Last June I picked the Boston Celtics to defeat the LA Lakers in the 2008 NBA Finals. Even as that happened, though, I warned that the Lakers would probably win in 2009. In fact, last October I argued the Lakers might be the team to beat beyond the 2008-09 season.
After 30 games the Lakers have been very good. The team’s efficiency differential (offensive efficiency minus defensive efficiency) currently stands at 9.90. To put that number in perspective, here are the top five differentials since 1973-74.
- Chicago Bulls (1995-96): 13.00
- Chicago Bulls (1996-97): 11.61
- Boston Celtics [2007-08]: 10.95
- Chicago Bulls (1991-92): 10.64
- Utah Jazz (1996-97): 9.39
In sum, the Lakers – as expected – have thus far been a dominant team. And when we look at the individual players we can see who is responsible for this outcome.
Table One: The LA Lakers in 2008-09 after 30 games
From Table One we see that the Lakers have four players – Pau Gasol, Kobe Bryant, Trevor Ariza, and Andrew Bynum – who have posted WP48 [Wins Produced per 48 minutes] marks above the 0.200 level. Remember, 0.100 mark is average, so each of these players are twice as good as an average player. And this quartet would be a quintet if Lamar Odom was producing as he did last year. Even with Odom’s decline (and the drop-off we see with respect to Bynum), the Lakers are still expected to win 67 games this season. Unfortunately for the Lakers (but fortunately for the rest of the league), the Lakers are not alone on top of the league.
Rondo Stops the Decline
Last year the Celtics won 66 games and took the title. Boston’s only significant defection in the off-season was James Posey, who produced 5.1 wins last season [with a 0.136 WP48]. With most of the team’s top producers of wins returning, one might expect a repeat performance. This, though, was not my expectation entering the season. Here is what I said last August:
“…given the age of its players, we should expect many of the top Celtics to actually decline. Consequently, just like we saw when Bird led Boston to titles in the 1980s, the Celtics are not likely to repeat as champions in 2009.”
Not sure what I meant by “many”, but Table Two indicates that both Kevin Garnett and Paul Pierce – two players who have surpassed 30 years in age – have indeed declined.
Table Two: The Boston Celtics in 2008-09 after 33 games
If nothing else besides the productivity of KG and Pierce had changed, Boston would only be on pace to win 62 games in 2008-09. In other words, Boston should have been passed by the Lakers. After 33 games, though, Boston’s efficiency differential is 10.85, or nearly exactly what this team did last year (and this is after their less than successful recent trip to the West Coast).
So how has Boston remained a dominant team? The answer is almost entirely Rajon Rondo. While KG and Pierce are on pace to produce 8.1 fewer wins this year, Rondo is on pace to provide 8.2 additional victories. Consequently, Rondo is now the most productive player on the Celtics. Yes, Rondo is offering more than KG, Pierce, or Ray Allen.
If we look at the individual stats, we see that Rondo – relative to last year – is a more efficient scoring, tallying more assists, and getting more steals. As a result, his WP48 has risen from 0.219 (which is very good) to 0.371 (which is amazing). Despite this performance, Rondo is not even listed among those receiving All-Star votes in the Eastern Conference.
Cleveland Completes the Trio
The play of Rondo has allowed Boston to keep pace (and currently slightly surpass) the Lakers. Although these teams are each having historic seasons, neither tops the league in efficiency differential. That spot is currently held by the Cleveland Cavaliers. Cleveland’s mark of 12.86 rivals the 1995-96 Chicago Bulls, and it’s the Cavaliers who currently have the best shot to surpass the seventy win mark (assuming current production levels remain unchanged).
Cleveland’s position at the top is certainly a surprise (at least to me). Last year Cleveland had a negative efficiency differential, so the improvement in this team has been astounding. Unlike the Celtics in 2007, Cleveland has not improved by adding two future Hall-of-Fame players. Cleveland’s only major addition last summer was Maurice Williams. Last year Williams posted a 0.122 WP48 mark in Milwaukee. This year his WP48 stands at 0.130. Although both marks are somewhat above average, Williams is simply not offering enough production to transform an average NBA team into the league’s best.
Such was the point I made last November. As I noted, the key to Cleveland’s transformation has been the play of Anderson Varejao, Ben Wallace, and Delonte West. Last year these players only produced 8.4 wins. This year – as Table Three indicates – this trio is on pace to produce nearly 30 victories. Part of this leap is tied to the increase in minutes these players are getting. And part of this is tied to the fact that Varejao and Wallace have essentially reverted to what we saw prior to last season.
Table Three: The Cleveland Cavaliers in 2008-09 after 31 games
The addition of Varejao, Wallace, and West are enough to transform the Cavaliers from a 40 win team (what their efficiency differential suggests they were last year) into a 60 win team. To get to 70 wins, we need to look primarily at the play of LeBron James. King James is producing 5.4 more wins this year than his play last year suggests. And if it weren’t for the play of Chris Paul (and I need to write a post on Paul and the Hornets), LeBron would be the leading candidate for Most Productive Player (M2P) in the league. Fortunately for LeBron, the MVP award is not based on Wins Produced. Consequently, LeBron should have a very good chance of winning his first MVP award in 2009.
Brief Comments on the Other Contenders
Entering this season I worried that 2008-09 season would be all about the Lakers. The dramatic improvement in the play of Rajon Rondo, as well as the changes we have seen in Cleveland, have produced a three-way race for the title.
Unless we see a major injury, the NBA championship will likely be won by one of these three teams. Although our focus will likely remain on Cleveland, Boston, and LA, one might also wonder about the other contenders (or should we say pretenders?).
Immediately after this trio, we have the Orlando Magic (who I discussed a few days ago). The Magic currently have an efficiency differential of 7.8. This is an outstanding mark, but suggests Orlando is not quite good enough to surpass either Boston or Cleveland in the East.
Although Orlando is not likely to catch the top three teams (barring injury), the Magic should maintain the fourth spot in the league efficiency differential rankings. The fifth best team in the NBA is currently Portland. The Trail Blazers efficiency differential is only 4.4, a mark that is barely ahead of the next seven teams on the list (New Orleans, Houston, Atlanta, Utah, San Antonio, Denver, and Dallas).
One should note that this list omits the Detroit Pistons. Last season the Pistons finished with an efficiency differential of 8.17, a differential that was only bested by Boston. Currently Detroit’s mark stands at 0.95. Thirteen NBA teams are currently surpassing what Detroit is doing this year. Even with increases in the production of wins by Amir Johnson [0.171 WP48] and Rodney Stuckey [0.155 WP48], Detroit has clearly fallen from the list of top contenders for the NBA title. This is odd, since seven of the eight players leading the Pistons in minutes played this season were in Detroit last year. The only exception is the addition of Allen Iverson, who currently ranks second among guards in voting for the Eastern Conference All-Star team.
As I noted a few days ago (and a few days before that), it’s difficult to look at this data and reach any other conclusion. Adding Iverson has simply not helped the Pistons.
Fortunately Iverson’s salary comes off the books this summer. This means the Pistons have a chance to acquire a talent (or two) to complement Johnson and Stuckey. So it’s possible the Pistons will return to the top of the NBA sooner (rather than later). And this means that maybe next year, the top teams in the NBA will be a quartet (or is that just wishful thinking on New Years?)
- DJ
The WoW Journal Comments Policy
Our research on the NBA was summarized HERE.
The Technical Notes at wagesofwins.com provides substantially more information on the published research behind Wins Produced and Win Score
Wins Produced, Win Score, and PAWSmin are also discussed in the following posts:
Simple Models of Player Performance
What Wins Produced Says and What It Does Not Say
Introducing PAWSmin — and a Defense of Box Score Statistics
Finally, A Guide to Evaluating Models contains useful hints on how to interpret and evaluate statistical models.
36 responses so far ↓
Italian Stallion // January 2, 2009 at 7:43 am
The Pistons have been playing better since they shuffled the lineup a little. I don’t think they are among the favorites to win the championship, but I suspect that looking at their point differential as a “moving average” could make them seem like more of a threat to upset one of the favorites by season end.
mrparker // January 2, 2009 at 10:52 am
Berri,
Great article as always. Nothing to add. I remember you asking for ideas in a previous post.
Maybe some post on the past 3 draft classes…since this always what I’m most interested in.
Also, I was thinking about what it means to be a negative player. Since a team can’t have negative wins, isn’t this concept a little off. If an all negative player team is fielded the worst they could do is get 0 wins. This might explain how simply subtracting Adam Morrison from having major minutes didn’t vault Charlotte into the playoffs last year. Maybe adding an average player(jason Richardson) can’t have an above average impact. Maybe you are working on this in the new book.
Solid work as always,
Parker
Jeremy // January 2, 2009 at 10:57 am
Rondo for All Star?? Rondo for MVP is more like it! I am as much of a Celtics homer as you will find anywhere, but even I would not have predicted before the season that Rondo would be the most efficient player on the Celtics AND the Lakers, by New Years Day.
The scary part about this, is that Rondo is just now “getting it.” Dude is still raw. When he puts it all together and has more confidence in his jump shot, I wouldnt be surprised to see his WP48 get over .40. Maybe not this year, but Rondo has shown us that he can improve at faster than light speed, so who knows?!?!?!
I just enjoy watching perhaps the leagues most underrated player spearhead Boston’s attack on Offense and Defense.
Michael // January 2, 2009 at 12:18 pm
Rondo for MVP is a little bit over the top. Cleveland looks to be a better team than Boston right now, and Lebron is much more productive than Rondo. Not to mention Chris Paul and Dwight Howard.
Rondo is definitely much improved though. Maybe by the end of the season he will get some attention, the mainstream always seems a little slow to catch on compared to this site!
PJ // January 2, 2009 at 2:06 pm
“a team can’t have negative wins”
Don’t we can these “losses”?
PJ // January 2, 2009 at 2:06 pm
“call,” not “can,” obviously
Westy // January 2, 2009 at 2:16 pm
Dave,
Would you have anticipated some of the fall-off in Bynum’s play due to his now playing next to Gasol (diminishing returns)? Or do you think it’s moreso his still recovering some from the injury?
If it is diminishing returns, it is interesting that Gasol’s WP hasn’t changed. It would be an interesting analysis to see whose individual statistics are suffering the most now that Gasol and Bynum are playing together. I guess a case could be made that it is Odom. No longer is he likely to be the 2nd leading rebounder, and so he must be ceding that role to the other two when he’s on the floor at the same time as them.
Vince G. // January 2, 2009 at 2:19 pm
The astonishing thing is that Rondo is that, much like Ben Wallace or Dennis Rodman, Rondo one of the few non-scorers in the league to be an elite player.
Even more astonishing is that he is doing it as a point guard, when the perception remains that he is still perhaps only the third or fourth best player on his team.
LeBron gets the MVP for two reasons. First, he has a better cast than Paul. Not only does winning grab media attention, but better casts mean that the stars have fewer WP48s than on poorer casts. It is no coincidence that Garnett was routinely the M2P during his years with the Timberwolves’ supporting cast, or that Paul is doing it right now without a player over .200+ WP48.
(P.S.: Okay, I’ll give James Posey a mulligan for the “modest-impact-on-stats-huge-impact-on-winning” credentials. But even then, Paul doesn’t have anyone approaching his caliber in New Orleans. Think Archie Manning without the pads or losing seasons.)
Second, part of LeBron’s trailing in the M2P race is due to playing fewer minutes. As it stands, the Cavs are forcing him to play about eight fewer games (compared to his career minutes) so that he’ll be fresh for the playoffs.
That, in turn, prevents him from building upon his monster WP48 and making it neck-and-neck, or even closer, to Paul.
But even then, for all practical purposes, it’s like picking between steak and lobster. Most GM’s would be absolutely thrilled to have one of those two on his starting roster. Danny Ferry? Nothing more need be said, save this:
Yo, Cavs, why trust the management of the team around your star player to a former player who not only LEFT the team that drafted him, but forced his way out by threatening to not play in the NBA at all?
And to get to Cleveland, the Cavs shipped off Ron Harper to get him in the first place! (For the uninitiated, Mr. Harper was a high-scoring guard who was, if we believe Win Scores, actually above average in Cleveland.)
Okay, enough ranting. Either way, James deserves MVP in my book.
dberri // January 2, 2009 at 2:43 pm
Westy,
Diminishing returns are a possible explanation for Bynum’s decline. But then we have to explain why Gasol, as you note, hasn’t really declined at all. It is possible that Bynum is also a) still recovering from his injury, b) not happy with his minutes (less than 30 per game), and/or c) actually demonstrating what he is generally capable of doing (in other words, last year was an aberration). At this point, I am not sure.
As for Odom… my sense is that Odom is really unhappy coming off the bench. He looks at himself as a player who should be starting and he doesn’t like his reduced role. Again, that is just my sense of things.
Michael // January 2, 2009 at 2:50 pm
Also theres the ever popular “this is his year” argument the media seems to go to, which would seem to fit James after he came so close to winning last year.
As far as Rondo being an elite non scorer, I think you can count Jason Kidd, Andris Biedrins,
Marcus Camby in that category as well, and none of those three get the attention they deserve either!
Do you know I saw a story the other day that the Knicks almost traded David Lee (0.297
after 30 games) to the Nuggets for Linas Klieza! (0.139 after 32) The only reason that it apparently didn’t go through was because George Karl didn’t approve! Crazy right? Well I checked on basketball reference and guess which player has the most games of 20 points or over? Klieza has nearly twice as many! So the Nuggets potentially had a chance to have a starting line up including Camby, Lee, Anthony, and Billups by this point of the season! Madness.
stephanie // January 2, 2009 at 3:14 pm
The most shocking thing to me this season is that the Cavs went from basically a .500 team to being a league dominating force with an ED better than the ‘96 Bulls. Um…what?
PJ // January 2, 2009 at 5:07 pm
stephanie: I suspect that the Cavs improved after the Gooden/Wallace trade and the one that netted Delonte West (along with Zoolander), though I don’t have the numbers to back me up (maybe the team took time to gel?). Then you add Mo Williams; then LeBron, Z, and Varejao all step up their performances; and… voila.
Anyway, the Cavs team that took Boston to seven very tough games in the ECF last year did not look like a .500 team.
Caleb // January 2, 2009 at 5:38 pm
I’m very curious to hear your thoughts on the Hornets. They are doing pretty well (19-9 as I write this, and about to play Portland), but had a rough start and got demolished by both Orlando and the Lakers recently. Though you posted in the offseason that Pargo leaving helps the Hornets… I think it actually hurt the team in the earlygoing this year. The Hornets blog http://www.atthehive.com has a post on this. http://www.atthehive.com/2008/11/29/675365/warning-this-may-blow-your
Fortunately, the Hornets acquired Antonio Daniels, and I think that helped them out as far as having a capable point guard on the floor while Paul is resting. It may be coincidental but thus far the Hornets have yet to lose when Daniels plays.
Also accounting for that slow start was the unexpectedly average play of Tyson Chandler… but he’s had a few games recently that suggests he’s getting back into last season’s form. As a Horents fan I hope so… his rebounding and defense are crucial to the team’s success. Anyway, I’ll be interested to see what you have to say on the 08-09 Hornets.
Vince Gagliano // January 2, 2009 at 8:55 pm
Mike:
I came close to adding Kidd, but didn’t. Here’s why.
Kidd doesn’t really score 20+ points a night. However, his career high was 18.7 a game in New Jersey, with four other seasons where he scored at least 15 per.
Rondo is pulling off his numbers with 11 points a night. Trust me, there’s a difference.
Kidd was, in reputation and in Win Scores, one one of the top three players on those Net teams. Rondo has to play second fiddle to the Big Three every night. (Sorry I had to invoke the broken record line.)
I will say this, it’s particularly tough to succeed at the non-scoring act as a point guard, shooting guard, or small forward. But ironically, Harper pulled it off on the second set of three-peating Bulls and the first two editions of the Shaq and Kobe championship Lakers.
Andrew // January 3, 2009 at 7:26 am
Rondo is a great player, a guy who could start on almost every team in the NBA and probably even score more on most of them.
But let’s imagine for a second that Devin Harris and Rondo swap places. If Rondo is one of the top two guys in a New Jersey system geared around his immense physical talents (which are pretty similar to Harris’s), I’d expect a similar result: much more gaudy scoring numbers with less efficient percentages.
Nobody is denying that Rondo is a great player who is integral to the success of the Celtics, but I don’t think it’s right to say that he “offers more” than Garnett, Pierce, and Allen because of WP48.
The idea that Rondo “has to play second fiddle” to the rest of the big three is ludicrous. Rondo is a different kind of player because his game is not based around getting a mid-range or perimeter shot. Rondo is a slashing player who has ridiculous talent once he gets into the paint, a lot like Tony Parker.
This year, Boston’s offense runs much more through Rondo than it did last year (Berri’s numbers would probably tell you that Garnett is worth fewer wins than last year because his assists are down, when in reality the only thing that has changed is the situations in which Garnett receives the ball–he’s still just as good a passer and makes the same decisions, but he’s more often in a position to make shots instead of taking it to the hoop or passing out–witness fewer FT attempts and fewer assists).
Rondo receives screens from two of the most skilled screen-setters in the league in Perkins and Garnett, and he doesn’t have to force any forays into the paint into shots because of his teammates. He has the ability to kick it to two 40% three-point shooters in Allen and Pierce, three of the deadliest mid-range shooters in the game in those two and Garnett, or two paint presences if the post defender switches (Garnett and Perkins).
Rondo is turning into a great player, but he’s not the most important piece on the Celtics. We have to recognize that Rondo is put in a position to excel because his fantastic teammates ensure that Rondo takes only his highest percentage shots.
So the idea that Rondo plays “second-fiddle” to the big three might be true in terms of shot attempts, but that’s exactly what allows Rondo to be one of the most efficient players in the league. He doesn’t have to take the shots that similarly talented players on worse teams do, because he has three of the best shotmakers in the NBA on the court with him most of the time.
This might seem like very specific criticism, and it is surely specific to this view of Rondo. But it goes back to what I think is the biggest weakness of the WP48 system: the WP numbers do a good job of predicting team performance, but there’s something wrong with how they break down player-by-player. Rondo should get credit for making an insane percentage of his shots, but what if his shot attempts aren’t created equally as say, Devin Harris’s? Maybe the league average starting point guard makes 50% of the attempts that Rondo gets, and 45% of the attempts that Harris gets.
mrparker // January 3, 2009 at 7:26 am
PJ,
You’re supposed to be able to add up WP and get close to the team win total. A team made up of the twelve worst players in the league would still have at least 0 wins. That is my only rub with wp.
Vince Gagliano // January 3, 2009 at 8:22 am
I meant Rondo played second fiddle in terms of popular perception, not actual play.
Barring a 20-points-per-game effort, if Boston wins another title or two, they’ll always be viewed as the Garnett-Pierce-(Ray) Allen Celtics, not the Rondo-Garnett-Allen Celtics or even the Rondo-Pierce-Allen-Garnett Celtics. They certainly weren’t the Garnett-Pierce-Allen-Posey Celtics in pop culture.
In revisionist history, people are always going to view KG and Allen as saviors to a once-proud franchise. Danny Ainge will be the mastermind behind the operation, much like Red was. The KG trade will become McHale’s Folly. And Clay Bennett supervising the trading of his own franchise player will be the beginning of the end of the Team Formerly Known as the Sonics in Seattle.
But we all know better than that.
Italian Stallion // January 3, 2009 at 8:48 am
Good insights Andrew.
Sometimes analyzing statistics is like looking into a rear view mirror. They tell you what happened, but they don’t always tell you why or what will/would happen with some changes in the team.
Ralph // January 3, 2009 at 11:09 pm
Article on Rondo in New York Timer today, http://www.nytimes.com/2009/01/04/sports/basketball/04celtics.html
Ralph // January 3, 2009 at 11:09 pm
Dave, The article agrees with you but says that other players are double teamed leaving Rondo open. How do the stats capture that this will make his shooting percentege higher?
Robert // January 4, 2009 at 12:08 am
The other bound that people are forgetting about (a criticism of WP48/wins produced in general) is it’s not possible for a team to have a WP48 >1.
In other words, the BEST 12 players (or the USA Olympic team) would be expected to go 82-0, yet the individual players who do more on other teams wouldn’t get proper credit. This is a failing of WP48.
About Andrew’s comment, I live in Cambridge and watch the Celtics religiously, I love Rondo as well, but he’s not the most important Celtic. It’s actually laughable to insinuate that, even with the propping of WP48.
Should he be an all star? Without a doubt. But is this the “Rondo led Celtics”? No. It’s KG’s team.
Hard To Find Good Help « RedsArmy.com: The Voice Of Celtics Fans // January 4, 2009 at 6:42 am
[...] there. TrueHoop had this link… which proves what we’ve been saying since day one: When Rondo is on, the Celtics are unbeatable. While KG and Pierce are on pace to produce 8.1 fewer wins this year, Rondo is on pace to provide [...]
PJ // January 4, 2009 at 7:34 am
“In other words, the BEST 12 players (or the USA Olympic team) would be expected to go 82-0, yet the individual players who do more on other teams wouldn’t get proper credit. This is a failing of WP48.”
If I follow you, this is what Berri would call “diminishing returns,” and applies not just to WP48, but most other stats (put great rebounders, scorers, assist-men on a team together, and their individual stats will likely go down, b/c there are only so many points, rebounds, and assists to be had).
Joe // January 4, 2009 at 7:54 am
The focus of the arguments here is on scoring…
Rondo creates a lot more possessions due to his superior rebounding and his crafty steals. In addition, his turnovers are very modest relative to his plump assist totals.
Rondo, in his 3rd season as a pro, has played ~1000 minutes. Devin Harris, in his 3rd season as a pro and a year older, played ~2000 minutes. Rondo has already surpassed his win production and there are ~50 games left in the season for the Celts. In other words, at 23, Harris wasn’t worth mentioning in the same breath as Rondo at 22. The same can be said at 21 and 20, as well as 22 and 21.
Put simple, at 23 Devin Harris wasn’t a scorer and was average. At 22, Rondo also isn’t a scorer, is much better than average.
Now, at 25, Harris is a scorer and is well above average(est ~.2 WP48), but not close to Rondo.(est .3+).
One should also note that before the whole “big 3″ Rajon Rondo was the most productive rookie from his class unless I am mistaken. Is Rondo going to continue to be a .3+ player? I’m not sure about that, but he certainly seems like he will be .2+ from here on out every single season with few or no exceptions. And most likely he hasn’t peaked, since most don’t at 22.
Note, if you were to simply drop Rondo to a guy shooting 44% at the same amount of shot attempts, he would still be in the .25 range from what I am seeing and thus still noticibly better than Devin Harris. And still on the upswing of his career.
Most valuable, taken literally, might apply here. I’ll cop out and say he is in the top 2 in value to the Celts along with Garnett.
Michael // January 4, 2009 at 8:48 am
“it’s not possible for a team to have a WP48 >1″
Of course it isn’t. How is a team going to produce more than one win per game?
There are only 1.00 wins to be had per game, the contributions of various players on the team lead to that win. Even a theoretical team of 0.200 players would not win 82 games because of diminishing returns and the effect of smaller roles (12 players can’t all start and play 30+ mins per game.) So its a nice idea to say in theory, but practically by seasons end some of those players would be at least below 0.200 and some games would be lost.
The same goes for teams theoretically producing ‘negative wins’ It is a cute idea but lets not go all theoretical physics here and start trying to divide by zero! There are valid criticisms of the wins produced model yes but not these two arguments!
Ralph // January 4, 2009 at 10:20 am
A negative win is a loss. Duh!
Michael // January 4, 2009 at 11:36 am
No it really is isn’t.
A PLAYER can produce negative wins and cost his team games, but a TEAM itself cannot produce negative wins. The worst a team could possible do is produce ZERO wins and lose every game, to lose more than every game you play is impossible. This is the reason that you can have negative win producing players, but not teams. A negative win is not a loss, rather it is the result of a player producing at such a low rate (or at a high rate in negative areas such as wasted possessions) compared to the average player that it will eventually cost the team games.
All I was saying is that hypothetical areas such as WP48s above 1.00 or team wins below zero are a practical impossibility and thus do not provide the basis for a criticism of the wins produced model.
Ralph // January 4, 2009 at 12:15 pm
Michael, it does provide a basis for criticism of the wins produced model. Wins produced is a good model but like all models it is a simplifying approximation of reality with a bunch of arbitrary assumptions. It does not control for interaction of the players. For example, if Garnett or Pierce are double teamed and Rondo is left with open shots his win score will be boosted artificially. Also, win score ignores defense except to the extent defense is reflected in (defensive) rebounds. That doesn’t make it a bad model. This stuff wouldn’t be reflected in box score stats. You just need to be aware of that context in looking at win score.
What I think would be a good statistical test of win score is how autocorrelated player performance is upon switching teams. (This might have been done. I’m not sure.)
Ralph // January 4, 2009 at 12:16 pm
(I understand your point now about negative wins. Sry for the glib reply.)
Robert // January 4, 2009 at 12:44 pm
Question regarding diminishing returns on a very good team (for a player like Rondo) – does the wins produced model than suggest on a terrible team, Rondo would have a higher WP48? Meaning, the worse a team generally is the more inflated individual stats are – this would make the converse true. (On really stacked teams, individuals suppress their own WP48 etc.)
if so, for addressing player movements across teams (ala Brand to the Sixers this offseason) some sort of scaling system should be used to readjust their assumed impact.
Michael // January 4, 2009 at 12:57 pm
What your talking about is the ecological fallacy criticism of the Wins produced model. This is defined (by Wiki I should add) as “an error in the interpretation of statistical data in an ecological study, whereby inferences about the nature of specific individuals are based solely upon aggregate statistics collected for the group to which those individuals belong.”
Basically the criticism suggests that success at the team level is not necessarily linked to success at the level of individual players. This is not the same as talking about negative wins or WP48 over 1.00 (as you seem to infer) but is a seperate argument.
I think Professor Berri has responded to the ecological fallacy argument successfully in the past when he has stated that wins produced may tell you how successful individual players are, but it cannot tell you what the causes are for that success (why.) For the why you have to engage your own mind, similar to how Andrew has above concerning Rondo.
I’d also like to add that the model is based on statistical regressions and not ‘arbitrary assumptions’, a criticism which is better left to metrics such as John Hollingers PER, which is modelled on assumptions.
What Might Have Been in Denver « The Wages of Wins Journal // January 4, 2009 at 11:10 pm
[...] NBA Mid-Season Analysis: 2008-09 ← Contending on New Years Day [...]
Andrew // January 6, 2009 at 5:45 am
I don’t necessarily disagree with Berri’s contention (if he so contends; I haven’t read this blog for that long) that a stat measuring player success doesn’t necessarily have to tell you why players are successful. But in that case, I do disagree with the idea that Rondo “offers more” than the Big Three; it seems like a much more responsible thing to say would be that Rondo helps Boston’s Wins Produced the most (and, one might add, that this fact might not necessarily mean that Rondo is the best or even the most valuable Celtic).
As for Joe’s criticism of my Harris thought experiment, I think his argument falls into a dangerous mode; namely, that he’s using numbers within a system to justify the system itself. In other words, when I say that Wins Produced might not effectively show player value, you can’t say “well, Harris has fewer Wins Produced, therefore your argument is moot.”
And I’m not just talking about scoring, I’m talking about the whole package. Harris is also a steals magnet (though not as good as Rondo). The biggest difference is their rebounding and youth, which I will agree makes Rondo a better player (I think Rondo would be the more valuable of the two to almost any team).
But let’s imagine for a second that Rondo was asked to take the number of shots that Harris is taking in New Jersey, and Harris was asked to take the number of shots that Rondo is taking in Boston. So instead of taking 8.3 shots, Rondo is taking 15.3. And maybe Rondo makes 46.5% of the shots that Harris takes instead of Harris’s 45.5%, and maybe Harris makes 50.7% of the shots that Rondo takes instead of Rondo’s 51.7%. That seems like a pretty generous assumption, considering that Rondo has never taken more than ten shots per game.
Also, imagine that whenever Harris dishes a potential assist, he has Boston’s shooters instead of New Jersey’s. That would do a nice number for his assist rate.
Basically, it’s not out of the realm that if we swapped Harris and Rondo, Harris would suddenly look like the poor man’s Chris Paul, and Rondo would be the merely solid player. I agree that Rondo is clearly the better player, and I think this just accentuates the issue with WP48: a better player can be said to have produced fewer wins than a lesser player because of team context.
Joe // January 7, 2009 at 8:49 am
Andrew,
I highly doubt it. Harris was in that exact position in Dallas with the same amount of experiecne as Rondo in Boston essentially. Rondo was significantly better. As I also sao, Rondo was the most productive rookie from his class before the big 3 arrived. Your argument has no real weight.
Andrew // January 7, 2009 at 11:15 am
From TrueHoop yesterday: Rondo is shooting a brilliant 52% this season, as good as any point guard. How can you leave him open? A peek at NBA.com’s hot zones reveals he has made 127 of his 205 shots at the rim. Everywhere else on the floor, however, he is a miserable 23 of 85. That’s right. According to NBA.com, Rondo has hit a total of 23 buckets not at the rim, all season. He has played 1100 minutes over 35 games, so that’s about one made jumper per 48 minutes played. And even though he is being left open game after game, he’s still taking it to the rim again and again — as reflected in his 205 shots at the cup, compared to 85 from elsewhere. Nevertheless, he’s a great player. With the right shooting coach, and some real dedication on his part, he could be insanely good.
If we use Harris’s zone percentages on the number of shot attempts that Rondo gets from each zone of the court, we get Harris shooting 50% on Rondo’s shot attempts (Rondo shot 51% on the same attempts).
And stop talking about Harris from years past. I’m merely using this example as an exercise to show that the type of shots that a player is allowed because of the other players on the floor has a profound impact on shooting percentage.
By the way, reversing the above exercise (applying Rondo’s zone percentages to the shots that Harris takes) yields 43.7% for Rondo, or 1.8% lower than Harris’s actual shooting percentage. If Harris were in an offense that allowed him to go to the cup every time, and if Rondo were forced to hoist midrange shots, the difference in their shooting percentages would be massive and would affect their WP48 accordingly.
Andrew // January 7, 2009 at 11:20 am
Also, just so I don’t come off as down on Rondo, consider this amazing stat. Rondo finishes at the dish an amazing 62% of the time. Harris stands at only 55%. Before dismissing that as bad, consider that 55% is also the rate at which Tony Parker makes his shots around the basket.