Be careful what you wish for...

[Collinto]

Greendreamer, I know good defense when I see it, because when I was in high school and college (played at a two year school at a low level, not really competitive and more like high level intramurals, but it was organized with practice and we worked at it). We watched game film there. Have you ever played hoops before? Or just a rec league?

This post belongs in the RealGM HOF. Good stuff.

[jfs1000d]

Greendreamer, I know good defense when I see it, because when I was in high school and college (played at a two year school at a low level, not really competitive and more like high level intramurals, but it was organized with practice and we worked at it). We watched game film there. Have you ever played hoops before? Or just a rec league?

This post belongs in the RealGM HOF. Good stuff.

Yeah, it's kind of corny, I know. LOL.

[tlee324]

I am kidding - mostly. Though Rondo's ability to fight through a pick sucks. he has always had that problem. It's his slight build. TA defends PG's far better then Rondo.

I think maybe TA fights through picks better than Rondo, but overall defending PGs, I'd go with Rondo over TA.

[GreenDreamer]

Greendreamer, I know good defense when I see it, because when I was in high school and college (played at a two year school at a low level, not really competitive and more like high level intramurals, but it was organized with practice and we worked at it). We watched game film there. Have you ever played hoops before? Or just a rec league?

Rondo gambles, which is very poor defense, and he loses sight of his man when that guy is off the ball. He should be a major pressure defender, but sometimes he picks up the dribbler too deep. When Rondo wasn't in the game, several other starters were out. Besides, Celtics had no bench in the playoffs. Everyone they put in stunk up the joint. Rondo is certainly better than anyone on their bench. Doesn't mean he played good defense.

Kid has the goods to be a Gary Payton type defender. Love his frame and quickness as a defender. He isn't that yet. He plays much too risky on defense and goes for steals.

Rondo played the pick and roll poorly and didn't communicate with his teammates.

Are we supposed to believe the stats and reject our immediate observatioins?

Getting blown by was an epidemic for the Celtics in the playoffs as a whole. Turkgulu killed the Celtics off the dribble. And on Rafer's poor stats...he shot poorly early in the series. He was WIDE open a lot of the time. How did Rafer do in games 6 and 7? better shooting. And, Anthony Johnson killed us in Game 3.

Here is the crux of my problems with this: I don't get what Wages of Wins is trying to measure. How can you prove or disprove it? As guyclinch said, there is no validity.

What's the formula, and what makes a good basketball team. The dismissal of the scorer is the worst part of the formula. It takes points to win games. There is a reason big-time scorers are always highly sought after.

These ultra efficient non-scorers need to play alongside the prolific scorer. That's why we have to be careful with Rondo. When Pierce and Allen go, and eventually KG, someone is going to need to score. If it has to be Rondo we are in big trouble.

Does WoW consider that the efficiency of Rondo has a correlation to the fact that the Celtics have two elite scorers on the wing and a good scoring option at PF? Ray Allen and PP were great before they played with Rondo. Rondo isn't the guy making them efficient.

They make Rondo efficient and hide his flaws. Rondo is never out there without one of the Big 3. If he was, the Celtics offense would struggle immensely.

The Gary Payton comment tells me a lot regarding how out of touch you are with the state of perimeter defense in teh NBA. I find this depressingly typical of many posters. I actually watched Payton talk and demonstrate how he would defend guys on NBA not that long ago, using OJ Mayo, I think, as his test dummy. Time and again he would finsih a sequence saying "but you can't do that anymore". The reason being that the rules are fundamentally different now. You cannot bump and grab on the perimeter in teh NBA. If you do not beat the man to the spot, and you hit him it is a foul. End of story.

That means fighting through picks, in teh attempt to stay in front of a guy, is a very dicey propositionn, especially if the pick setter is a guy like Dwight Howard who is going to hip check you as you are going by and not get called for it. Hell, he actually flattened Rondo with a blatant lowered shoulder and didn't pick up the call, the refs instead electing to call one on Veal. That was just insane. They weren't so reserved in calling fouls on him, though, if he bumped his man, which is basically the norm in the league. Seriously, when was the last time you saw somebody fight over a pick and stop Rondo? It isn't like they never go over the top, the's just gone if they do. I saw Alston do it once. Its like Doc said in the regualr season when asked about hypothetically playing Rondo He said that with the old rules he'd be able to take him, but with the new ones that he wouldn't have a chance. This is NOT high school. This is NOT ccollege. It is the NBA, and the rules are DIFFERENT. With slower point guards it is doable, but with the quick ones, forget about it.

The whole gambling thing is overplayed, IMO. Most of his poke from behind moves come on teh pick and roll. He comes over the top, and if that guy shows the ball, Rondo takes a poke at it. He's following the guy anyways. Ian Thomsen from SI polled some scouts for end of season awards picks, and they picked Rondo for first team defense, specifically PRAISING his pick and roll defense. It is very HARD to be as good as he is within the constaraints of teh rules.

Alston scored 26 points on 10 for 25 shooting over teh last two games. He did hit 5 of 11 threes, mostly on kickouts. Better, but not by a whole lot.

[jfs1000d]

On the changing rules, I have no doubt Gary Payton would adapt and find a way to become a great defender. Players adapt to the game they are playing. Good defenders will be good defenders under any conditions. You adapt.

This argument is getting circular. Mind as well end.

It's interesting that we see two totally different players out there. I see a good, but flawed player, whose weaknesses can be exploited, but is stilleffective because largely those flaws are shielded by the talent of the Big 3.

You see a unique player of rare productivity and efficiency that creates a huge positive for the team and will soon eclipse the Big 3 as a star. Rondo it would seem to you is on the same level as an in prime Jason Kidd or Magic Johnson,or at least heading there.

I guess you feel Rondo can stand alone, and I think he is very much a product of the Big 3.

[Slartibartfast]

Thanks to Drza for a pretty even-handed analysis. You met Berri halfway at a less ambitious conclusion.

Pete, you are placing a lot of value in your intuition. Your intuitive judgment of a player's pretty goodness, which you substantiate by comparing it to the general consensus, is hardly the stuff of a compelling argument. Your eyes and intuition are as likely (I believe more likely) to generate false projections and evaluations. And at least stat-driven analysis opens up discussions that are forbidden according to the dictates of media-driven convention.

There's a lot of apple-to-oranges argumentation going on in here that I think Drza sliced through rather nicely through his rejection of the "player a is better than player b" framework, and instead focusing on specific areas.

You construct the argument that: a. Kobe is better than Rondo, as evidenced by elementary understanding of basketball and conventional wisdom, b. WoW lists Rondo as better than Kobe, and c. therefore WoW is obviously wrong and thus useless.

As dzra pointed out, "better" is impossibly vague. Maybe you disagree with WoW's take on "better," but at least adequately define your judgment of better.

I like substantive input from the opposition. The rebuttals in this thread, with the exception of drza and a few solitary points from Pete, have made this mostly an exercise in identifying logical fallacies. Appeals to authority, ad hominems (Pete seems to be hellbent on proving that GreenDreamer is a Rondo cultist instead of just answering his points), tautologies and many more.

[jfs1000d]

It's not intuition it's observation. Why would you collect numbers and interpret those numbers without viewing how they were created with the same critical eye?

What you need is some field work. Empirical evidence is by no means superior to observation. In some instances, one works best,in other instances, another works best.

This formula is an advanced synthesis of a box score that doesn't take into account the manner the stat was created. If rondo misses a layup,it's treated the same as missing a 15 foot contested jumper.Am I correct in this?

That to me is a pretty serious limitation. Observation is much more important in this instance,especially the NBA where the players are so codependent on others and one player can make such an extreme difference in performance.

Baseball? Give me advanced stats. Hoops? I got to see the guy play. It's a more accurate evaluation tool.

The more complex the interaction of players (such as football) the harder it comes to quantify outside of the traditional compilations.

[GreenDreamer]

It's not intuition it's observation. Why would you collect numbers and interpret those numbers without viewing how they were created with the same critical eye?

What you need is some field work. Empirical evidence is by no means superior to observation. In some instances, one works best,in other instances, another works best.

This formula is an advanced synthesis of a box score that doesn't take into account the manner the stat was created. If rondo misses a layup,it's treated the same as missing a 15 foot contested jumper.Am I correct in this?

That to me is a pretty serious limitation. Observation is much more important in this instance,especially the NBA where the players are so codependent on others and one player can make such an extreme difference in performance.

Baseball? Give me advanced stats. Hoops? I got to see the guy play. It's a more accurate evaluation tool.

The more complex the interaction of players (such as football) the harder it comes to quantify outside of the traditional compilations.

Your opinion is a more accurate observation tool? I could be wrong here, but you were probably in the "Rondo was stealing rebounds from his teammates camp". Well that turned out to be a complete load of crap. You don't average almost 10 rebounds game stealing them from your teammates, and rebounding is important. The whole opinion thing gets blown to bits by the hard data.

It isn't as if blindly follow these things. I certainly don't think that Scalabrine is the worst Celtic of the last 30 years, but I also understand that his usage dictates his effectiveness. Put him outthere with a Kidd or Rondo, and he can be very useful. Have him play without guys who can set him up, and his good defense can't offfset his offensive liabities. That was even true this season Just check out his player pairs

http://www.82games.com/0809/0809BOSP.HTM

The difference is quite stark. What your opinion has yet to answer is why Rondo's stats are the way that I are. This entire thread is full of "I think this way because I think this way" statements, yet no real explanation as to why the statistical data would read so much to contrary. Not just WoW stats, but plus/minus, offensive and defensive ratings, opponent's performance, oncourt/offcourt data, NBA efficiency , etc. The team perorms much better with him on teh floor and much worse with him off of it. Why is that?

There are all kindds of "opinions" out there which get droned over again and again. One popular one is the "He plays with three great players" one. People continued to say that through and after the playoffs. That one of those guys hadn't stepped on the court in weeks seemed to make no difference. Garnett being on the roster was enough. "his defense is overrated", yet his guy significantly underperforms and the etam's defense is much better with him in than out... yet this "opinion" keeps getting repeated.

Here are some things to think about, which may be escaping your observations

1. What happens when Rondo is not "blown by"? How difficult is it for his guy to do anything constructive?

2. How effective is he at doubling down and getting back to his man?

3. How often is he attacked on mismatches?

4. How much area does he cover in our zone defense?

5. How effective is his ball pressure in stalling an offense?

6. How often does he cash in on his gambles.?

7. How much help is he given on the offensive end? How much do the other Celtics cheat towards him?

8. How important a component is defensive rebounding to playing effective defense? The Celtics grabbed the second highest percentage of defensive rebounds of the 16 playoff teams, and Rondo was second on the team in defensive rebounds and defensive rebound percentage. Did you know that Rondo's DRB% was 21.9 in these playoffs,and Pau Gasol's was 22.5%? The Lakers actually grabbed 4% fewer defensive rebounds? That means that Rondo was clearing the glass at the same rate as an All-NBA F/C who is 7 feet tall, in an effort which actually made his team a better rebounding team. Who else was going to do that?

9. How effective is he at ball denial?

These are things to keep in mind.

[wigglestrue]

"Which composite stat has the highest correlation to actually winning games?"

Which one correlates better, PER or WoW?

[GuyClinch]

Pete, you are placing a lot of value in your intuition. Your intuitive judgment of a player's pretty goodness, which you substantiate by comparing it to the general consensus, is hardly the stuff of a compelling argument. Your eyes and intuition are as likely (I believe more likely) to generate false projections and evaluations. And at least stat-driven analysis opens up discussions that are forbidden according to the dictates of media-driven convention

First off I wasn't talking about MY judgement per se - but the collective judgement of the NBA (gms scouts fans) as being superior to the stat driven "judgement" of the wages of win.

Again the wages of win HAS NO ESTABLISHED VALIDITY. That means its just a collection of rehashed box scores that have some number produced from it. While it might be intellectually appealing to believe this "hard number" has some useful value.. It might not have any useful value at all. Your just assuming it does.

Its much akin to assuming that astrology is a real science just because they use quantified data about venus rising so and so many degrees and so on. Yes there are real "numbers' there but they might not apply to our lives.

The human mind is far superior at evaluating players to this stat, IMHO. When we watch a game our data set is the entire game. Whereas this stat operates from a limited data set (just the stuff in the box score) and then attempts to quantify those results using rules that originate from nowhere..

Why you think that would out perform basketball professionals is beyond me. Is it better then nothing? Sure - just reading box scores is better then just guessing a players value. PER and WoW are useful - maybe intellectually interesting. But we have very good reasons to question their real world validity when it comes to finally tuned assements of player value.

I don't get in tizzy when I see Ainge offer Rondo for Stuckey and Hamiliton. I have more faith in his insight then wages of win. He has a championship under his belt which validates his insight.

Pete

[wigglestrue]

The human mind is far superior at evaluating players

I agree with Pete. (!)

[Slartibartfast]

Pete, you are placing a lot of value in your intuition. Your intuitive judgment of a player's pretty goodness, which you substantiate by comparing it to the general consensus, is hardly the stuff of a compelling argument. Your eyes and intuition are as likely (I believe more likely) to generate false projections and evaluations. And at least stat-driven analysis opens up discussions that are forbidden according to the dictates of media-driven convention

First off I wasn't talking about MY judgement per se - but the collective judgement of the NBA (gms scouts fans) as being superior to the stat driven "judgement" of the wages of win.

Again the wages of win HAS NO ESTABLISHED VALIDITY. That means its just a collection of rehashed box scores that have some number produced from it. While it might be intellectually appealing to believe this "hard number" has some useful value.. It might not have any useful value at all. Your just assuming it does.

Its much akin to assuming that astrology is a real science just because they use quantified data about venus rising so and so many degrees and so on. Yes there are real "numbers' there but they might not apply to our lives.

The human mind is far superior at evaluating players to this stat, IMHO. When we watch a game our data set is the entire game. Whereas this stat operates from a limited data set (just the stuff in the box score) and then attempts to quantify those results using rules that originate from nowhere..

Why you think that would out perform basketball professionals is beyond me. Is it better then nothing? Sure - just reading box scores is better then just guessing a players value. PER and WoW are useful - maybe intellectually interesting. But we have very good reasons to question their real world validity when it comes to finally tuned assements of player value.

I don't get in tizzy when I see Ainge offer Rondo for Stuckey and Hamiliton. I have more faith in his insight then wages of win. He has a championship under his belt which validates his insight.

Pete

What are your criteria for established validity? The esteem of the public? It should be results. WoW has loved Ainge's best moves and panned his worst ones. WoW predicted the Celtics 2008 title and the Lakers 2009 title. It predicted the failure of Iverson in Detroit, esteemed the Rocket's role players that "overachieved," the failure of the Stephon Marbury and Mikki Moore signings and much more.

You honestly believe Berri just pulled this data out of his butt, or got lucky with an arbitrary system comparable to astrology? I believe there is meaningful correlation between his data and real world results, and his predictions have been in many cases more logical and accurate than those espoused by professionals.

While the noise starts to build when you try and pull off apples-and-oranges comparisons between players like Rondo and Kobe, and to a limited extent, predictions on match-up-weighted contests between teams like the Magic and the Cavs, I think the overall validity of the stat is fairly impressive, and certainly worthy of being a conversation starter for player comparisons.

[wigglestrue]

"Which composite stat has the highest correlation to actually winning games?"

Which one correlates better, PER or WoW?

Is this a stupid question?

Or has no one calculated the correlation coefficient of composite stats to winning percentage?

[GreenDreamer]

"Which composite stat has the highest correlation to actually winning games?"

Which one correlates better, PER or WoW?

None of the above. basketball-reference.com's is better than both, but I think that WoW might have the overall edge on a stat like PER. Why? It translates better. Paul's WoW stats haven't changed all that much, for example, whereas his PER took a dive when Ray and KG showed up.

To really put that into perspective many would agree that last season was his best season ever. His overall game was at its peak, on both ends, and he had a great year. His PER 19.6. The worst season of his prime was 2003-2004. He dealt with nagging injuries, slipped into playing a bad brand of basketball and had very little help around him. Basically a very off year for Paul, barely breaking 40% from the field and shooting below 30% from the 3pt line for the only time of his career. His PER? 19.4. So tell me, how good is a system which rates a guy's best year equal to with his worst one? Why did he rate the same? Well he scored more that season and had one more rebound a game. According to Bsketball-reference.com, though, he had 12.2 winshares last season, his second highest total of his career, and tyhe only reason that his 2001-2002 performance totalled more is that Paul played more minutes that season. In 2003-2004 he had only a 7.3, which makes it easily his worst full season according to Winshare, which is exactly what it was.

Now take Ray. A lot of people thought that this season was his best ever, in many ways. His PER, though, was an "above average" 17.3 (15.0 is Hollinger's statistical average player for that season). His winshare total, though, was the second highest of his career, second only to his excellent 2000-2001 campaignwhen his career was his own career high 22.9. Yet again, though, the only reason that he had more total winshares is that he played more minutes that season. His efficiency numbers were better this season.

Last season Paul finished 10th in the league in winshare, second to KG on the team. This season Ray finished 8th in the league, leading the team. Now that is what I call a system which has some validity to it. One which weighs your numbers against what yourteam did. Guys who are simply stat stuffers, but who don't help their team win are penalized. Those who get those stats while playing winning basketball get more love. It isn't perfect, but it is damn good. Rondo? He finished 14th.

So forgive me for not giving PER all that much respect, because I don't think it is that great of a system. It has its own value, and even Basketball-reference.com records it, but used in a vacuum it can be very misleading. In addition, I find it kind of funny that you are trying to hammer rthis thing home. Rondo had a better PER this season than either Paul or Ray. Does that mean that he had a better year?

Edit: If you add up the winshares for this year's team, you get 61.3. We finished 62-20.

[drza]

"Which composite stat has the highest correlation to actually winning games?"

Which one correlates better, PER or WoW?

Is this a stupid question?

Or has no one calculated the correlation coefficient of composite stats to winning percentage?

I don't know if anyone has ever done the calculation you're asking for. Berri has claimed that his measure is 90 - 95% accurate at estimating a team's wins and losses based on the box score stats of the individuals, but I'm not sure if that's what you're asking for.

If it is, if you go to his 2008 team review page ( http://dberri.wordpress.com/nba-team-reviews-2007-08/ ) and click on the individual teams, it gives the wins produced for each player then adds them up to get a predicted # of wins for the team. That gives an idea of about how close his stat correlates to team winning percentage (i.e. the Hawks had 36.3 wins produced as a team, and won 37 games. The Celtics had 68.3 wins produced, and won 66 games. The Bobcats had 29.5 wins produced and won 32 games. etc.)

Now, that's not to say that he is apportioning the credit for a win "correctly" and many complain that he uses fudge factors to get his numbers closer...I can't speak on that as I don't know. But the actual data that he publishes seems to indicate that his measure does a reasonable job correlating with team wins after the fact.

[drza]

"Which composite stat has the highest correlation to actually winning games?"

Which one correlates better, PER or WoW?

None of the above. basketball-reference.com's is better than both, but I think that WoW might have the overall edge on a stat like PER. Why? It translates better. Paul's WoW stats haven't changed all that much, for example, whereas his PER took a dive when Ray and KG showed up.

To really put that into perspective many would agree that last season was his best season ever. His overall game was at its peak, on both ends, and he had a great year. His PER 19.6. The worst season of his prime was 2003-2004. He dealt with nagging injuries, slipped into playing a bad brand of basketball and had very little help around him. Basically a very off year for Paul, barely breaking 40% from the field and shooting below 30% from the 3pt line for the only time of his career. His PER? 19.4. So tell me, how good is a system which rates a guy's best year equal to with his worst one? Why did he rate the same? Well he scored more that season and had one more rebound a game. According to Bsketball-reference.com, though, he had 12.2 winshares last season, his second highest total of his career, and tyhe only reason that his 2001-2002 performance totalled more is that Paul played more minutes that season. In 2003-2004 he had only a 7.3, which makes it easily his worst full season according to Winshare, which is exactly what it was.

Now take Ray. A lot of people thought that this season was his best ever, in many ways. His PER, though, was an "above average" 17.3 (15.0 is Hollinger's statistical average player for that season). His winshare total, though, was the second highest of his career, second only to his excellent 2000-2001 campaignwhen his career was his own career high 22.9. Yet again, though, the only reason that he had more total winshares is that he played more minutes that season. His efficiency numbers were better this season.

Last season Paul finished 10th in the league in winshare, second to KG on the team. This season Ray finished 8th in the league, leading the team. Now that is what I call a system which has some validity to it. One which weighs your numbers against what yourteam did. Guys who are simply stat stuffers, but who don't help their team win are penalized. Those who get those stats while playing winning basketball get more love. It isn't perfect, but it is damn good. Rondo? He finished 14th.

So forgive me for not giving PER all that much respect, because I don't think it is that great of a system. It has its own value, and even Basketball-reference.com records it, but used in a vacuum it can be very misleading. In addition, I find it kind of funny that you are trying to hammer rthis thing home. Rondo had a better PER this season than either Paul or Ray. Does that mean that he had a better year?

Like I've been saying, every advanced stat has its weaknesses. Win shares has them as well. One such weakness is that defensive win shares rely upon defensive rating, and defensive rating is heavily skewed by team results. Essentially, defensive rating is the defensive part of pure +/-. So just like you would expect all of the starters on the best team to have great pure +/- numbers, likewise, all of the starters on a team with excellent defense will have good defensive ratings. Now, there will be a relative difference among the starters...i.e. the best defender on the team will likely have a better defensive rating than the 4th best, but where the stat is skewed is that the 4th best defender may also have a better rating than the best defender on another team that is actually a better defensive player. Likewise, 2 great and equal defenders on different teams can have dramatically different defensive ratings based upon the caliber of their teammates. And defensive rating is the basis for defensive win shares.

That's why when we were discussing Rondo's defense yesterday, I don't give the most credence to the fact that he had a high number of defensive win shares. He played a lot of minutes on a stifling team defense, which is great, but his on-court/off-court stats suggest (to me, anyway) that the defense is stifling regardless of whether he is present or not. That, to me, is a much better indication of his actual defensive impact than a skewed stat like defensive rating or defensive win shares.

Hey, you see what I'm doing here? I'm INTERPRETING the individual stats based upon their strengths and weaknesses. Now, that doesn't mean that defensive rating or win shares are WORTHLESS...they also bring information to the table, and are worth considering, but they aren't a 1-stop we-know-this-is-right-because-the-stat-says-so analysis either. The exact same thing is true for PER, or for wins produced, or for the family of +/- stats. They are all different ways to measure this complex phenomenon we call basketball. Like Pete and Wiggles have been saying, the human mind is a vital part of basketball analysis, but it works best when it has the most information to work with. I'll repeat again, no 1 stat can answer "who is the best player?" by itself. The question is too complex, and any 1 stat that claims its the holy grail stat to answer it is being used incorrectly. But in my opinion, dismissing a stat out of hand because it can't do something that no one stat can ever legitimately do is just as ignorant as hanging onto any one stat as the holy grail.

[GreenDreamer]

"Which composite stat has the highest correlation to actually winning games?"

Which one correlates better, PER or WoW?

None of the above. basketball-reference.com's is better than both, but I think that WoW might have the overall edge on a stat like PER. Why? It translates better. Paul's WoW stats haven't changed all that much, for example, whereas his PER took a dive when Ray and KG showed up.

To really put that into perspective many would agree that last season was his best season ever. His overall game was at its peak, on both ends, and he had a great year. His PER 19.6. The worst season of his prime was 2003-2004. He dealt with nagging injuries, slipped into playing a bad brand of basketball and had very little help around him. Basically a very off year for Paul, barely breaking 40% from the field and shooting below 30% from the 3pt line for the only time of his career. His PER? 19.4. So tell me, how good is a system which rates a guy's best year equal to with his worst one? Why did he rate the same? Well he scored more that season and had one more rebound a game. According to Bsketball-reference.com, though, he had 12.2 winshares last season, his second highest total of his career, and tyhe only reason that his 2001-2002 performance totalled more is that Paul played more minutes that season. In 2003-2004 he had only a 7.3, which makes it easily his worst full season according to Winshare, which is exactly what it was.

Now take Ray. A lot of people thought that this season was his best ever, in many ways. His PER, though, was an "above average" 17.3 (15.0 is Hollinger's statistical average player for that season). His winshare total, though, was the second highest of his career, second only to his excellent 2000-2001 campaignwhen his career was his own career high 22.9. Yet again, though, the only reason that he had more total winshares is that he played more minutes that season. His efficiency numbers were better this season.

Last season Paul finished 10th in the league in winshare, second to KG on the team. This season Ray finished 8th in the league, leading the team. Now that is what I call a system which has some validity to it. One which weighs your numbers against what yourteam did. Guys who are simply stat stuffers, but who don't help their team win are penalized. Those who get those stats while playing winning basketball get more love. It isn't perfect, but it is damn good. Rondo? He finished 14th.

So forgive me for not giving PER all that much respect, because I don't think it is that great of a system. It has its own value, and even Basketball-reference.com records it, but used in a vacuum it can be very misleading. In addition, I find it kind of funny that you are trying to hammer rthis thing home. Rondo had a better PER this season than either Paul or Ray. Does that mean that he had a better year?

Like I've been saying, every advanced stat has its weaknesses. Win shares has them as well. One such weakness is that defensive win shares rely upon defensive rating, and defensive rating is heavily skewed by team results. Essentially, defensive rating is the defensive part of pure +/-. So just like you would expect all of the starters on the best team to have great pure +/- numbers, likewise, all of the starters on a team with excellent defense will have good defensive ratings. Now, there will be a relative difference among the starters...i.e. the best defender on the team will likely have a better defensive rating than the 4th best, but where the stat is skewed is that the 4th best defender may also have a better rating than the best defender on another team that is actually a better defensive player. Likewise, 2 great and equal defenders on different teams can have dramatically different defensive ratings based upon the caliber of their teammates. And defensive rating is the basis for defensive win shares.

That's why when we were discussing Rondo's defense yesterday, I don't give the most credence to the fact that he had a high number of defensive win shares. He played a lot of minutes on a stifling team defense, which is great, but his on-court/off-court stats suggest (to me, anyway) that the defense is stifling regardless of whether he is present or not. That, to me, is a much better indication of his actual defensive impact than a skewed stat like defensive rating or defensive win shares.

Hey, you see what I'm doing here? I'm INTERPRETING the individual stats based upon their strengths and weaknesses. Now, that doesn't mean that defensive rating or win shares are WORTHLESS...they also bring information to the table, and are worth considering, but they aren't a 1-stop we-know-this-is-right-because-the-stat-says-so analysis either. The exact same thing is true for PER, or for wins produced, or for the family of +/- stats. They are all different ways to measure this complex phenomenon we call basketball. Like Pete and Wiggles have been saying, the human mind is a vital part of basketball analysis, but it works best when it has the most information to work with. I'll repeat again, no 1 stat can answer "who is the best player?" by itself. The question is too complex, and any 1 stat that claims its the holy grail stat to answer it is being used incorrectly. But in my opinion, dismissing a stat out of hand because it can't do something that no one stat can ever legitimately do is just as ignorant as hanging onto any one stat as the holy grail.

I have already said my piece about Rondo's defensive winshare and the ratings. The defensive ratings identified Perk as our second best defensive influence after Garnett (who was second in rating), and Rondo as third. What happens in the playoffs? Perk has a -10.2 in oncourt/offcourt, and Rondo a -9.5. The heat got turned way up, and whatthe bench did against the league as a whole didn't matter anymore.The best defenders showed their worth. The second unit scrubs of inferior teams weren't out there anymore.

In addition to that, as I said the only other guaards to finish as high as Rondo did this season were Jordan and Kidd, who each made 9 all defense teams. You don't accidentally stumble into doing that. Team defense is a great thing, and great coaching makes a huge difference, but great defensive teams ride great defenders. You don't just wheel anybodyout there and get the job done. Can a gerat defemder be underrated in this system because he is playing on a bad defensive team? Sure. Even tehn, though, his great defense will be in stark contrast to his teammates on that team, statistically. Check out the rankings on Basketball-reference.com. 9 of the 10 memebers of the NBA All-Defensive teams for this season are in the top 20 in this stats. The only one who didn't make it was Battier, who missed the first quarter of the season with Injuries. So is this stat really overlooking that many people?

http://www.basketball-reference.com/lea ... aders.html

[GuyClinch]

But in my opinion, dismissing a stat out of hand because it can't do something that no one stat can ever legitimately do is just as ignorant as hanging onto any one stat as the holy grail.

From an analytical perspective you should dismiss unproven statistics. Like I said just because someone has a 'system' we don't know that system actually correlates with real life results. The problem with dberri regression is that its regressive.

What you SHOULD be able to do is track player movement, plug in the win shares, and then predict the wins of a team. Much like with a car I could find the horsepower, peak torque, weight of the car, grip of the tires - and then predict ACCURATELY its quartermile result.

Its not that I don't believe in statistics. Its that your engaging in sloppy thinking. Your saying well you can't "dismiss" a statistic. The onus is on the developer of the statistic to prove its worth. In the scientific world you don't assume theories are correct. The evidence has to be gathered - and then we can say that it MAY be correct.

Pete

[GreenDreamer]

But in my opinion, dismissing a stat out of hand because it can't do something that no one stat can ever legitimately do is just as ignorant as hanging onto any one stat as the holy grail.

From an analytical perspective you should dismiss unproven statistics. Like I said just because someone has a 'system' we don't know that system actually correlates with real life results. The problem with dberri regression is that its regressive.

What you SHOULD be able to do is track player movement, plug in the win shares, and then predict the wins of a team. Much like with a car I could find the horsepower, peak torque, weight of the car, grip of the tires - and then predict ACCURATELY its quartermile result.

Its not that I don't believe in statistics. Its that your engaging in sloppy thinking. Your saying well you can't "dismiss" a statistic. The onus is on the developer of the statistic to prove its worth. In the scientific world you don't assume theories are correct. The evidence has to be gathered - and then we can say that it MAY be correct.

Pete

So, in that case where do the opinions of a bunch of message board yahoos rate on the divine ladder of perfection?

[GuyClinch]

So, in that case where do the opinions of a bunch of message board yahoos rate on the divine ladder of perfection?

On par with the creators of unvalidated statistics. Your issue is you try to dismiss contrary viewpoints on the basis of your stat. It doesn't have the kind of credibility to do that. If you were talking about something more easily quantifiable like free throw percentage then a stat could settle this argument.