Quote:The PER favors players on winning teams by using a higher points per minute to determine the +/-.
As stated briefly in the original thread, +/- is NOT part of PER. Even if it were, playing on a winning team does not necessarily translate into a higher +/- for a player. In fact, the opposite may be true. Brand, Marbury, and Kirilenko are all notable examples of players on losing teams who have consistently rated among the league leaders in +/-. The argument would be that great players on losing teams have an advantage in the +/- system because the difference between their team with them on the court versus their team with them off the court is much greater than the difference between a winning team with its great player on the court versus the winning team with its great player off the court. Being on a winning team surrounded by other great players thus actually may DEPRESS a great player's +/- statistic. Though, overall, it is true that winning teams will have more players with positive +/- statistics than losing teams, that doesn't mean that the stat is subjective. In fact, it is the chicken and the egg problem. A team doesn't have more players with positive +/- numbers because it is a winning team; a team is a winner because it has more players with positive +/- numbers.
Either way, we all recognize that +/- is not a perfect measure and no one has suggested to the contrary. It does seem to be fairly consistent, however, in its correlation to the consensus rankings of the league's top players. EVERY year that I've paid attention to the +/- stat, the MVP candidates and most recognized top players rate very highly in the +/- rankings -- Duncan, Kidd, Garnett, Nash, Nowitzki, O'Neal, Ginobili, Brand, etc. all have been near the top of the +/- leaderboard in the last 3 years. Sure, there may be some surprising exceptions, but, some of those surprising exceptions may not be so surprising to fans and analysts who truly follow the game.
Quote:Chic I don't remember you ever addressing my concern about the PER rating system. Your choice to move that discussion has me curious as to motive. edit added: but it may be a good idea to move it to another thread and free up the Jazz forum. Thanks for the idea.
Motive? Because I didn't want to spend time addressing an out-of-left field comment that had nothing to do with the current thread means that I have some sort of motive? You have raised this point before, and I have addressed it before. As a mod, I'm sure you have the capability to pull up old threads and find all of the discussions of the +/- statistic to find it.
Quote:When a statistic says that a team does +9 points better when a player is in that's a +. When a team does -9 point better when a player is out that's a -. All +/- stats are based on how well the team does when a player is on the floor or not.
This is true. But please see my response above to understand why your original premise does not mean that there is a systemic bias for players on winning teams. As I noted above, players on losing teams may have an advantage because they theoretically should have more opportunity to affect their team's comparative performance.
Quote:Even the PER Roland and WINVAL [are +/- stats . . . based on how well the team does when a player is on the floor or not].
Let me repeat: PER is not a +/- statistic. It is a statistic that combines a variety of individual statistics and then normalizes the combined individual statistics for league-wide performance compared to a baseline performance year. Unless you can point me to some concrete evidence that would convince me to change my understanding of PER, your mere repetition of the argument that it is a +/- statistic affected by team performance is not going to make me change my mind.
Quote:I don't need to waste my money on a book explaining the next holy grail of basketball statistics,
That statement tells me everything I need to know about your true interest in this topic.
Quote:"a myth blindly chased despite the fact that it does not exist." "No single rating system can ever provide enough subtlety, enough nuance, that it will be all we need. A balanced perspective, one that considers multiple rating systems as well as personal observation, is the only way to go." -- Dean Oliver, the author of Basketball on Paper.
Amen to that. This has been my point each time I refer to various statistics and rating systems to bolster arguments. We SHOULD consider all evidence in support of a position -- available statistics, rating systems, and personal observation. We can't just throw one statistic or system out the window dismissively because we -- who "don't need to waste their money" to learn more about them -- view them as hokum, hogwash, or quackery without even trying to understand them. To dismiss PER or +/- as adding nothing to our understanding of a player's value without even purporting to be interested in understanding or learning more about the statistic or rating system strikes me as deliberately myopic and closed minded.
------------------------ "Hockey is a sport for white men. Basketball is a sport for black men. Golf is a sport for white men dressed like black pimps." ...Tiger Woods
Quote:To dismiss PER or +/- as adding nothing to our understanding of a player's value without even purporting to be interested in understanding or learning more about the statistic or rating system strikes me as deliberately myopic and closed minded.
Chic (and others), just for future reference when posts are edited/deleted... to tell a person that he is myopic or closed minded is a personal attack. Intent is very difficult to judge on a message board and many times it comes down to interpretation by a moderator. Talking "down" to people can also be interpreted as a personal attack. It's best to just leave smack, baiting or name calling out of your post and attack the idea, not the person. Chances are when people react to the baiting everyones posts will be edited one way or another.
Just because I don't buy a persons book does not mean I am dismissing a persons work or ideas. I just look at the whole picture and not one persons opinion (PER) or statistical base.
Here is an excellent article on the evolution of statistical analysis in basketball.
Over there is TENDEX, Doug Heeren's revolutionary first strike for using player statistics in a comprehensive rating form. Beside it lies Manley Credits, the simplification of TENDEX with no complicated pace factor to complicate, now resurrected as NBA.com's " Efficiency Rating System".
Scattered throughout are other linear-weights formulas, with various values for blocks, steals, and whatever else. Squint hard enough and you might even find my own VORP system.
None of these efforts, however, has succeeded in its quest to provide one rating system that's all we need to evaluate players.
Dean Oliver, the author of Basketball on Paper, thinks there's a good reason for that -- the Holy Grail, at least the NBA version, is apocryphal, a myth blindly chased despite the fact that it does not exist.
Oliver has won me over as a convert. No single rating system can ever provide enough subtlety, enough nuance, that it will be all we need. A balanced perspective, one that considers multiple rating systems as well as personal observation, is the only way to go.
But that doesn't mean we can't make better rating systems.
The current trend in NBA analysis is away from so-called linear weights systems, which attach a value to every counted stat, usually determined by theory, and towards those using the performance of a team with a given player on the floor. In its most basic form, this is plus-minus.
There are a lot of things to like about using plus-minus ratings. One of the frequent criticisms of statistical analysts' efforts is that they fail to take into account the unmeasured contributions a player makes on the court, like setting good screens or playing position defense. These are very real things, but they are simply ignored in linear-weights rating systems.
Dallas Mavericks owner Mark Cuban exemplified this thinking in his "Moneyball for the NBA" entry in his blog.
"Yes, there are stats that are out there that could be used to better build an NBA team, but No, they cant be used for building an NBA team, because the stats that most likely most correlate to a player and teams success are not being collected," Cuban wrote.
That's the beauty of plus-minus. If a player sets a great pick, it doesn't affect his individual statistics, but it does affect the performance of his team, and plus-minus -- in theory -- captures this.
Thanks to the exceptional effort of Roland Beech and his 82games.com, NBA fans -- and teams and writers -- have been able to see plus-minus numbers, amongst other things, updated on a regular basis throughout this season -- for free. There is no doubt that this has helped our understanding of the league immensely, and we owe Beech an enormous debt of gratitude.
At the same time, as useful is Beech's Holy Grail attempt, the so-called "Roland Rating" that takes the difference between the team's plus-minus when a team is on the court and when he is on the bench, both taken on a per-48-minute basis, it still falls short in some regards.
Take, for example, Richard Jefferson. Jefferson had the seventh-best Roland Rating in the league this season, but four of his Nets teammates, all starters, also appeared in the top 31. His Roland Rating appears to be capturing not his own individual contributions, but the dramatic gap between the Nets' starters and their bench.
Or what about the curious case of Michael Redd? Redd went from a solid 2.9 Roland Rating last year to -8.8 this season, worst on the Bucks. A similar drop in an individual statistic, like field-goal percentage, would be one of the most stunning declines in recent memory -- and Redd did that as a 24-year-old who made his first All-Star appearance!
There is a way to improve on what the Roland Ratings tell us, and that is to take into account the context each isolated plus-minus rating was earned in -- the player's teammates, the lineup of the opposition, the score, and the location of the game. Essentially, this is at the heart of Jeff Sagarin and Wayne Winston's celebrated WINVAL system, which is used by the Dallas Mavericks and has been used by Indiana, Seattle, and Toronto, according to various reports.
I haven't been kind to WINVAL in this space, and I would be lying if I told you it didn't have something to do with its creators' attitude.
In 2001, for example, Sagarin told the Indiana Daily Student about other rating systems, "nobody has the mathematical ability to take it to the level that we are."
Not only is that arrogant, it also happens to be wrong. Beech said in a Q&A with me in January that two college professors were working on their own versions of WINVAL, using 82games.com's data. I have since found out that one of these professors is Dan T. Rosenbaum, the UNC-Greensboro economist better known to NBA fans for his work with the luxury tax (a role in which his work was recently cited by ESPN.com's Marc Stein). And after a few months to refine his methods, with the help of the APBR_analysis discussion group, Rosenbaum has invited me to help him share his work with the world.
The $64,000 question is, besides for the fact that he's not charging thousands of dollars for his work and has opened his methodology up for scrutiny, what makes Rosenbaum's work any better than WINVAL?
Let's start with the issue of sample size. From what they've released to the public, Winston and Sagarin have calculated their ratings on a year-by-year basis. Unfortunately, 82 games aren't always enough to produce a statistically reliable sample for the data-hungry WINVAL methodology. The issue is that there are not always enough different lineups to provide enough points of comparison for a single player.
The only time Winston and Sagarin have publicly reported standard error for their measures was in this Washington Times article, which states, "Winston says he can predict a player's future rating within four points but only about 60 percent of the time". In general, not reporting standard errors has made WINVAL results appear more certain than they really are.
"Standard errors are essential in evaluating how much faith to put in a rating system like WINVAL that often produces counterintuitive results results that often reflect little more than sampling variation (or luck, for non-statisticians)," Rosenbaum told me via e-mail. "Communicating when their results are reliable and when they are not is critically important, and Winston and Sagarin in their press accounts often seem to go out of their way to obscure this crucial issue."
According to Rosenbaum's 2002-03 calculations, only around 50 players can be found to be statistically differentiable from a replacement-level player, using standard statistical significance methods. Rosenbaum helps make his results less noisy by using data from both the 2002-03 and 2003-04 seasons. Using both seasons reduces the standard errors Rosenbaum reports by almost 50%, a significant difference that allows him to distinguish about 200 players as being above replacement level four times what he could do with just one year and near the right number.
We also know that Rosenbaum adjusts the weight on a particular plus-minus observation (which can be considered the time between substitutions) based on the score of the game, addressing Mark Cuban's point that, "Scoring when you are behind by 20 with 1 minute to go, and cutting it to 15, isn't the same plus-5 as down by three and scoring four with 10 seconds to go," as he told the Seattle Post-Intelligencer last year. WINVAL might make this adjustment; because of its proprietary nature, we don't really know.
Far more important may be Rosenbaum's ability to make sense of what the data is telling him. One of WINVAL's biggest shortcomings, in my opinion, has always been Winston's and Sagarin's general inability to explain unusual results besides a stock answer that defense is underrated by conventional analysis. To wit, this comment:
"This guy, Butler, I have no idea what he does," Winston told the Washington Times, "I'm sure he doesn't have flashy stats. But when he's in, the Wizards play great. What's his first name? Mitchell?"
I don't think I need to tell you that this is not how it should be. Being able to interpret results beyond "good" and "bad" is a critical part of any rating system that truly aspires to Holy Grail status. In order to understand his results, what Rosenbaum has done is regressed them against individual statistics.
That is, he's created a linear-weights system that approximates a player's adjusted plus-minus rating based on the average value of each major statistical category (points, blocks, steals, etc., all per 40 minutes) as well as some other ones that he has found significant, most intriguingly the product of points per 40 minutes times rebounds per 40 minutes times assists per 40 minutes, the strongly positive sign on which indicates that versatile players tend to be more valuable to their teams than their individual statistics would indicate.
"Lots of really smart people have worked on creating player ratings, and as an academic I believe it is important to do my homework and learn from them," said Rosenbaum. "I think these regression results are a place where we can start to answer a lot of questions that have been bandied about for years
Rosenbaum has also used these projected ratings, in comparison with actual ratings, to theorize about the possibility that young players are overrated by their individual statistics -- a fascinating proposition that, if true, would have a significant impact on the way we analyze the value of players on their rookie contracts.
The last important thing Rosenbaum has done with his projected plus-minus ratings is combine them with actual adjusted plus-minus ratings to produce his final rating for each player. In this manner, while going beyond individual statistics, Rosenbaum does not entirely throw them out. The resulting ratings match with the reality we observe as NBA followers much better than WINVAL.
One of the tricky things about any Holy Grail system is how much it should differ from these observations of reality. A system has to be close enough to our expectations to maintain credibility; it must not fail "the laugh test" (once called the "Shaq" test, by yours truly, but O'Neal is no longer dominant enough that any system that doesn't rank him number one must be questioned). At the same time, if all a system did was give us what we expected, why would the math be necessary at all?
I think Rosenbaum has done an excellent job of straddling these two extremes. Of the top ten players by his system:
Rk Player Rtg 1 Kevin Garnett 16.2 2 Tracy McGrady 12.6 3 Andrei Kirilenko 12.3 4 Tim Duncan 12.1 5 Shaquille O'Neal 11.8 6 Kobe Bryant 10.0 7 Dirk Nowitzki 9.9 8 Ray Allen 8.5 9 Baron Davis 7.6 10 Vince Carter 7.6 every one was an All-Star this season, four were on the All-NBA first team, and seven made one of the All-NBA teams.
At the same time, there are surprises. Kirilenko rating as an MVP-caliber player is one of them, I imagine, for most fans. He has always been well-liked by these types of "impact" rating systems -- WINVAL had him as the league's second-best player as a rookie in 2001-02, which made less sense, given Kirilenko was then a part-time starter who barely scored double-figures.
Carter's rating is also a bit surprising and may indicate that his poor reputation league-wide has obscured what he actually does (or does not do) on the court.
In terms of Holy Grail immortality, like everything else, Rosenbaum's system comes up short. But I suspect that the skeleton of this system lies a little closer to that mythical location than any other system ever has gotten.
For more information about Rosenbaum's work and a more detailed explanation of his methods, see "Measuring How NBA Players Help Their Teams Win" and "Picking the Difference Makers for the All-NBA Teams".
Quote: www.82games.com/newuser.htm Using a formula based upon John Hollinger's well crafted PER ratings, we can assess what kind of performance teams are getting at all 5 floor spots. Similarly we can take individual players, say Yao Ming, and calculate the "PER Difference" between their production and that of the counterpart on the other team.
Linear weights - The most common way of evaluating players' overall ability is through the use of what's known as "linear weights" formulas, so named because they assign a weight to each statistic (rebounds, steals, points, etc.) and add or subtract them. The most commonly used linear weights are the WNBA.com Efficiency System and John Hollinger's Player Efficiency Rating (PER). 15 is average for Hollinger's PER, bigger numbers better and smaller worse - except when rating the performance of a player defensively, when you want the opponents to have smaller numbers.
Hollingers system uses parts of linear and plus minus. In order to asses the plus/minus play of individuals against other players (PER Difference) it has to also use a +/- weight in it's formula. It also uses linear measures that are subjective to the personal bias of the person who writes the formula. Both types have their faults.
Great players will always excell in plus minus stats. But two non-star identical players statistic's are affected by their teams play and winning percentage. The player on a winning team will draw his +/- ratings from a larger base of points scored by his team. The opposite being true of the player on the poor team.
We can all have our opinions and discuss things in a mature manner without name calling. I enjoy your (and others) posts here, and respect your opinion. I hope you (and others) will respect mine. A good discussion with lots of different opinions is always a pleasure.
Notty -- I appreciate your views and your generally thoughtful approach and responses on various topics. You are correct to observe that my suggestion that those who simply throw statistics out the window dismissively in their discussions of a player's value strike me as myopic or nearsighted was inappropriate. Not to justify this comment, but to explain where it comes from, however, the history of statistical discussions on this board typically involves me citing a statistic to support a stated opinion only to have a host of other board members criticize me for using a statistic in the first place and suggest that personal observation is the only valid way to evaluate a player (okay, that's a slight exaggeration, but only a slight one). I have been on this board for approximately 18-20 months, and, nearly monthly -- if not more -- have had to defend a statistic that I've cited from attack by posters who belittle my opinion by telling me I rely to much on statistics and then criticize the statistic merely because the statistic disagrees with the poster's views even if the poster shows nothing but disdain for, and no interest in understanding, the statistic.
I responded as I did to you because your turn of phrase that you "don't need to waste [your] money on a book explaining the next holy grail of basketball statistics" again seemed dismissive and derisive of the very statistic we were discussing.
As for the articles you cite, during the past couple of years, I have read all of them and find most to be interesting and illuminating.
Finally, addressing your conclusion that PER uses part +/-, I have never seen that in any article nor does it appear that any component of +/- appears in the actual formula. I still don't understand the basis for your conclusion. I don't see it in any of the articles you cite. Yes, Dan Rosenbaum's new formula relies on the +/- statistic, but I have seen nothing to suggest that John Hollinger's PER statistic is colored in any way by it or includes any component of the +/- statistic. Please point me to a specific citation that says something to the contrary, because I would be very interested to know more about this topic.
------------------------ "Hockey is a sport for white men. Basketball is a sport for black men. Golf is a sport for white men dressed like black pimps." ...Tiger Woods
Chic you are not the only one to defend statistics, there are others. Peter Glans use to write stat articles here using his own formula for +/- . Combined with personal observations and news articles statistics pretty much are the source of many peoples opinions on players. Is it possible that these people tried to get you into a discussion, and you felt they were being confrontational? I really don't know because I can't spreak for others on the board. You could ask them.
I have been told by many including you I believe, that John Hollinger's PER stats were not just linear stats like they use in fantasy leagues. If I am correct (again going by what I have read from you and others) no one know his formula, but he has added weights and such to take into account various aspects that other linear stats have not, and this is what makes his different. If no one knows his formula, how could they write articles about it?
Using this quote from above...
"Similarly we can take individual players, say Yao Ming, and calculate the "PER Difference" between their production and that of the counterpart on the other team."
It is my opinion that he could not do this with only linear stats and not be like all the other linear stats formulas that have existed over the years. The only way to calculate how an individual player will do against another, is to include a +/- weight in his formula. I can't prove this, and you can't disprove it. But it makes sense to me, otherwise his "new" statistical formula is just a linear formula with more bells and whistles.
Do you feel that his formula is strictly linear (such as fantasy leagues), and therefore he gives the weights and numbers to what he theorizes they should be?
Quote:Chic you are not the only one to defend statistics, there are others.
Looking back at it, I see that what I wrote wasn't entirely clear. I didn't mean to suggest that I am the lone defender of statistics on this board, What I meant by "the history of statistical discussions on this board" are those discussions that I have had with statistical detractors. I understand that there are and have been other discussions about statistics outside my limited universe.
Quote:Combined with personal observations and news articles statistics pretty much are the source of many peoples opinions on players.
I would think so, but, in response to almost any post that reports some non-traditional statistic -- such as PER or +/- -- I almost always get some variation of the dismissive "there are lies, damn lies, and statistics" line fed back to me. Many posters have responded with more direct derision -- not necessarily personal attacks, but derision for the sited statistic:
"With all due respect, you always rely way too much on statistics."
"as said, I don't pay any attention to that PER stat so I skipped the last part of your post"
"I get the feeling that you don't even watch the games and just look at statistics."
"I would rather believe Sloan than some stupid made up statistic."
Etc.
See the thread that started this discussion as a prime example of a typical exchange: ME: Here is an interesting statistic. ME: Explanation as to why the statistical rankings might be different than Fox's rankings. OTHER POSTER: "How many times must we go over this? [Your statement] is just a complete misconception." ME: I think you misunderstood my statement, and here is a more detailed explanation of why. OTHER POSTER: I understand your statement, but still disagree. ME: But your restatement of my statement micharacterizes my original post to undercut the statistic. OTHER POSTER: I don't believe you meant what you said. ME: Blah, blah, blah, blah, blah.
Quote:Is it possible that these people tried to get you into a discussion, and you felt they were being confrontational? I really don't know because I can't spreak for others on the board. You could ask them.
I think that nearly all of us come across as confrontational at times on this board -- intentionally or not. And, I think that I have probably been involved in more "confrontational" exchanges than the average poster on this board for 2 reasons. First, I sometimes voice strong opinions about unpopular subjects that may elicit a strong response. Second, I have often responded in kind when the tone of a response appears to be personal or condescending. I have tried to be more reasoned in my responses and remove any trace of emotion from the prose, but I agree that some of my posts could be viewed as confrontational -- sometimes intentionally, sometimes not.
Quote:I have been told by many including you I believe, that John Hollinger's PER stats were not just linear stats like they use in fantasy leagues. If I am correct (again going by what I have read from you and others) no one know his formula, but he has added weights and such to take into account various aspects that other linear stats have not, and this is what makes his different. If no one knows his formula, how could they write articles about it?
I don't believe that I have said that nobody knows his formula. In fact, I posted what I believe is his formula in the very thread to which you were responding:
"Similarly we can take individual players, say Yao Ming, and calculate the "PER Difference" between their production and that of the counterpart on the other team."
"PER Difference" is a different statistic than PER. The way this is calculated is to take each player's PER -- which is calculated as set forth in the formula above, and then simply compare the player's PER to the aggregate PER of his counterparts from his opposing teams over the course of the season. But, again, this is a different statistic than PER (though I think PER Difference actually may tell you more about the player than PER -- since it incorporates a more direct defensive element).
Quote:It is my opinion that he could not do this with only linear stats and not be like all the other linear stats formulas that have existed over the years. The only way to calculate how an individual player will do against another, is to include a +/- weight in his formula. I can't prove this, and you can't disprove it. But it makes sense to me, otherwise his "new" statistical formula is just a linear formula with more bells and whistles.
Do you feel that his formula is strictly linear (such as fantasy leagues), and therefore he gives the weights and numbers to what he theorizes they should be?
As I have posted before to explain PER to others, PER is a per-minute rating developed by John Hollinger, author of one of the most highly regarded basketball analytical publications around, Pro Basketball Forecast. In Mr. Hollinger's own words:
Quote:The formula, which I call the Player Efficiency Rating (PER), adds the good (made shots, steals, assists, rebounds, blocked shots, free throws), and subtracts the bad (missed shots, turnovers, fouls) by assigning a point value to each item (I arrive at the point values in a fairly tortuous way, and that's one of the parts I'm saving for the book). The rating for each player is then adjusted to a per-minute basis (so that, for example, you can compare subs with starters in the frequent 'he should start ahead of so-and-so' debates), and also adjusted for the team's pace. In the end, one number sums up the players' accomplishments (the statistical ones, anyway) for that season. I've set it up so that the league average, every season, is 15.00, which produces sort of a handy reference guide:
A Year For the Ages: 35.0 Runaway MVP Candidate: 30.0 Strong MVP Candidate: 27.5 Weak MVP Candidate: 25.0 Bona fide All-Star: 22.5 Borderline All-Star: 20.0 Solid 2nd option: 18.0 3rd Banana: 16.5 Pretty good player: 15.0 In the rotation: 13.0 Scrounging for minutes: 11.0 Definitely renting: 9.0 On next plane to Yakima: 5.0
It's superior to other ratings I've seen . . . for a number of reasons. First of all, most ratings tend to overweigh the non-scoring categories, which is great if you're in an 8-category fantasy league but produces results that don't jive with reality. The "IBM Award" is the most notable of these; the fact that Michael Jordan didn't win it during any of his five MVP years says a lot (David Robinson, on the other hand, won this award five times. I propose renaming it the 'IBM Roto MVP Award').
There are a number of variations on the "Tendex" system, which basically adds points for good stuff and subtracts for bad stuff, just like the ratings here do. The problem is the weights they put on categories are screwy (most add exactly one point for all the 'good' and subtract the same for all the 'bad'), and so tend to distort the importance of some categories. For instance, I've seen versions from last year that didn't even have Allen Iverson in the top 20 because they overrated missed shots. It's worse if you go back in history; use a Tendex-type method for the early '90's and you'll consistently see David Robinson rated higher than Michael Jordan, which borders on blasphemy."
Anyway, based on the publications that I have read, the PER formula is:
Some of the less obvious components of the formula -- factor, VOP, and DRBP -- are adjustments for overall league performance during a given player's season or era and are defined as follows:
------------------------ "Hockey is a sport for white men. Basketball is a sport for black men. Golf is a sport for white men dressed like black pimps." ...Tiger Woods
Quote:I would think so, but, in response to almost any post that reports some non-traditional statistic -- such as PER or +/- -- I almost always get some variation of the dismissive "there are lies, damn lies, and statistics" line fed back to me. Many posters have responded with more direct derision -- not necessarily personal attacks, but derision for the sited statistic:
"With all due respect, you always rely way too much on statistics."
"as said, I don't pay any attention to that PER stat so I skipped the last part of your post"
"I get the feeling that you don't even watch the games and just look at statistics."
"I would rather believe Sloan than some stupid made up statistic."
On a message board people are going to disagree with you. But we need to make an effort to be less confrontational and accept the fact that others have a right to their opinions also.
From a mods point of view, once someone steps over the line, most others in the thread starts stepping over and we have a lot of posts that need to be edited. Most other sites just delete the multiple over the top posts in entirety, and the mods here are going to start doing that much more often in frequent exchanges. We all need to be more civil and mature by accepting others opinions without name calling. Doing the multiple deletions also deter those that try to get edited as a red badge of courage.
The formula means nothing without the values assigned. I can scratch a chicken better than that. -j/k
So his "new" statistical formula is just an old linear formula (fantasy league type) with more bells and whistles? I can live without paying the price of the book to find out his fairly tortuous way he arrives at his formula. In time we will learn it anyway, and by that time someone will come up with a new formula and release his book or start his own website.
I admit sometimes I get confused as to your use of the PER stats (linear) and other times your use of 82 games stats (+/-). Which is right?
I don't know where this discussion is anymore. If everybody agrees that PER is a subjective formula, but a useful tool, is applied objectively, though due to the nature of the formula it MIGHT favor certain types players over others, and isn't inherently more valuable than any other measure (though I do like it better than most), than I'm satsified. Someday maybe I'll actually run some numbers through the formula and see how much I agree with it. But that day is not today.
------------------------------ "It has basketball monkey and it wants to return to play."
Pro Basketball The NBA Tries to Make Teamwork a Science Coaches are crunching numbers and consulting computers to find winning lineups
By RUSSELL ADAMS Staff Reporter of THE WALL STREET JOURNAL November 5, 2005; Page P6
Players in the National Basketball Association will find a new category in their report cards this fall: "Plays well with others."
In a league long dominated by high-flying superstars, more teams are focusing this season on teamwork -- and turning to surprisingly scientific methods to measure it. New technology makes it easier to track the performance of every combination of five players that steps on the court, in a long list of game situations, from out-of-bounds plays to pick-and-rolls to zone defenses. As different player mixes yield different results, teams are beginning to quantify the elusive concept known as chemistry.
Say, for example, that after a coach inserts two particular players into a game, the opposing team has trouble scoring. Getting ready for the next opponent, the coach might flip open his laptop, punch a few keys, and see how his team did defensively in other games when the same two players were on the court together. He's able to do this because teams are increasingly turning to software that dissects plays, follows every pass and shot and tracks each player's part in every possession.
For basketball, it's something of a catch-up game. While combining video and statistical analysis has long been used to gain an edge in baseball and football, it's a fairly new phenomenon on the hardwood court. NBA teams have been slower to adopt that approach to dissecting games because it requires someone actually record every possession of every game, including which players were on the court. Los Angeles Clippers coach Mike Dunleavy says that when he took the job in 2003, the team's entire video-scouting system consisted of two VCRs.
Now, the NBA is realizing how well the approach can actually work. Software developed by XOS Technologies helps a coach see things like how the team fared in quarters when certain players were on the court together. He could run any number of different scenarios: To determine whether one player helped his fellow defenders more than another, he could isolate moments in games when the two played separately but with the same four teammates. "Coaching went from being very subjective to an exact science of what guys do," says Mr. Dunleavy.
Of course, the raw material for this kind of study has been around for some time. The NBA started keeping such records in 1996, and over the years more teams have been developing or purchasing software programs to help them extract patterns from the data. But now the data are more widely available thanks to independent Web sites like 82games.com and academics who record them for teams or for publication. Fans were the first to catch on to 82games, but the site's founder, Roland Beech, says more NBA teams are coming to him now with questions about his data.
In addition, all but two NBA teams now use XOS's system that packages statistics and video for instant evaluation of what is working and what isn't. And XOS is testing a wireless system that allows NBA coaches to show plays on a laptop during timeouts.
Over the course of an NBA season, the average team uses about 500 different five-man lineups, according to statistician Wayne Winston. The new push toward more number-crunching analysis could have profound implications for how games are played.
"The ability to do that with numbers is huge," says Dean Oliver, a statistical consultant for the Seattle Supersonics. "The NBA and a lot of sports have always been about who are the best players. But basketball is such an intricate game it really is about how you fit them all together."
Though fans have grown used to Michael Jordan, Kobe Bryant or Tracy McGrady pouring in basket after basket while his pals mostly stand around and watch, there is growing evidence that savvier teamwork can take a team all the way to the top. Last year's finalists, the Detroit Pistons and San Antonio Spurs, played a team-oriented game, with only one player between them averaging more than 20 points a game last season. And don't think the rest of the NBA hasn't noticed the success that comes from the team approach.
Indeed, as coaches track every dribble of every game, they're discovering some surprising things about their own rosters. Among the revelations: Eddie Griffin gave the Minnesota Timberwolves a 10-point boost when he came off the bench to join superstar Kevin Garnett on the court last year, according to 82games.com. And the Houston Rockets' most effective pairing in net points (see chart nearby) was all-star Tracy McGrady and journeyman Jon Barry, who together made Houston 13 points better than opponents.
Of course, the idea that a good player isn't the same thing as a good teammate is probably as old as the NBA itself. And duos that just seem to click on the court aren't a new phenomenon; you might remember some fellows named Stockton and Malone or Cousy and Russell.
But technology is taking the guesswork out of finding player combinations that work -- or don't. Shot-location data, for example, have often confirmed teams' hunches that pairing guys who like to shoot from the same part of the court can inhibit both players' scoring.
And teams that have bad chemistry often see it blow up in their faces. Exhibit A: The 2003-04 Los Angeles Lakers, who added future Hall of Famers Gary Payton and Karl Malone to an already stacked roster -- and got worse. That team was hurt by injuries and the distraction of Mr. Bryant's legal troubles, but Mr. Payton's inability to accept a reduced role and play a team game was widely seen as a big part of the team's ultimate undoing.
The Miami Heat took a major roll of the dice when it this off-season jettisoned Eddie Jones and Damon Jones, perhaps the two best teammates stars Shaquille O'Neal and Dwayne Wade could ask for, and replaced them with two highly skilled players -- Mr. Payton and Antoine Walker -- who have had trouble fitting in on equally talented teams. The Heat didn't return phone calls seeking comment.
Science suggests the pairing of dominant scorers like the Philadelphia 76ers' Allen Iverson and Chris Webber is likely to run into problems. Statisticians Jeff Sagarin and Mr. Winston created a calculator that rates the performance of the many different lineups each team uses during a season and found that one of the lineups the 76ers used most after Mr. Webber was acquired last year was more than 21 points per game worse than an average NBA lineup against similar competition.
[The NBA's Dynamic Duos] - source: 82games.com These teammates significantly boosted their team's fortunes when they were together on the court last season. Net points per 48 minutes is a measure of how many more points than their opponents the team scored per game while using that pairing
Quote:In addition, all but two NBA teams now use XOS's system that packages statistics and video for instant evaluation of what is working and what isn't.
Quote:How many think the Jazz is one of the two teams?
I've got my hand raised. Talk of statistics around the Jazz front office and coaching staff was probably met with grumbling, talk of "hokum," and someone recalling "back when I was playing . . . ." -- much the same way a few posters on this board react to any discussion of statistics.
