who have the ball more than his teammates. Most statistical systems favors players who do more things with the ball. Sometimes players do things with the ball that hurt, rather than help, the team. Yet they are rewarded in most conventional statistical systems. Hence, "selfish" players are usually rewarded by these stats. 2. The system measure basically one thing: the point differential of the team when a player is on the court versus when he is on the bench. This is usually called the plus-minus stats. Let me repeat in case somebody misses what I said in the previous point: THE PLUS-MINUS MEASUREMENT HAS NOTHING TO DO WITH ANY OF THE CONVENTIONAL STATISTICAL NUMBERS OF THE PLAYER. 3. The +/- method is used by most NBA coaching staffs to some extent to aid them in guaging which players are more effective on the floor. All coaches know that conventional stats, while helpful, don't give the whole picture in terms of how a player is impacting the outcome of the game. So this concept is NOT new and is well-received by NBA coaches. Anyone who disses this basic approach is dissing a well-used method by professional basketball people. 4. The +/- method is good in that it tacitly accounts for things that conventional stats don't cover. These thing can be as basic as pick setting, boxing out, spacing, movement without the ball, passing to the right teammate at the right time etc. They also include more intangible things such as defensive presence, intimidation, on-court leadership, etc. 5. The basic assumption of this method is that if you do the good things, whether accounted for by conventional stats or not, your team's performance will increase. Conversely, if you do the wrong things, the team's performance suffers. And when you sit, the team loses those good things or bad things that come with your presence on the court, and the team performance benefits or suffers. In other words, it tries to account for how well a player AFFECTS the team, not how he well he PLAYS because how well one "plays" is very ill-defined by conventional stats. 6. The simple +/- system is useful for coaching decision within a single team, or more accurately, within a single position. The coach can compare how a starter and a bench player each impacts the team's performance and decide how much he wants to play each player. However, it is not good to compare players across different teams, or even across different positions. The simple reason is that it only tells you the relative effectiveness of a player compared to his backup. And it also doesn't account for the quality of people playing alongside you and against you on the court. 7. This is how Rosenbaum's method comes in. He tries to adjust the +/- measurement of a player by factoring a his teammates' impact, and his opponents' impact. It is very similar to the "schedule strength factor" when people do power ratings for teams, except that doing it for individual players is much more complex and tedious. Basically, you have to factor in all the other 11 players when a player is playing. You also have to factor in the players that are NOT playing, and so forth. 8. Now you can see how complicated this can go. This is not something that an ordinary basketball guy can do. You have to have the knowledge of mathematical models together with the help of powerful database computing to get it done. So those who diss the method as something done by "book people who don't know basketball" do not understand the complexity of it. I don't pretend to understand all the technical details because I don't have the professional training. But all you need to know is the basic principle behind all these formulae. 9. I have pointed out that while this method is very compelling, there are still some things that it cannot account for. These include coaching, fitness of a player within a certain style of team, a thing generally called "chemistry" (sound scientific, doesn't it ), performance variation, fatique, game schedule (back to back road games are very different from 2 days rest between home games). It is flawed. But it is better than most other conventional statistical systems, imo. 10. Why are there aberrations such as the now infamous Ostertag rating? I don't know enough details of the calculation to answer. There are some possibilities. First, it might be that some players are severely underrated/overrated by conventional evaluation methods (including both statistical and subjective "human" methods). e.g. Could it be that Brian Cardinal is really that good? (Hmm, maybe Jerry West has been using this thing for years and they all thought he's a genius? ) Could it be that AK47 is really a more valuable player than Kobe? Kobe is the typical beneficiary of the "do more and get more" type of conventional rating methods. Second, it might be that some players' ratings are affected by the inherent deficiencies I pointed out above. Third, it might be that some players are just lucky/unlucky that catches the performance variation of his teammates and opponent just right/wrong. OK, I have spelled it out for you. Now you have no excuses and no pretense. Fire away. . . or go away.
that's lot of input. but the key here is that nobody ( i think) is dissing the general approach, what some people, including me, have issue with is the particular application by this professor of the general math model to NBA players. and even this type of general math/statistic/calculous model in general is so much flawed: it's most widely used for marketting researchs, and if it really work, i can immediatelly become a billinair by applying it to stock market. heck. i once even tried to help an economic professor to code his model for studying deparment store promotion activities' impact on its sales, and to tell the truth, it did not work well. do not through too much respect to those guys before a math approach succeeds in predicting market up and downs as acurate as we predict weather changes.
Yes, there are at least a few people dissing it without even understanding the general approach. I have no problem with people who know math questioning his formula. But a lot of criticism stems from seeing some aberration (particularly Yao) in the ranking. People say things like, "Any system that ranks Ostertag above Yao must be worthless." And after repeated explanations, they still say the same thing. That's what is frustrating discussing issues with these people.
When I saw AK4 is rated at 2. I know the system is not working for NBA. Stop talk the sytem, NBA is not math.