Statistically, Rondo Should Have Been ROY

I don’t really care that Roy won the award, because it seems like that will be all he ever wins in his career. A) He is in Portland. B) He doesn’t have star potential. C) He is an average at best defender.

I think when all is said and done Rondo will have much more success in the NBA, win wise and award wise. If Rondo develops even an average jump shot he will be one of the top 5 PGs in the NBA. Rondo just does everything.

Also, this was the weakest draft that I can remember. That’s like winning a gold in the special olympics.

I don’t want to get into another huge argument with the likes of Scintan and ehmunro, but these guys from the Wages of Wins have it nailed. The comparison of Duncan and Garnett in the article immediately preceding the article about the best rookies is also right on the money.

I was at the Vegas Summer League for Rondo’s first ever Celtic games…Everyone in the building knew Rajon was the real deal…Everyone except Doc Rivers I guess.

Wow. That’s an amazing statistic.

I’m not aware of this WP48 stat…any idea how it is calculated or where we can find information about it for other players?

Jaysus, with that reasoning, Linden Larouch (sp?) should be president.

LaRouche =P

gotta love Boston homers

Uggh. Not the wages of win again what a bunch of crap psuedoscience..

I love Rondo. I love attempts at using new advanced statistics to get at more subtle contributions and effectiveness. But I’ve read Wages of Wins, and Berri et al. have such gaping holes in the logic behind their calculations, and don’t even blush when they claim that Iverson(who I don’t love but recognize his ability) is an average NBA player at best or that Michael Jordan didn’t make those who played with him better (because his teammates stats went up when he “retired” – no kidding, there’s some advanced thinking), that their credibility is a little thin.

Logic this one out. The Cs were 4-31 when Pierce did not play and in two of those wins Rondo played 6 and 12 minutes. So they were 20-27 with Pierce, so it stands to reason about 15 of those wins are because of Pierce. Jefferson is clearly the main other reason any games were won. So exactly how does Rondo figure into 7 Wins Produced?

Where most of the advanced stats fall down is in crediting garbage time minutes and minutes played against second teamers the same way as competitive minutes against starters which matters little for full-time starters, but does for comparing bench players/part-time starters with consistent starters. Rondo had a fairly sizeable amount of these minutes which elevates his /- and pretty much every other stat. What also does not seem make a whole lot of sense, is that The Wages of Wins authors claim one thing that separates them is factoring in efficiency a lot more than say the NBA Efficiency stat. So how does a guy like Rondo who had a mediocre at best FG% and FT% among rookies rate so highly?

Again, I think Rondo is potentially a very fine player, I advocated him as the starting point at the beginning of the year. And yes, this was such an atrocious year for rookies, with Roy the only guy having what I wold consider a notable first year, that Rondo being among the best rookies is not surprising. But, this 7 Wins Produced stat for Rondo just doesn’t make much sense when Bargnani and Garbajosa, guys who played meaningful minutes on a 47 win team, basically have no Wins Produced when their numbers are combined. But Rondo has 7, on a 24 win team that may have been a 10 win team without Pierce all year? Please.

Even when you disagree with Hollinger, you can see all the numbers and formulas he’s using, laid bare. You can question his assumptions, etc. With the WoW people – they’re just important looking numbers, with no science behond them – at least, no science that they’re willing to show. Except that they overvalue rebounds to a truly dramatic extent.

I love Rondo. But did he have a better rookie year then Bargnani? Not buying that one. Not to mention that Garnett vs. Duncan article he had before it. Yikes.

I don’t know much about the stats here, but I’m pretty impressed by any system that has David Lee as the most valuable player in the league.

This stat, like all others, cannot be used in a vacuum. Personally, I don’t care too much for stats that are subjective in nature, either. Still, it is yet another indicator of Rondo’s VALUE to his team. The thing about Rajon is that while he really came on offensively at the end of the year, it was his defense that really set him apart throughout the season. Defense is the hardest aspect of the game to guage by the numbers, but it isn’t too hard to see Rondo’s effect on the Celtics’ team defense. When he was on the court the Celtics had the same defensive rating as the Detroit Pistons, when he was off of the court they had the defensive rating of the Grizzlies.

Did his coming off of the bench have something do to with that? Yes, but he played almost half of his minutes as a starter, and he also finished quite a few NON blowout games for the team coming off of the bench. I also don’t think that the whole “garbage time” argument carries too much water with regards to Rondo. Paul, Al and Wally all had their highest player pair plus/minus ratings with Rajon. Who do they have their next highest ratings with? They have them with each other. In other words, the four best players on the team had the most success when playing wth each other. It isn’t rocket science.

What I find to be the most interesting comparison between Roy and Rondo is their offensive and defensive ratings. Roy had an offensive rating of 107.2 and a defensive rating of 113.1. Rondo had an offensive rating of 104.9 and a defensive rating of 104.1. With offensive ratings higher is good, with defensive ratings higher is bad. Rondo’s offensive rating was 1.0 more points per 100 poss. for the Celtics when he was on the court than off of it, and his defensive rating represented a 8.3 fewer points surrendered. Roy’s offensive rating represented a 2.6 point increase for his team, but his defensive rating represented a 4 point increase (worse) than they had with him off of the court. So Roy’s offensive helped his team score more points, but with his extra offense came an even greater defensive hit. Rondo not only brought extra offense, he made his team MUCH better defensively. I think that is why his stat is so high in this case.

One final observation. Doc had to get really inventive in throwing games at the end of this past season, didn’t he? Paul was out, Al was out for a while, yet Doc still had to coach his worst to ensure the losses. When we were going through that 18 game losing streak, we couldn’t buy a win, yet at the end of the Doc was forced to scramble for losses. Al played throughout the losing streak, but who was starting at the end of the season, often without Al? Rondo. Does that mean that I think Rondo is better than Al? No, I firmly believe that the pecking order on this team is Pierce, Jefferson, Rondo. Still, Rondo’s defensive impact was so large that the team remained competitive when he was out there without our two best players.

Roy had the glitzy stats, but Rondo had the more profound impact upon his own team.

At the very least, maybe we can put to rest the idea of drafting a PG and moving Rondo. I don’t understand the value of applying fantasy stats to team players, but Rondo is a player, and a Rondo/West PG situation is not the neediest poistion for this draft.

Figures don’t lie, but liars can figure.

Stats analysis is for people with too much time on their hands (kinda like blogging). The fact is Roy was selected as ROY and Rondo was not. Perhaps Rondo will turn out to be the better player; perhaps not. I’d still take Bargnani or Tyrus Thomas over either player.

Ha! Hank beat me to it. What he said.

The Wages of Wins guys have some serious problems with their approach to statistics (for starters, they have practically an entire chapter about the fact that Iverson is secretly a really bad player). There are entire threads on the ABPR site about the problems with their approach.

I’m a big believer in the use of stats to help organize and understand our world, but if something doesn’t pass the laugh test, then you’ve got a problem.

And I’m sorry, as much as I love Rondo, this one just doesn’t pass the laugh test.

The numbers that Berri, et. al. use are perfectly logical if you understand regression analysis. These are the same techniques that economists use in other areas. Most of those who debunk these methods have never read the book, or if they did, they didn’t understand it.

Lets see. The hedge fund Long-Term Credit had three, or was it four, Nobel prize-winning economists on board. They knew all the formulas for financial success. Two years later the fund went bankrupt and nearly brought down the entire international financial system. Are these guys now working for Rondo or the Celts?

Having read and understood the book perfectly well, I beg to differ. Simply because a well-known and accepted tool has been applied, does not mean that it has been applied correctly. There are numerous examples of the mis-application of regression analysis in the literature. One of regressions major limitations is that it assumes that the relationship between variables travels in a straight line. If that assumption is not, in fact, a good one, then the results of the analysis can be essentially garbage.

Also, the state of the art in analysis has moved well beyond regression at this point (in large measure due to the limitations of traditional regression). Analysts use tools such as support vector machines, genetic algorithms, and decision trees more and more frequently, as the limitations of regression are recognized.

For a full discussion of some of the limitations of the Wages of Wins by the top statisticians in basketball, check out this thread http://sonicscentral.com/apbrmetrics/viewtopic.php?t=1002&highlight=wages wins on the ABPR Metrics board. There is an earlier thread on the topic as well, if you care to search for it.

Oh, don’t get Pretentious Petula.

The regression methology isn’t perfect, but it’s better than 90 percent of the other basketball analytical tools out there, most of which are biased in favor of offense over defense.

When I watch basketball games, I know instinctively that players like Rondo, Balkman and Millsap contribute more to success than many other more highly touted players.

They were right about Iverson weren’t they? I’m not going to rehash this discussion further, which went on ad nauseum at the time of the proposed Iverson trade.

Let me add that I am not impressed with links to competing analytical organizations like abprmetrics and 82games.com which have a vested interest in debunking the Wages of Wins methodology in favor of their own.

First of all, Brickowski I am not the one who “got pretentious” by implying that anyone who disagrees with the Wages of Wins is either uninformed or stupid.

Secondly if you ever checked out ABPR you would learn that it is not a uniform organization with a position. It is a place where people interested in basketball stats (with wildly varying viewpoints at times) hash out their differences, educate, and generally try to advance the state of the art in basketball analytics.

If you use regression analysis without a strong understanding of the variables that impact it you can get some funky findngs, period. That’s why you can’t be an economist one day and then say “now I’m a basketball analyst” and necessarily be using regression appropriately; Berri et al are apparently football and baseball experts also as they have chapters on tha also. Most college sophomores taking a stat class can run a regression analysis and get an equation, and there are plenty of PHD academics out there who have a shaky understanding of the limits of the anytics they use. Regression analysis may not even be the best analytic tool for what they are trying to do. Most of the people who have taken issue with Berri et al. are either academics (as they are) and/or experienced basketball analysts. People quite capable of understanding the Berri book.

Most of the basketball analysts out there will tell you that even the metrics they have produced themselves have flaws. Everyone is trying to get a handle on the complexity of the interactions that happen in a dynamic sport. For example, I actually agree with Berri et al. that Iverson is overrated because of the inefficiencies in his game, I just don’t believe he is an average NBA player at best as they do (page 136 of Wages ..“the net value of Iverson during his career is a bit below the average NBA player.” They cite that he only led the Sixers in Wins Produced one time in his career. Again, that just does not make sense. Berri et al. don’t understand that some players are forced to be inefficient in a way that enhances the efficiency of their teammates. There are times when a team is forced into a tough shot, that Iverson may hit only 30% of th etime, but if anyone ese tried it it would only go in 10% of the time. Players like Kobe, Pierce, Iverson etc. are forced into inefficiency at times because they were the only one capable of pulling off certain shots. If you never attemp a tough shot, your efficiency looks great which is why players like Tyson Chandler, David Lee, Mutumbio often have inflated offensive efficiency scores.

Berri et al. have contributed to the discussion of plyer evaluation in a provocative way, but logic indicates thatthe conclusions they reach based on their stats point to gaps in their derivations. I think they spend too much me defending themselve than reflecting on how they can improve their metrics.

Their conclusions on Iverson make perfect sense to me.

Let’s leave it at this. I’ve been playing, coaching and watching basketball since before you were born (probably). The first NBA game I can remember watching was in 1960 when my father took me to MSG.

So it’s good for an opinionated old man’s ego when some young Turks come up with a statistical method that confirms with about 98% accuracy the subjective conclusions that the old man has reached about players by simply watching games, even if the statistical method itself is viewed by some as pseudo-scientific balderdash.

There are two flaws that I can see with almost every numerical system for evaluating baksetball. First, offense is overvalued relative to defense. That’s because we don’t keep good defensive statistics. For example, where’s the statistic for the guy who boxes out his man so that a teammate can get the rebound?

Even when you go to the offensive side of the ball, where’s the statistic for saving a possession by diving on the floor for an errant pass? Oh sure, if it was the other team’s ball, it’s a steal. But who counts the number of his own team’s possessions that a player saves with those kind of hustle plays?

Second, there is poor understanding of the synergy among players, and although the authors of the Wages of Wins raise this issue, they haven’t solved it, and neither has anyone else. We compare player A with player B. But we can’t capture very well why certain groups of players play better together than other groups, or how groups affect indivdual performance.

But I’m tired of the argument. Maybe statistic enhance one’s enjoyment of a glacially slow game like baseball, but basketball is much faster paced and much more artistic, like modern dance. I glance at box scores like everyone else, but the numbers don’t really capture what I see.

Brick – I agree with both your points of what the primary weaknesses about basketball stats, I’ve said similar things myself. Hopefullydefensive stats can improve some by capturing stops, double teams, who is guaurdig who etc.. The synergy issue is a bigger hurdle and may well always be fairly subjective and much more art than science.

Believe it or not Brick there are others over the age of 40 on this site that know a thing or two about basketball and some have subjective opinions that are supported by other facts. In almost any field of interest there are knowledgable people who disagree and find data to support their views, and they obviously defend and gravitate toward that information. For me, after watching nearly every Celtics game this year, to say Rondo has 7 “Wins Produced” and Bargnani and Garbajosa have a combined zero makes little sense to me; and I love Rondo, and have since he was at Kentucky, and think he’s going to be very good. Logically though that just does not resonate with me, and have some ideas of why it does not work as a metric.

But you’re tired of the argument and do not sound like what Berri et al. or other analysts have to say interests your more artistic bent (by he way, I also think basketball can never be the statistical equivalent of baseball, there are too many interactions,. Given that, I’m surprised you so vigorously defended them. If stats are only valuable when they agree with your viewpoint, what good are they anyway? Some confirmation is nice, but I want some insight into long-term trend across players and teams that are difficult to pick up by only being able to watch a fraction of all NBA games.

Well the “wins” score makes perfect sense to me. And I believe the scores, however flawed, have predictive validity. A player with a high score will help you win games more than a player with a low score becase the player with the higher score does more constructive things with his time on the floor.

Would you trade Rondo for Roy? I wouldn’t.

That’s a tough one. Roy was my favorite player in last year’s draft, I like his all around game – I could not fathom how anyone could take Morrison or Shelden Williams ahead of him. However, I like and value more than a combo guard, a true point guard with defensive quickness and an ability to run a team, and I just like what I see, so I stick with Rondo. But to say he had the best rookie year, that’s what I question. If you use the metric as a predictive measure, you might look at WP/48 to account for time played and then Balkman blows everyone away.

The predictive vaidity question is interesting, I’m not sure Berri et al. have used their scores to predict future performance but most measures do ok at the very highs and very lows, but how well do they distinguish those mid-level performers is the most telling. This rookie class would be an interestng test because the performances are so ordinary. It will be telling if Rondo, Roy, Balkman, and Sergio Rodriquez (their top 4 per 48 scorers) end up among the best players in this draft and Morrison (any measure including watching him play should tell you he stinks), Bargnani, Gay, and Carney the busts, and whether Wins Produced/48 does better than PER, NBA Efficiency or some of the other popular metrics.

Actually I think that the true ROY was Millsap. He’s the only one still playing, and he’s made a very positive impact in two tough playoff series.

The other “bust” is Marcus Williams, who can’t defend a chair, and hardly saw the floor in the playoffs.

Defensive statistics are not going to ever be kept as defense is usually almost always a team oriented. I agree with those that say defensive stats are not represented well in determining ones value or one’s measure of leading a team to wins.

One of my most notorious discussions is about Kwame Brown’s defensive rebounding stats. It looks like he is a below average rebounder. Kwame boxes out his man as well as anyone in the NBA. He may not get the rebound himself and shoot, the Lakers may even give up the Offensive rebound at times. However Kwame usually finds a big body and takes him out completely. Lamar Odom had one of the higher defensive rebounding stat line in the league for power forwards/small forwards yet his numbers were alot higher when Kwame was the one at Center. It looks like Odom is the more valuable guy but the Kwame’s of the world are needed just as much as the Odom’s of the world when defensive rebounding.

Brick, i disagree about Marcus Williams. You forget that Jason Kidd plays point guard for the Nets. He was playing 43-45 minutes per game. Kidd and Marcus can not play the 2guard. Therefore, Marcus’ minutes would be limited in this situation. Marcus impacted games just as much as Rondo at times when he came into games. Marcus’ scoring ability should not be underestimated.