I get bored very easily and when I’m bored it’s hard to keep my attention focused. My mind wonders until I find something of interest to keep me entertained (exhibit A: this blog :-))
I bring this up because I got bored at one point for a few hours during the weekend during a christmas related activity. Bored to tears actually and my mind wandered to the question of value.
Now the idea of value came to my head because of a post I did a few days ago. In it I debuted my version 2 of opponent adjusted Wins Produced Model (see the Basics for details). The idea was to take into account not just what a player does in terms of point generation but how his direct opposition does as well. Granted the model has some limitations (i.e as of now I can only do opponent adjustment at the position level and this tends to give a flatter distribution because exceptional defenders (in terms of limiting productivity/point generation) may be grouped with normal or below average ones).
This post got this blog some attention, because when the numbers where crunched over the last ten games, Kevin Garnett was the MVP. I then proceeded to apply this analysis to the 2009-2010 season (with some surprising and unsurprising results). After that I was all fired up to throw the new metric at the current season and provide some sanity to the MVP discussion (and maybe get all that ridiculous Rose and Amare MVP talk squashed). This brings us back to the start and my musings: Just what is value in Basketball?
Value in basketball terms for a player to me can be easily defined in terms of Wins Produced for the team and we know those are highly correlated to point differential. By that logic, in the simplest terms the MVP must be the player that has generated the most Wins or point differential for his team. Now there are some additional factors to consider. Where you play your homegames is important (i.e. the Jazz and Nuggets players are teeing off from the kiddie tee and I’ll probably expand on this in the future) . How good your team is also very important (i.e. wins on Bad teams are easier to come by than wins on good teams again something that will come up again possibly even in the comments for this post).
For this initial version we’ll keep it simple, the MVP will the player generating the most bang (in terms of wins and points for his team). In the future we’ll increase the complexity (gives you something to look forward to). So my narrative path was all set, I’ll write a post in which I look at the numbers and prove that one of my favorite players is back at the top of the league.
Now before we get into I wanted to give some perspective. Here’s the table of the all time Wins Produced leaders since the 1977-1978 season:
These numbers are current thru the start of the season. So KG is making an assault on mount Olympus. By the end of his career he should sit comfortably at the top of this list. The exciting part is that while it looked for a while there like he might limp to the finish, this season KG looks like he’ll take the top spot going away.
Now the big question that remains is do the numbers match the narrative? Not quite:
Lot’s of work for a familiar answer. The MVP so far?
As for the adjustments? Well the Heat play in the flattest of states and have the highest differential and Wins Produced of any team, so Lebron’s numbers will only improve by comparison. KG does crack the top ten (as does Kobe). Rose and Amare are not close (Rose at least manages a positive contribution and a top 50 appearance. Turns out the Heat are really,really good. Who knew?
some dude
12/27/2010
MVP doesn’t mean most wins produced or points created even “best player.” Wade is 5th in WP on the season, so how valuable is James to the team in relation to say, Chris Paul to New Orleans?
The problem I have with the analysis, besides that, is that if Love was first, he’d be the MVP on a sub .300 team. You’re not that valuable if you can’t win games, either!
It’s a balance between winning and contribution and context. Wade and Lebron cancel each other out. You could argue that Kobe and Pau do the same as do Rondo and KG to an extent. Dwight and CP3 and Dirk are top 3 IMO.
I’d argue it has to be Dirk, so far. Have you seen his +/- splits? His team is +15 when on and -17 when off. I know the issues with +/-, but that is a ridiculous split. The team essentially is elite with him and cannot function without him.
And when I watch Dallas play, this is exactly what I see. The team is horrible every time it isn’t run through him.
arturogalletti
12/27/2010
sd,
As I said, I’m keeping it simple (right now) but I agree with you on love (and i’ll get to it can’t spend all my bullets at once). That said the Heat are playing at a ridiculous level right now. I wished the numbers showed differently but they really don’t. I have a tough time seeing anyone other than a fully healthy Celts team with homecourt beating them. As a number 1 seed, they’d be really frightening.
I have a feeling, some of my preseason predictions are going to very quickly approach reality.
some dude
12/27/2010
The Heat’s numbers don’t mean Lebron or Wade has to win MVP. it just means they’re the best team. I mean, if you want to define it as the single best player in the league, then sure. But I don’t believe that and that’s not usually how it’s used. And the “value” aspect is unique to everyone. You could say Lebron produced the most wins, so he’s the most valuable. But then I can shoot back that he’s only x% above the 2nd best player on the team while CP3 is (X+y)% greater. I mean, there’s so many ways to do it and everyone has their own way of doing it (there’s no universal way). Right now I believe the 3 I meantioned mean more to their team through the 1st third of the season than either Lebron or Wade (now, if it was the MVP combo…).
Regarding the Heat, I don’t agree at all. I think people are falling into the old Cavs issues again. #1 team by most metrics, but basketball people could see the obvious flaws. Take the win against the Lakers. Ignoring the fact that Bynum still hasn’t come back, yet, Miami made quite a lot of outside shots. Lebron started 5/5 from 3. Big Z shot what, 4/5 from long 2s? They took a ton of long 2s and made them. Through the first 3 quarters, I thought L.A. did a very good job keeping them out of the paint. Now, Miami’s defense, especially in the 2nd half was quite good, but nothing about their overall play made me think they’d win in the postseason when the officiating and style changes. They won’t shoot that good from long 2 (their 1st 3 quarters was really high) over the course of a series and it’s a recipe for disaster.
Then you go back to the Dallas Mavericks game and yet again, they can’t do anything at the end of a game besides get to the FT line. Their end of a close game offense is horrific. Just like Cleveland (it’s largely the same offense!). And I also don’t believe guys like Chalmers will perform in the playoffs. Call it the Mo Williams corrolary if you will.
Until i see a change in their offensive strategy and synergy, I won’t believe it til I see it.
One more thing I’d like to comment on. I don’t agree at all on the HCA thing. In fact, I think they’re better on the road. I know it sounds weird, but they are way more focused on the road, IMO. The booing gets them pumped up. Have you noticed how lackluster they look at home? it’s because they have the WORST crowd I have ever seen for a title contender. Really, it’s a shame such a terrible sports city has this team. They crowd doesn’t what to ever do, hardly cheers properly, etc, and it does translate to the team. I’m not suggesting they want to play in Boston, but outside of maybe Boston, I think the Heat would be better off not playing at home.
John
12/28/2010
I think your MVP argument is quite ridiculous. So we should penalize LeBron because he plays with another superstar? Should Jordan not have won any MVPs merely because he played along Scottie Pippen for almost his entire career? Does Shaq not deserve his because he played with Kobe? Duncan because of Robinson? Malone because of Stockton? What about Magic Johnson? or Larry Bird?
All of these MVP winners played along with at least 1 other superstar who also put up amazing numbers. Why should we discount their great seasons just because another player on their team also had a great season? Or just because if we take them off the team, their team wouldn’t be decimated? It’s silly.
The award is Most Valuable Player. It doesn’t specify Most Valuable Player to his team. You’re right in saying that everybody has a different definition of “value,” but looking back at history it’s incorrect to say that most people don’t view the MVP award as the best player in the league. Over the past 25-30 years, more MVPs have been chosen based on best player in league (Duncan, LeBron, Kobe, Shaq, Malone, Jordan, Magic, Bird, etc) over most valuable to their team (Nash, Garnett).
some dude
12/28/2010
mighty fine straw man you built there. All I argued was that the “best player” or “most productive” player isn’t the de facto MVP.
I argued value is tough to define and unique to everyone. But the notion that context shouldn’t play a part is ridiculous. Lebron can certainly be the MVP playing next to Wade (and I’m if not this year, at some point it will happen). I just don’t think he’s the MVP this season, yet, regardless of what WP or points created says. I think Dirk has been more valuable to his team. Anyway, the best players tend to win so they tend to be MVPs. That is kind of a duh moment.
Also, I resent the fact that KGs MVP wasn’t part of the “best player” ilk in 2004. Nash’s MVPs were the only true mistakes, anyway.
John
12/28/2010
Right, I understand that’s what you argued. That’s why I pointed out that going back through history it’s rarely been regarded as the most valuable player to his team. And I don’t think that’s the correct construct to view the MVP. The MVP award has almost exclusively gone to the person viewed as the best player in the league during that given year regardless of with whom he played. And almost never to a player who was more valuable to his team than any other player in the league to their team.
So I’m just not sure I like the context you’re trying to introduce into the debate.
Also, I put Garnett into the latter category because he was the best player in the league, but without him his team would have been horrible, which can’t be said about the others. But then I realized that I put LeBron in the Best player category. So that was inconsistent and a mistake. So we’ll move Garnett to the Best player side. Sorry about that, feel better?
some dude
12/28/2010
I think through recent NBA history it’s been more like the “one of the best player on one of the best teams.” And they try their hardest to rotate it in some sort of strange fairness doctrine. But other than Nash, you could make compelling arguments. But you are wrong about “with him he played.” How do you explain Shaq only having one MVP. Surely he should have had 2 in a row, but Kobe’s emergence knocked him down. As did the year Nash got his second and he was partnered with Wade.
But it’s wrong to base our arguments for MVP based on the media. I think we can all agree the media’s analysis is very basic and really you just need to score a lot. How else do you explain Duncan with 0 DPOY awards and Camby having 1?
MVP is not best player. MJ didn’t always win, Shaq only has 1, KG only has 1, etc. So context does matter. Again, I’m not trying to say playing with a superstar does or should knock you down automatically. I just think in the context of this season, it sure does for Lebron. The fact that the team with Wade in the game but not Lebron has been better than visa-versa impacts it. Context. And until they start beating playoff teams with regularity, Lebron is not the MVP to me. context.
Grith
12/27/2010
One question about home-court advantage that’s been bothering me. Since Assists are somewhat subjective, have you taken local scoring tendencies into consideration? For instance, I’ve seen it mentioned a couple times that Phoenix is hard to get an assist in. If it’s hard there, somewhere else it must be easy.
nerdnumbers
12/27/2010
Point of clarification Arturo,
If you use the Nerd Numbered powered seasons then Magic also breaks 300 wins 🙂 I will not back down on the Magic > Charles debate!
Also looking forward to what we see about reliability of some of the defensive aspects. Keep up the awesome work.
arturogalletti
12/27/2010
For the first baseline, I ran the numbers for 2009-10 vs the 2010-11 numbers and got a 38% correlation. About what I expected (I know I need to do some more work on this).
Guy
12/27/2010
Arturo, I don’t think you are recognizing the enormity of your own discovery here. The opposing players’ AdjP48 isn’t just a little “extra” defensive information. It can tell you whether WP48 is working in the first place. And at least for the big rebounders, the answer is a very clear “no.” Love is supposed to be generating an extra 9.7 rebounds per 48 for his team. But he clearly isn’t taking them from opposing PFs, who are actually much more productive than average. What is your theory — that Love is a rebounding monster who somehow takes rebounds only from opposing guards? Isn’t it time to concede the obvious here?
And even with a guy like Evans, who posts a positive WPopp, the disparity is between what oldWP predicts and the reality revealed by oppADJP48 is enormous: he is generating +9.9 rebounds per 48 (allegedly) but it looks like opposing PFs are only -2. Where are his other 8 rebounds coming from?
Simply taking an average of the two WP’s and calling that a rating, when you’ve just demonstrated conclusively that the metric wasn’t working in the first place, isn’t a very satisfactory solution.
bduran
12/28/2010
You don’t think that a good rebounding PF only takes rebounds from the opposing PF do you? He takes them from everyone. His team, along with the opponents bigs and guards. I’m not sure how much you can glean from this with respect to rebounds, especially since opp AdjP48 oppReb48.
Guy
12/28/2010
No, I don’t think a PF would take every rebound from same-position opponents. But I would expect maybe 40%-45% to come from the opponent, on average. At a minimum, 75% should come from opposing bigs (their normal proportion). But what we see here is much, much less. For Evans it looks like maybe 20% (assuming ALL of the opposing PFs’ diminished AdjP48 stems from rebounding). And look at the others: for guys like Camby, Wallace, and Love we see virtually no evidence of an impact on opposing position players.
And yes, oppAdjP48 is not oppReb48. But to calculate the former, Arturo must have the latter. So if the oppReb48 numbers somehow vindicate oldWP48, I’m quite sure he will let us know.
C’mon, bduran, you must see the writing on the wall here…..
dm
12/28/2010
somedude I think it’s about you read chapter one of Stumbling on Wins that deals with human irrationality it cites many interesting studies on human “experts” being inferior to statistical models. I don’t if it’s intentional but you make it sound like your “observations” are infallible.
Just because a stat doesn’t fit your pre-existing opinion doesn’t mean it’s worthless, that’s not how science works. Time for some humble pie your eye ball test is neither interesting nor a compelling argument.
And to Arturo I hope that you continue to try to examine and advance wins produced from every angle objectively I honestly feel the direction of this blog is being dominated by the two loud mouths (somedude and Guy) and their “gut feelings”.
some dude
12/28/2010
The problem is that there is irrationality built in to the idea that there is no OVB or that the assumptions are not flawed. Arturo’s own work in his last (real big) post demonstrates a big flaw in WP48. Something many other statistics gurus has been trying to point out for a long time. And other statisticians use other metrics with differing results.
What you’re really saying is you want me to base everything on the formulas of one person even though not everyone agrees with those formulas, issues seem to be quite prevalent with the formulas, and the formulas are based on incomplete data (even the formula maker agrees that defense is accurately tabulated into the formula and then there are issues like charges taken, shots challenged, good screens, etc not accounted for whatsoever).
My observations are not infallible. I’ve said on here I’ve been wrong (cough, Kings, cough). And if you want to argue that Lebron is the MVP, and you have multiple statistical analyses to back it up, I’ll definitely listen. And then I can fire back with my own. And we can throw some observation in there too. There is no single basketball formula yet, so we all have to make mental adjustments to the data sets. Anyone claiming pure observation or pure statistics is clearly doing it wrong. You must combine the two (and yes, stats should have more weight).
The funny part is that comments like yours are no different than the “human” experts being irrational. And yes, most so-called basketball “experts” hardly are (and neither am I). And I don’t use an eyeball test. I can give other stats to back up any argument i generally have or can at least sit down and point out on screen to what I’m referring.
As for being a “loud mouth,” I take it as a compliment. It’s much better than being a sheep. Nothing ever gets done if no one is around to critique. I am here because I want to find the best statistical analysis possible at the individual and team level. I respect Arturo a heck of a lot for the hard work he puts in on his free time for “fun,” that he tries to address our questions, and that he is very cool about everything and doesn’t silence detractors and criticism as I have been censored on Berri’s blog (with no response from him in 2 weeks after doing what he asked. sigh).
Neal Frazier
12/28/2010
OK, I am going to dive in with an ‘eyeball’ thing I noticed long ago that may square the circle of Some Dude’s feelings about rebounds being over-rated. A team with 2 really solid rebounding big men can leak its guards and sf out for fastbreaks confident in the belief that they will not give up offenseive rebounds/putbacks to their opponents. Those great rebounding big men create fast break and transition opportunities for their guards. Teams that lack solid rebounding big men have to send all 5 guys to the defensive glass and end up with no fast break or transition points. So these great rebounding bigs are taking rebounds from their guards but are giving their guards easy scoring opportunities in the process. On the other end of the court those great rebounding bigs not only create extra possessions with their offensive rebounds, but they also cause the opponents to abandon their fast break and transition offense because they have to send all 5 players to the defensive glass rather than having their guards leak out. The regression analysis captures this and gives the credit for the easy buckets to the rebounding bigs who free up their guards to run and credit for their opponent’s lack of easy buckets because they had to abandon the fast break.
Thus those rebounds can be taken from team-mates while simultaneously creating value for the team by freeing up those team-mates for other valuable activities.
Guy
12/28/2010
Neal: This is an interesting theory, but I’m afraid it is incorrect. If the big rebounders were improving teammates’ shooting efficiency, as you suggest, WP48 will capture that in those players’ shooting stats. And opponent efficiency is also captured by the team defensive adjustment. All those payoffs, if they exist, are already captured in the metric. So there can’t be a “hidden” benefit that just happens to offset the failure to deliver the promised rebounds, or else WP48 couldn’t successfully predict team wins.
This “black box” theory of Wins Produced — players don’t actually produce the rebounds they are supposed to, but this is somehow offset by a previously-undiscovered contribution elsewhere on the court — has actually been proposed by many others. But never, interestingly, by Dr. Berri. The WP regression “works” because the sum of player Reb48s equals team Reb48, and WP credits rebounders with the value of all those possessions (while giving zero overall credit to shooters), not because high-reb48 players improve the performance of their teammates. Indeed, Dr. Berri believes that good players REDUCE the productivity of their teammates — just the opposite of your theory.
nerdnumbers
12/28/2010
Guy,
Point of clarification (new favorite phrase on this thread!), the term Black Box when used in modeling refers to a model that has inputs and outputs and that the inner workings are “hidden” from others. For example in Neural Networks there is a hidden layer that is trained and is responsible for the logic but may not be known by the user.
The Wins Produced metric was made via regression, the terms and combinations are known and exposed (http://www.stumblingonwins.com/CalculatingWinsProduced.html). So this model doesn’t count as black box. . . at all. Now your claims as to why things are wrong actually do count as a block box. The more you know.
Guy
12/28/2010
Nerd: I couldn’t have been more clear in saying that Neal’s interpretation of WP was a “black box,” because it says that WP is wrong about rebounds but for somewhat mysterious — and completely unproven — reasons manages to be correct anyway. I even went out of my way to point out that Dr. Berri himself does not subscribe to this interpretation. I never said, nor do I believe, that WP is a black box.
I’m sure it’s upsetting to you to have Arturo (of all people) drive the final nails into WP’s coffin. But please don’t take out your frustration on me….
some dude
12/28/2010
i would love to see this regression analysis.
Two of the fastest paced teams, Phx and GSW over the years, have been 2 of the worst rebounding teams…
Guy
12/28/2010
It’s worth noting that Arturo’s data also demonstrates quite conclusively that that “bad rebounders” rated so poorly by oldWP are in fact much better players (or much less bad) than Wins Produced claims. Look at the bottom of the table. Bargnani holds opposing players to below-average productivity. How could that be possible if he were really giving away 4.7 reb per 48 minutes? Or look at Brooks Lopez: he is allegedly giving away 3.9 reb/48, but opposing centers post an average productivity — why aren’t these centers grabbing a ton of rebounds against the hapless Lopez?
Lopez is a great example of how following oldWP can lead you into blind alleys. Prof. Berri has a post up now discussing how worthless Lopez is and how the Nets should be happy to get rid of a player who has put up league-average WS48 at center from age 20-22. This is based almost entirely on Lopez’s low rebound total (his low reb48 accounts for 86% of the difference between Lopez’ WP48 and an average center). But New Jersey is an average rebounding team overall, and Lopez’s opposing centers are no better than average. Where is the evidence that Lopez is losing a huge number of games? There just isn’t any. A broken metric will often give you bad analysis.
bduran
12/28/2010
The other bigs on the Nets have historically been decent to good rebounders. This year they are still decent to good rebounders, yet the Nets are a slightly below average rebounding team. Lopez has been a very poor rebounder this season. I think maybe there’s a link.
Guy
12/28/2010
bduran: If Lopez is losing 4 rebounds a game, why isn’t NJ a terrible rebounding team? Why aren’t they a -4 reb48 team? Or at least a -2 reb48 team? And please don’t say a few “great rebounders” are balancing the scales, because that’s a confirmation of diminishing returns, not a rebuttal. That’s the whole point of DR: every “bad rebounder” ends up paired with some good rebounders, while every “great rebounder” is matched with some very bad rebounders. Find me some exceptions, and we’ll talk. (You too, flowers.)
bduran
12/28/2010
No ones arguing against DR. The size of the effect maybe, but not it’s existence. Your post seems to state that Lopez’s poor rebounding isn’t a cause for concern. I think it is. Sure, his team mates maybe picking up some of the slack, but not enough to make them an above average team, so I believe that his poor rebounding is hurting the teams rebounding.
You also ignore the rest of what Berri said about Lopez, that in addition to being a poor rebounder, he’s a poor scorer. This year his FGAs are up, but his TS% and EFG% are way down. According to Dr. Berri’s post he’s also below average with respect to the other box score stats. So I think it’s safe to say that he’s playing below avearge and is costing his team games.
Guy
12/28/2010
bduran: You like to move the goalposts on me. WP and Berri are saying Lopez is nearly the worst player in the NBA, and his team would be better off trading him for a bag of balls. That’s the view I was disagreeing with. Do you agree with WP/Berri, or do you have a different view? I never said that Lopez’s low rebound total wasn’t “a cause for concern,” or claimed it had no effect on his team — “diminishing” is not a synonym for “zero.” Maybe Lopez does cost his team 1 rebound a game or so. But the difference between costing your team 1 rebound and costing them 4 rebounds is huge (about 8 wins over a season). So if you’re point is simply that Lopez is costing his team “something,” that he is at least a tiny bit below average, I won’t argue with that.
The reason that WP/Berri considers Lopez a totally useless player is clearly his low reb48. Yes, Lopez’ efficiency is down this year, likely related to his big increase in usage. But if he were average in Reb48, he’d be nearly average in WP even with his poor shooting this season, and clearly a player with the potential to be above-average.
And to say “everyone agrees on DR” seems rather beside the point, when Prof. Berri (and perhaps you?) believes they are relatively “small” and that accounting for them makes “almost no difference” in players’ ratings, while others believe they are extremely large (at least 70% on drebs) and that accounting for them would yield very different player ratings. Again, where do you stand? I notice you like to pop in with criticisms, but never quite take a position yourself.
Gabe
12/28/2010
“I notice you like to pop in with criticisms, but never quite take a position yourself.”
…
bduran
12/29/2010
Interestingly enough, I agree with some of what you say. I am concerned about the rebounding issue and have started familiarizing myself with R in order to try to answer some of the questions for myself. The thing is, I often disagree with how you make your arguments.
For example, I disagree that the Lopez evidence your represented supports this statement
” Where is the evidence that Lopez is losing a huge number of games? There just isn’t any. A broken metric will often give you bad analysis.”
I don’t know about huge, but Lopez is below average in most everything. He’s also a very poor rebounder on a slightly below average rebounding team so I don’t think DRs is the issue. I agree with Dr. Berri’s assessment that Lopez has been playing poorly this year.
Here’s another example.
“Maybe Lopez does cost his team 1 rebound a game or so.”
Yes, maybe.
Guy
12/29/2010
bduran: You say “He’s also a very poor rebounder on a slightly below average rebounding team so I don’t think DRs is the issue.” If a team having a “terrible rebounder” at center yet being average in rebounding does not suggest diminishing returns to you, what would? This seems like an excellent example. What do you think diminishing returns would look like?
I guess we’ll just have to disagree about each others’ style of making arguments. For example, here you preface your rebuttal with the hedge “I don’t know about huge.” But “huge” is everything, it’s what I was talking about — WP/Berri is saying Lopez is one of the worst players in the NBA, and has virtually no value. That conclusion is based mainly on his allegedly terrible rebounding, but Arturo’s data appears to show opposing centers aren’t grabbing a lot of rebounds (which was the point of my comment). Having a bad 30 games and being one of the league’s worst players are not the same thing. Losing 1 rebound is not the same as losing 4 rebounds (the difference is the total productivity of an average NBA player). Scale matters.
Similarly, when I pointed out Arturo’s discovery that opposing players of great rebounders often have average or better productivity, you responded by pointing out that every single rebound won’t be taken from same-position opponent. Well, duh! But surely a significant proportion will. Yet Arturo’s data shows that the big rebounders are taking, at most, a very small fraction of their credited rebounds from opposing position players — Kevin Love appears to have no impact at all. Instead of discussing that evidence, you chose to rebut a claim I never made.
I think the discussion will work better if we all address the arguments that people actually make, rather than creating strawmen and then knocking them down.
bduran
12/29/2010
What does “huge” mean? That was my point. It’s not very clear. He clearly producing much less than an average center across the board. I think that’s “huge”. Do to DR he may not be costing his team as much as WP suggests, but how much less? Still could be “huge”
“If a team having a “terrible rebounder” at center yet being average in rebounding does not suggest diminishing returns to you, what would? ”
like I said, I believe in DR. You keep saying people don’t, when everyone says they do. What I meant was, that doesn’t seem to be Lopez’s issue. Certainly it looks like Humphries is picking up some, but not all, of the slack.
“Similarly, when I pointed out Arturo’s discovery that opposing players of great rebounders often have average or better productivity”
You pointed out lots of anecdotal evidence. Also, the evidence you used was “productivity” not rebounds and you compared position to position. They don’t operate in isolation and productivy is not just rebounds and your evidence is anecdotal. The plural of anecdote is not data.
You may be right in everything that you say, please use data to show it.
Guy
12/29/2010
I fear we may need to just agree to disagree. If feels to me like we speak different languages (which isn’t to say yours is “wrong,” but it makes communication hard). You say “Humphries is picking up the slack” and I say “diminishing returns.” The whole point of DR is that SOMEONE ALWAYS PICKS UP THE SLACK — the way to show DR is relatively small, rather than large, is to demonstrate there are many exceptions. And why in the world would you claim that I “keep saying people don’t [believe in DR] when everyone says they do,” when you know I’ve acknowledged this many times (including this thread) but nonetheless believe that one reaches fundamentally different conclusions about player productivity if you accept the Berri/WP theory that DR for rebounds are “small” and “make little difference,” as opposed to the consensus in the analytical community that they are very large. Unless you think this is a trivial distinction, I don’t get your point.
On Arturo’s oppAdjP48 data, you are correct I have not done a systematic analysis. If/when he posts his oppReb48 data, maybe I will. But sometimes you can see the answer just by looking at the data. We already know two things: 1) the variance in opponent AdjP48 is very small, and 2) Reb48 produces most of the variance in WP48. So it will be shocking, virtually impossible, to learn that the variation in oppReb48 is large. If you want to wait to see the rebounding data, that’s fine (but I’m afraid I’ve already spoiled the ending for you). I would just point out that the evidentiary burden lies just as much with Berri, Arturo, and anyone else who believes in “small diminishing returns,” as it does with those who believe in large DR. Whatever one’s opinion of the studies showing large DR, it remains true after all these years that neither Berri nor anyone else has presented one shred of evidence for the theory of small DR on rebounds. So I hope I will hear your voice calling for rigorous statistical evidence, and not just anecdotes, for that theory as well.
ilikeflowers
12/28/2010
Guy: This model is broken because it overvalues rebounds.
Me: That’s a myth.
Guy: Here’s some regressions about apples.
Me: We’re talking about oranges.
Guy: Here’s some anecdotes about apple pie.
Me: We’re talking about oranges.
Guy: Here’s some regressions about apples.
Me: We’re talking about oranges.
Guy: I reject your oranges and substitute it for my apples.
Me: This is a waste of time, I’m outta here.
Guy: regression…anecdote…apple strudel.
Guy: anecdote…anecdote…apple tart.
.
.
.
50 Guy posts later…
Guy: These apples are delicious.
Guy: The silence is deafening!
Guy: Well, looks like we’ve reached a consensus here about the model being broken. How ya like them apples!
.
.
.
weeks later…
Guy: This model is broken because it overvalues rebounds.
Me: Must…resist…provoking…more…apple…talk…can…not…resist!!!! [click].
Gabe
12/28/2010
ilikeflowers:
+1
some dude
12/28/2010
it makes sense you’d use fruit in your metaphor, giving that Guy’s been feeding a troll.
Jon
12/28/2010
Guy, clearly NJ is an average rebounding team because Kris Humphries is an MVP candidate! Don’t let his greatness blind you from Lopez’s feebleness.
Guy
12/28/2010
Jon: my bad.
greyline
12/28/2010
A question about the box score (and also a measure of my ignorance):
Looking at the box score (in the Star Tribune) for yesterday’s Clippers-Kings tilt, and it lists something called “team rebounds.” The Clippers had 11 and the Kings had 8. What exactly is a “team rebound?” Is it when the ball bounces off the rim and out-of-bounds? Is it the non-rebound that occurs after a missed free-throw (when there’s more than one being awarded)? Both? Neither?
some dude
12/28/2010
Team rebounds are rebounds that go out of bounds and over the back fouls. Can be a blocked shot, a tip out, bounce off the rim out, etc. I think, but may be wrong, that even last second heaves where the time expires is also credited with a rebound.
basically, it’s a rebound that no single player is given credit for.
EvanZ
12/28/2010
I am now capturing these in my model., interestingly enough. I split the credit evenly among the teammates.
Alvy
12/28/2010
I think its rebounds that are not contributed to a single player. It’s a way of accounting essentially. And yeah, its like a ball hitting the rim and going out of bounds.
greyline
12/28/2010
Thanks for clarifying–I thought it was just an accounting thing, but couldn’t quite reconcile why it didn’t match the total of missed shots.
Follow up question, when we talk about teams out-rebounding one another, do we use team rebounds + rebounds for each player, or just rebounds for each player?
Bill Gish
12/29/2010
While we’re bringing up box score questions. I have never heard anyone address the advantage that bigs have because of how players lineup to rebound missed free throws.
I am six feet tall and 65 years old but I know enough about boxing out and have big enough butt that I could snag a rebound or two if you always let me be the guy who lined up next to the hoop on free throw attempts.
Does this make the rebounding by someone like Landry Fields even more impressive?