My latest for The Huffington Post asks whether or not Carmelo Anthony is worth the money the Nuggets are currently offering. Readers of this forum – as the following few posts illustrate – have seen the argument that Melo is overrated before.
Kevin Durant vs. Carmelo Anthony
Melo, King James, and the Human Highlight Film
The latest “Melo is overrated” argument begins with a column written by Tom Haberstroh at ESPN.com. The column requires insider access, so everyone may not get to read what I think is a great offering from Haberstroh.
What’s so great about this story? As I note at Huffington Post, Haberstroh argues that Melo is overrated because his scoring efficiency is not exceptional — and beyond taking shots — Melo doesn’t do much more than an average small forward with respect to any other aspect of the game.
What’s interesting about this argument is that it is essentially the Wins Produced story. Wins Produced is a single number that estimates a player’s impact on wins. Although having a single number is useful, people tend to focus too much on the number and less on the story the number is telling. And that is why the Haberstroh story is so good. Wins Produced argues that players should be evaluated in terms of shooting efficiency and the non-scoring factors that impact outcomes. Furthermore, players should be evaluated relative to the average performance at the player’s position. Haberstroh makes all these points in the discussion of Melo.
So hopefully this will help more and more people understand why focusing on just scoring – or scoring based metrics like PER – leads to incorrect conclusions. Again – as Haberstroh demonstrates — you don’t need Wins Produced to see this point (although I still think it is useful to have one number to summarize a player’s value).
– DJ
P.S. The Wins Produced numbers offered for Anthony, LeBron James, and Kevin Durant at The Huffington Post are just for what these players did at small forward. James and Anthony probably played some power forward, so the Wins Produced numbers you see at Huffington Post are higher than what has been reported previously.
Wages of Wins Journal 
robbieomalley
August 9, 2010
Have you ever watched basketball? I doubt it. If you had you would know that Melo is one of the best players in the NBA. What part of 28.2 PPG do you not understand? Of course LeBron James and Kevin Durant are technically better, they averaged more points.
I’m going to cite all the tough shots he takes and how the defense has all five men guarding only Carmelo. I am also going to ignore the fact that James and Durant play in the same role as Anthony and face the same kind of defensive attention and are still very productive. I am also going to ignore the fact that the Carmelo Anthony has better team mates than either James or Durant.
some dude
August 10, 2010
I agree that Anthony is overrated. I agree he’s not worth the max in basketball terms, though perhaps in investment terms he is. I also believe his defense is mediocre for his talent level and while he’s a good interior passer, he doesn’t really do anything else well besides score.
But to be fair, PER doesn’t say he’s a max player, either. His last 2 seasons are 22 and 19 in PER. Far from a max player, especially when considering defense. I’m no advocate of PER by any means, but if you’re going to take a shot at a colleague, at least base it on actual facts.
To correct Robbie up there, it’s unfair to say
A. he has better teammates. This is debatable (billups and nene might be good, but everyone else is weak).
B. Teams don’t defend Melo or Durant the same as James. Teams respect their mid-range game. I’m not arguing Melo faces tougher defenses, but to characterize them as playing in the same role (Durant curls off screens and off the ball way more while Lebron is way more a ball dominant player) or are defended in the same manner is incorrect.
But Melo is definitely worse than those 2.
some dude
August 10, 2010
I’d also love to know why people keep referencing Kobe’s stats during the season. It was evident to everyone with eyes that he was playing hurt, especially with the finger. Almost any other player would have taken time off.
His FG% dropped drastically from like 49% until New Years to like 44%.
If anything, his FT% is the best indicator. a career 85% shooter, shooting that rate until New Years, suddenly shoots like 78% the rest of the year. That could not happen without an injury.
I’m not trying to be a Kobe apologist. Let me get to my point.
His season numbers post New Years didn’t matter. It dragged down everything. his WP48 pre and post New years are vastly different. Same with his PER.
His playoff PER was nearly 25 and his WP48 was over .2 again. The turning point was getting his knee drained.
All I’m saying is stop using last season’s numbers to prove anything related to Kobe or other players in comparison to Kobe. His playoff numbers are far more indicative of his true level of production. You wouldn’t even compare Melo to Kobe as you did in the article if you did it properly.
robbieomalley
August 10, 2010
Somedude,
This is one of the more annoying aspects of writing on the internet. You can say one thing and it can be misinterpreted really easily. Because this isn’t a true conversation, it is difficult to qualify your words.
A. I think his teammates are better. This is especially true in terms of making shots. And if we’re talking about how Anthony’s volume scoring isn’t exceptional then it’s an important point.
B. I did not say Anthony, Durant, and James are defended in the same manner. This is obviously not the case, all three play the game in different ways. I said they face the same kind of defensive attention. I think it’s fair to say that when these three are on the court that they receive the majority of the defenses attention. Also, in terms of role I meant in the very general sense of being the primary scorer or “Go-to guy.”
some dude
August 10, 2010
A. I disagree with Lebron. Durant yeah, but not Lebron. Lebron’s teammates were better. I’d much rather have a Mo/Varajeo/Jamison/Parker/West/Moon/Hickson/Shaq supporting cast than Billups/Nene/KMart/Afflalo/Lawson/Anderson/JR Smith
Billups is massively overrated, Kmart is a bonehead, JR Smith is awful and straky, and Birdman is also very overrated. I like Nene a lot, however.
Cleveland had been on of the top defensive and 3 point shooting teams for years now. Denver is mediocre on those fronts. You cannot tell me the differences in Melo and Lebron account for this or that Mike Brown is a better coach than Karl.
B. Then by your meaning of “defensive attention,” you’ve essentially said nothing at all.
MegaloArenas
August 10, 2010
http://forums.myspace.com/t/4820333.aspx?fuseaction=forums.viewthread
Full article
khandor
August 10, 2010
—————————
re: Furthermore, players should be evaluated relative to the average performance at the player’s position. -dberri
—————————
Please explain the rationale of this claim.
[e.g. the specific reason any one player should ever be evaluated relative to “the average” performance at the players’s position rather than relative to “the best performances” by other players at the same position]
Tom Mandel
August 10, 2010
Dave — I agree that WP48 being *a single number* stands in the way of acceptance, making it appear at once magical and over-simplified. Perhaps especially because the number reflects statistical operations that a reader typically can’t duplicate.
On the other hand, having the single number makes comparison so much easier!
One *problem* with the single number — which it would be a good idea to acknowledge — is that it doesn’t allow one to note areas where improvement is possible. This is especially a problem when the number is used to sum up young players (where improvement is more likely) and where the number obscures the rather small data set on which it is based — again that’s mostly true of young players.
A classic example is commentary here on John Wall. Not to mention judgment of his likely future as a player based on 4 Summer League games! I’ve been pretty amazed by the mis-use of WP48 in this way — especially by you.
Perhaps this is the key point. However useful and well-developed a metric is, it can be misused with ease. Moreover, however valid a *statistical* metric is, it will offer many false judgments of individuals (what I’ve called, in the case of WP48, “the Nick Fazekas problem”).
Chicago Tim
August 10, 2010
Tom Mandel
I’m assuming you refer to it as the Nick Fazekas problem because Fazekas did not make it in the NBA. But from what I’ve read, he also wasn’t given much chance to play in the NBA. Now he’s playing in France, but I’ve been unable to look up his stats. Do you know how he is doing? Note that Josh Childress was forced to play in Europe despite putting up great stats in the NBA. NBA teams can be wrong.
That being said, it’s also my understanding that college stats are notoriously suspect because of unequal level of competition, the young ages of the players, and other variables. So I don’t think anyone has claimed that WP48 can predict the best player to draft just by looking at college stats.
Tommy_Grand
August 10, 2010
1. Josh Childress was FORCED to play in Europe?
2. The professor wrote: “And Durant is not quite the same kind of passer as LeBron.”
Quite an understatement. I love Durant, but KD’s passing stats do not hold a candle to Lebron’s. The reason Lebron rises above every other star SF in basketball is James’ ability and willingness to dish out assists. In addition to scoring 25+ points per game (usually shooting better than 50% from the field) it’s normal for James to facilitate 13-14 easy shots (uncontested jumpers, lay ups, & dunks) for his teamates. Durant does this maybe 5-6 times per game – which is good but not stellar.
nerdnumbers
August 10, 2010
Tommy Mandel and Grand,
First off I agree Wp48 as a single number makes life easier. I also like seeing what players do well so I split it into three numbers to see what players excel at. For instance you point out Durant is not the passer Lebron is. You are absolutely right:
http://nerdnumbers.wordpress.com/2010/08/08/mvp-race-preseason/
It boils down like this: Lebron good at everything. Durant good at Scoring, Rebounding but bad at passing. Melo is ok at scoring and bad at everything else.
Tom Mandel
August 10, 2010
Chicago Tim (I’m from Chicago too, btw, though I don’t live there. So this is from CT to CT):
You point out that “college stats are notoriously suspect”, but that’s my point rather than being an argument against my point. Or, really, it’s not that the stats are suspect — they are what they are — but rather *what they mean.* Obviously, the larger the data set (therefore the older the player) the more they tell you about what the player’s future will be.
The problem comes when 1) judgments are made about players based on extremely small data sets, 2) a single number (i.e. PAWS40 in the case of college players) obscures what’s in the stats (e.g. a very young player has one notable flaw that can perhaps be corrected: TOs in the case of John Wall), and 3) the accuracy of a measure *statistically* (its predictions across a large number of players) is mistaken for accuracy in an *individual* case.
There is no question that people use PAWS40/WP48 in these problematic ways — I can point to many instances, including from Dave. Please note, however, that I’m neither complaining about this exactly (it happens to pretty much every useful idea!) nor am I suggesting that it is evidence *against* the correctness or utility of PAWS40/WP48 (it isn’t!).
In fact, I’d be willing to bet that if you mapped productivity of NBA players drafted as juniors and seniors against their combined PAWS40 as juniors and seniors vs. where they were picked in the draft, their PAWS40 would be *substantially* more correlated to performance than draft position. But this is *very different* from looking at e.g. John Wall’s PAWS40 as a college freshman and predicting his future.
Nick Fazekas didn’t fail in the NBA because GMs and coaches weren’t able to tell that he was good. If you believe that, then you must also believe that Bryan Zoubek, who had the 6th highest PAWS40 of any NCAA player in the 2010 draft, should have been picked 6th. Or, better, you have to explain why you *don’t* think that.
Tom Mandel
August 10, 2010
Khandor, do you *really* fail to understand why one ought to compare a player’s performance to the average? Doing so tells you whether the player is above average, average or below average, and exactly where in respect of average the player stands.
Perhaps — even though people typically and usefully make such comparisons in virtually every field of life — you don’t understand why one would want to know where a player stands in respect of average? For example, you’d like to know how your car’s reliability compares to the most reliable car in the world rather than how it compares to average reliability?
I do note that Dave and commenters here are extraordinarily polite to you; for that reason I’ll stop right here.
Jon
August 10, 2010
Could someone point me in the direction of analysis that shows that Wins Produced or W/48 is more correlated to actual wins than PER? I’ve been reading this site alot lately and it talks alot about how “correct” its stats are and how bad other ones are, so i’d like to see the proof. There must be an interesting analysis on this somewhere?
nerdnumbers
August 10, 2010
http://arturogalletti.wordpress.com/2010/07/20/predictive-stats-bad-metrics-correlation-in-the-nba/
I’m pretty sure the rule has become if you have a question DJ hasn’t answered in the last 5 years that Arturo has answered it or it happens to be his next post. WoW Ninja!
Italian Stallion
August 10, 2010
Since I didn’t contribute to any of the Kobe debate I feel compelled to say something controversial.
I think Melo has the basketball talent to be a much more productive player than he has been for Denver. If he were traded to the Knicks for example, I think coach D’Antoni would reduce his usage, get him better looks, and increase his efficiency fairly significantly because there would be other good scorers on the team, plenty of space for him to operate in, and more easy fast break points (all of which would help Melo)
Chicago Tim
August 10, 2010
Tom Mandel
Prof. Berri is usually pretty careful to qualify his judgments of college players — indeed, it is usually a guest columnist who writes about the college draft. But in his books, Prof. Berri and his coauthors also note that doing well in the NCAA tournament and getting picked high in the draft can benefit a player in terms of playing time and salary throughout his career, long after one would think he has proven to be a bust. Similarly, players for small schools who don’t boast a high scoring average could be victims of prejudgment throughout their professional careers, and may never get that golden opportunity to show their abilities. So without picking any specific names, let’s just say that I reserve judgment on any professional basketball player under the age of 23, including those who have, so far, failed to find a job in the NBA, and especially if they seemed to play well in college. They might be victims of irrational prejudgment. Or they might suck. I don’t know.
So getting back to the names you mention, you say:
“Nick Fazekas didn’t fail in the NBA because GMs and coaches weren’t able to tell that he was good. If you believe that, then you must also believe that Bryan Zoubek, who had the 6th highest PAWS40 of any NCAA player in the 2010 draft, should have been picked 6th. Or, better, you have to explain why you *don’t* think that.”
How can you be sure that Fazekas and Zoubek are not as good as their college stats would indicate? Surely it isn’t because you trust the judgments of NBA GMs. I’m not the one claiming that they are great; you are the one claiming that they are clearly bad. Why do you think that, and how can you be sure?
nerdnumbers
August 10, 2010
Italian,
Something Arturo mentioned to me and something I’ve felt is that Melo doesn’t live up to his potential. Mosi and DJ have shown that Melo has played very well in the Olympics and last year he did quite well in the playoffs. That said at this point I just don’t buy that he’d change enough in any system to be worth the money. It’s a shame because a 6-8 guy with his size and skill should be a force (see Lebron James) and not just a middle of the road SF. That said if Dantoni (who wasn’t shown to increase player productivity) can do it and the Nuggs can get some cap space and maybe picks in S+T I’m all for it.
robbieomalley
August 10, 2010
Somedude,
How am I saying nothing at all? It seems like I’m talking about one thing and you want me to note on something else. Sorry for not doing that. I thought this concept was pretty simple but I’ll have to say it again.
We’re talking about Carmelo Anthony not being as productive as commonly thought. This is in part because he is not an exceptionally efficient scorer. A common defense of high volume, low efficiency scorers is that they have all the defenses attention thus making their shots more difficult than average players. What I tried to say earlier is that, if this argument if brought up, you have to explain why other players like LeBron James and Kevin Durant face roughly the same level of defensive attention and are still very efficient scorers.
As an analogy if this were a Wins Produced topic, what I’m talking about is three players produce roughly the same amount of wins. What I’m saying is that three players have the same WP48/Wins Produced. What you seem to want to know is exactly how they produce those wins (Scoring, passing, rebounding, etc.). That’s not what I’m talking about. Sorry. I know the three players are defended in specifically different ways but you can’t say scoring is made more difficult by defenses for Carmelo than it is for LeBron James or Kevin Durant.
Tom Mandel
August 10, 2010
Chicago Tim —
You don’t need to defend Dave Berri from me!! I’m a *huge* fan. And a convinced believer and user (as an amateur, obviously — they don’t let me draft unfortunately!) of PAWS40. :)
Moreover, I don’t think (and didn’t claim) that either Nick Fazekas or Bryan Zoubek is “bad.” For example, I don’t think you can start for Duke as BZ did, put up a terrific PAWS40, and be bad. Far from it.
Actually, you would expect that having put up the 4th best PAWS40 in the NCAA and starting on the team that won the tournament, Zoubek would have been drafted high. In fact, he wasn’t drafted at all; he has a non-guaranteed contract w/ New Jersey, however. Hope he excels.
As far as that goes, if the Washington Wizards would listen to my lobbying they’d have drafted or at least signed post-draft Artsiom Parakhouski; he had the *3d* best PAWS40 in the NCAA, and he played pretty well for Boston in the Summer League in July. I like him — as indeed I like Zoubek (Which doesn’t mean I’d have drafted either of them high in round 1).
I was making a completely different point. It was about the use of a statistical measure to predict an individual case. As to Dave being “careful to qualify” — sometimes he is, but sometimes he’s not! I’m not trying to convict him of anything here. It’s just the wrong way to use a statistical “fact.”
Tom Mandel
August 10, 2010
Occasionally — tho not often — a guy Carmelo Anthony’s age changes teams (coaches and “systems” too) and becomes a much better player.
The most obvious example I can think of is Detlef Schrempf. He went to the Pacers when he was about to turn 27 I believe and immediately became a much better player.
So *maybe* a change of scene would help Anthony? Or maybe he should make a point of meeting Detlef Schrempf?
khandor
August 10, 2010
Tom Mandel,
Is the goal in your life to actually “win” the league championship in a closed set with 29 other competitors, which includes successfully advancing through 4 rounds of playoffs, against only the best teams in the group, in which 4 victories are needed from a series of 7 games each?
If it is, then, perhaps you … and others … will actually get further by not comparing your performance with the “average” performance of anything from this entire group of 29 but rather from a comparison with only the best performances of a highly specific sub-set.
Since, it is important to be polite, I will simply stop there.
khandor
August 10, 2010
IS,
———————————————————-
re: I think Melo has the basketball talent to be a much more productive player than he has been for Denver.
———————————————————-
Amen, brutha.
ODU
August 10, 2010
What benefit would be derived from comparing players to the best at a given position rather than average? Don’t you just end up in the same place — saying that Player X is “worse” than the best instead of “better than average?”
The bigger question I have about comparing players to the average at their position is identifying exactly what position guys are playing. Many players slide between positions, and players at the same position often have differing roles. I’m not sure what the solution is, but I remain skeptical of position adjustments when positions are not well defined.
Edmond
August 10, 2010
Khandor,
It’s funny–I don’t think there’s quite as much difference in outlook between you and the other posters here as we all tend think. I think most of us agree that on-court decisions (i.e. don’t shoot that three!) have a large impact on player productivity. The difference between Stephen Jackson and a good small forward has more to do with the *way* he plays than with the *sort of* player that he *is*.
I think many of us would agree that decision making can be improved and also that skills can be developed. I think the main difference is that I think the barriers to such change are pretty high.
I guess my question is: do you think WP is a fairly accurate measure of productivity (as opposed to “skill” or “potential”)?
Italian Stallion
August 10, 2010
ODU,
“Many players slide between positions, and players at the same position often have differing roles. ”
This is an issue I bring up from time to time.
Unless there’s an 11th commandment I never heard about, I don’t see why a 7’0″ player can’t be a very good outside shooter and above average playmaker as long as you have other players that are above average rebounders that can score in the paint etc… Yet a player like that would get penalized badly relative to the typical C that is efficient around the basket and gets a lot of rebounds.
To me, the goal is to put together a balanced team that can defend whatever the opposing team puts on the court and do all the things required for success. It doesn’t matter which position they play.
todd2
August 10, 2010
Anthony’s a tweener, he isn’t big and he isn’t fast. He’s a deluxe role player like Nowitzki. Can he make his team mates better and what does he contribute when he doesn’t have the ball?
nerdnumbers
August 10, 2010
Italian,
I’ve been posting like mad on Role Players that are position adjusted (starting here http://nerdnumbers.wordpress.com/2010/07/23/thursday-morning-musing-2010s-top-role-players/). With how the Wins Produced formula works, there is nothing penalizing your 7 Footer (Nowitzki I assume) In fact as the average Center does not pass a lot, this player may be slightly rewarded for passing as long as you pointed out that they keep up on rebounding (which Centers are expected to do)
There is something important to realize in that each stat has a value. When your player is on the floor they need to be adding value to their team. One of the hardest ways to do this is by just focusing on scoring (although Reggie Miller for instance was very successful at this) In this way players hurt themselves by not focusing on things that actually win games.
Now the last point you guys make has some merit. Players do slide between positions. However, in the Wins Produced world unless a player slides at least two position (e.g. SF to PG) for a big chunk of time, it won’t have a big influence. While I can buy arguing “Turk is really an SF not a PF”, I doubt anyone here would suddenly claim that all this time Dwyane Wade was really a Power Forward. Sure there’s the weird case (Barkley and Magic), but these are not the norm.
Italian Stallion
August 10, 2010
nerdnumbers,
I agree with a lot of what you say, but it’s demonstrable that the average TS% of Cs is slightly higher than for PGs and SGs. The disparity increases if you break players out by inside scorers and outside scorers instead of by position. So I do think that a C that plays on the perimeter would get penalized relative to one that plays inside.
In addition, there is some evidence that playing on the perimeter reduces the chances for grabbing offensive rebounds.
If you concede that good outside shooting is a requirement of a successful basketball team and you get that from your C instead of your SG, then in some strange way he needs to be compared to a combination of a C and SG because that’s what he is.
some dude
August 10, 2010
“What I tried to say earlier is that, if this argument if brought up, you have to explain why other players like LeBron James and Kevin Durant face roughly the same level of defensive attention and are still very efficient scorers.”
They receive the same defensive attention but are played defensively differently.
does not compute.
If by defensive attention you solely mean “he’s their #1 so teams take note of that,” then like I said, you said nothing at all.
You base how much defensive attention each receives based on how they are played defensively.
reservoirgod
August 10, 2010
Italian Stallion:
Bargnani sucks. Let it go.
Italian Stallion
August 10, 2010
reservoirgod,
Bargnani is an example of a misclassified player, but there are a multitude of players in the NBA that are classified to positions even though their skill is nothing like the standard for the position. I think we need more flexibility in the system to account for those exceptions.
Here’s a recent article on the subject:
http://www.basketballprospectus.com/article.php?articleid=1190
some dude
August 10, 2010
doesn’t 82games.com classify bargs as a PF or a 4?
arturogalletti
August 10, 2010
Hey all, working today. Did I miss anything? :-)
IS,
I’m going to point you to this:
http://arturogalletti.wordpress.com/2010/07/29/the-short-supply-of-tall-people-revisited/
Big Men (F/C) are on average more productive than everyone else . They in fact account for 50% of all gross productivity. If you have a 7 footer who plays and produces like a SG and takes minutes from Centers you will be at a competitive disadvantage to other teams. WP48 loves Bigs who produce like bigs but also shoot from distance (Love, Nowitski,Garnett come to mind). We saw these last three season that rebounding/size wins championships. I can’t think of a team that won it all without a >.200 big on their roster (and I looked the 91-93 bulls had Grant, 96-98 Bulls had Rodman)
nerdnumbers
August 10, 2010
IS,
Gotta agree with Arturo here. If you have a 7 footer shooting inefficiently from 3 instead of pulling down boards then you have a problem. Also for perspective on Bargnani, let’s call him a PG (which is just not true, how do you account for those tiny 6 footers who were on the floor the same time as him?), his WP48 would have been .046 (average is 0.099). So if you somehow want to claim Bargnani is really a PG, well he goes from being really bad to being. . . bad (not to mention that would mean Toronto was playing some 6 footer as a Center).
reservoirgod
August 10, 2010
Italian Stallion:
I saw that article but couldn’t stomach reading the whole thing. I thought it was rubbish. I don’t think it makes sense to come up with new positions to accommodate a few square pegs.
todd2
August 11, 2010
Re Is, “To me, the goal is to put together a balanced team that can defend whatever the opposing team puts on the court and do all the things required for success. It doesn’t matter which position they play.” Can we get an Amen from the congregation! That’s probably the gist of this site; measuring players by their productivity and tossing out traditional categories like SF, SG, etc. and physical characteristics. Billy Beane turned the baseball world upside down by tossing out typical combine evaluations and focusing on productivity: do hitters get on base and do pitchers get people out? It didn’t matter how. Successful basketball teams need some combination of players that can pass, shoot, dribble, rebound and defend. That combination can be achieved in a million different ways—the Bulls were very successful with a point-guard-by-committee approach and the Bad Boys won with a jump shooting center.
Tom Mandel
August 11, 2010
Arturo — Love shoots from the outside? I guess it’s true in a sense. He has a low overall FG% for a big (45% last year) but it gets a little bump from his making @1 three per 48 (@ 33%). He’s not even 22 yet; if he gets better as an outside shooter, he could be a HoF player!
Khandor — to *understand* a player, you compare him to the average (i.e. try to place him on the bell-shaped curve of productivity). What this has to do with setting a goal to have (or become one of) the best players is… nothing. Nor with how one goes about *acquiring* better players — except in the sense that if you *can’t compare* players to the average you never know how to acquire good players.
khandor
August 11, 2010
————————–
re: It’s funny–I don’t think there’s quite as much difference in outlook between you and the other posters here as we all tend think. – Edmond
————————–
Agreed. :-) The key differences are actually quite subtle … and, tend not to be looked at properly – at least, initially – by those with an antipathy towards disparate thinking.
————————–
re: do you think WP is a fairly accurate measure of productivity (as opposed to “skill” or “potential”)? – Edmond
————————–
WP is a fairly accurate measure of a player’s level of on-court production, related to the few “conventional” stats it actually measures. Unfortunately, however: i. All the stats are not measured; and, at least, some of those stats which are measured are not measured properly [in regard to their effect on winning/losing the league championship … which is, afterall, what the primary goal is].
————————–
re: Billy Beane turned the baseball world upside down by tossing out typical combine evaluations and focusing on productivity – todd2
————————–
Has Billy Beane’s team ever won the World Series?
Baseball and basketball are two very different games … despite a few statistical similarities.
khandor
August 11, 2010
————————–
re: Khandor — to *understand* a player, you compare him to the average (i.e. try to place him on the bell-shaped curve of productivity). What this has to do with setting a goal to have (or become one of) the best players is… nothing. Nor with how one goes about *acquiring* better players — except in the sense that if you *can’t compare* players to the average you never know how to acquire good players. – Tom Mandel
————————–
Actually … that’s not what you do at all.
When you compare a player to the average player … what you get is precisely that, i.e. a comparison between that player and an average player, only.
Conversely, if/when you compare a player to a highly specific sub-set of the best players, exclusively, what you get is precisely that, i.e. a comparison between that player and a highly specific sub-set of the best players, only.
When the primary goal is geared towards the winning of the League Championship it is crucial that a player is able to compete effectively in a match-up against the best players in the league, given their unique combinations of strengths and weaknesses.
e.g. Although “you” might want to know how “your” car compares, in terms of reliability, with the “average” car, if the primary goal is to actually have the “best” car on the road, then, it would be more helpful to know precisely that … i.e. what the best cars actually look like … from the get-go.
[PLEASE NOTE: If this line of logical reasoning fails to resonate with you, you are certainly free to thinking differently, if you wish.]
Leon
August 11, 2010
I know a bit off-topic, but I saw this on nba.com and it is refreshing to hear. Kevin Love on what he needs to do for team USA:
“For me, it’s going to be really attacking the glass on the offensive and defensive end,” Love said, “getting put-backs, getting extra possessions, knocking down the three ball, getting to the free throw line and really being that international presence at the four or five spot.”
He knows what he needs to do to be productive, his coach does as well. Why do other players (Melo being a good example) not follow this train of thought?
Italian Stallion
August 11, 2010
Guys,
This is not about Bargnani. It’s about many things.
Some SGs do most of their scoring as slashers while others do most of their scoring as outside shooters.
So SFs do most of their scoring as slashers while others do most of their scoring as outside shooters.
A good team will usually have both slashers and outside shooters.
Great slashers are extremely valuable and tend to be more efficient scorers than outside shooters because they get into to the paint and take easier shots . But you absolutely NEED outside shooting to build a winning team too. So why should the SG that slashes be considered superior to the SG that’s an outside shooter when they are filling two different required roles?
In some cases that slashing SG should be compared to the typical SF and that outside shooting SF should be compared to the typical SG.
The examples like this are endless.
In the case of Bargnani, I’m not arguing he’s a great player, defender or anything else.
I’m arguing that he’s not as bad as he is considered here.
His style of play simply requires the team to pair him with another hybrid player. Since he can drag big men away from the inside with his outside shooting, he needs to be paired with a player that’s a good offensive rebounder that can also take advantage of the space he creates – basically filling the role of the typical C but without the height and strength to defend Cs.
It’s not so hard to understand that a team needs all the basketball skills to succeed. It’s harder to understand that each of the skills doesn’t have to come from typical position. When they don’t, IMO players should be compared to other players with similar required skill sets.
Schermeister
August 11, 2010
Nick Fazekas European statst
http://www.eurobasket.com/player.asp?Cntry=USA&PlayerID=43948
And I am sick of people quoting one failure of WP48 like that defines the system. GM’s have made an amazing amount of failures per team and per draft. furthermore Fazekas was never given much of a chance. Paul Millsap played in the WAC as well, I was lucky too see both play. And they both dominated completely within that division. Millsap seems to be playing fine now.
His stats seem good. Notice the 0 turnovers
todd2
August 11, 2010
Has Billy Beane’s team ever won the World Series? For a time he got the most “bang for the buck.” A recurring theme on this site is not evaluating players by ppg/rings/titles.
marparker
August 11, 2010
Billy Beane exploited an inefficiency that was subsequently closed because he told everyone which inefficiency he was exploiting. If he were really some kind of baseball genius he would have already been exploiting the next inefficiency. Instead it seems(from admiring the A’s young home grown roster) he has an advantage at drafting players but has lost the ability to sign good veteran players for cheap.
nerdnumbers
August 11, 2010
Scherm,
Nick Fazekas had a great WP48 in his 260 minutes of play with the Clippers. His Wins Produced in 2008 were more than 5 other players on the Clips that year that got 1000+ minutes of play. The failure there is with NBA GMs. Would you rather spend a lot of money on a 30+ vet with nothing left in the tank (Boston) or spend the same money on a young tall player like Fazekas begging to get in? The Nuggets waived him, which makes me so happy given our complaints complaints on needing a big man. . .
reservoirgod
August 11, 2010
I agree w/ khandor that the object of the game is to win championships & players need to be evaluated for that purpose. With that said, evaluating players against the avg does serve that purpose because only half the league makes it to the post-season, which is a prerequisite for winning the championship. But after that, your team needs to have advantages against other teams that have above avg players. I think this is what Arturo has been trying to get at w/ his playoff review series, 30 to 16 to 1.
brgulker
August 11, 2010
Italian,
Here’s the problem I have with your line of reasoning about Bargs. You said,
What is easier to find:
A) A typical PF/C that patrols the paint, rebounds, shoots a high volume of high percentage shots, and defends the interior?
B) Bargs + another hybrid frontcourt player(s) that can make up for all the deficiencies that Bargs’ “unique” play style creates?
To me, it seems almost too obvious that it’s easier to build a successful team utilizing option A, because that type of player is easier to find.
brgulker
August 11, 2010
Forgot to sub to comments … sorry double post.
some dude
August 11, 2010
brgulker hit the nail on the head.
A 7 footer who is terrible at rebounding and help defense (and not good at man to man, either) puts your team in a tough position, not just on the floor, but off it as well. It’s very hard to find a complimentary player that will allow the team to utilize bargs skills successfully.
You don’t want your 7 footers to be your SGs. In almost every case, you can’t find SGs to be your rebounders and post defenders, while while in a 6’8 and up league Bargs might be MVP, in the actual NBA he’s a liability.
arturogalletti
August 11, 2010
Hey all,
Just put up a piece on the 4 (and a half) team trade today.
http://wp.me/pYIAy-6Y
some dude
August 11, 2010
arturo, your analysis has a flaw in it IMO.
You say NOH got much worse and a lot is based on the loss of roughly 5 wins from Collison. But CP3 is expected to take those minutes back from Collison. On Indiana, his WP projection may be 5 wins, but on New Orleans it’s probably much lower because of a lac of playing time, no? Also, I think you’re very optimistic on Posey’s WP number given his 3 year decline and age.
Evan
August 11, 2010
is it me or is Arturo rapidly surpassing DJ for WoW analysis?
Tom Mandel
August 11, 2010
Schermeister writes “I am sick of people quoting one failure of WP48 like that defines the system.”
Well, context points to me as the object of your frustration. But, I didn’t actually suggest that the case of Nick Fazekas “defines” WP48, nor even that his case constitutes a “failure of WP48.”
What I did say is that a *statistical* truth can’t be applied mechanically to any given individual player. Coming up with the WP48 methodology was an analytic achievement of no mean merit. Doing arithmetic on an individual player, on the other hand, does not constitute analysis at all. I would hope that was obvious.
Nick is playing well in France — maybe we’ll see him back in the NBA. Hope so. My point was *not* that Nick Fazekas can’t play basketball. In ’06-07, he had *the highest* PAWS40 of any 4 or 5 in the NCAA. Would you have drafted him first that year over Oden? Would you have drafted him 2d over Durant? How about 3d over Horford? Heck… how about 9th over Noah?
If not, then you can’t turn around and point at an individual with a *low* PAWS40 relative to his draft position and conclude *automatically* that he was a poor pick. Yet, over the *whole draft*, PAWS40 will do a better job than actual draft order in predicting NBA success. This is not a contradiction but a pointer to the difference between a *statistical* truth and an individual one.
John Giagnorio
August 11, 2010
The frustration with players like Fazekas is that they get no chance at all. Teams happily recycle expensive garbage players like Nazr Muhammad and Glen Davis when there are numerous 2nd round picks or undrafted players who could step in and provide similar, possibly superior, production.
John Giagnorio
August 11, 2010
Or what Andres said 12 hours before :) These marathon comment posts can be frustrating as well.
arturogalletti
August 11, 2010
Tom Mandel,
That was a brilliant comment. I’ll have to dig up my draft data set to test it out. Will PAWS/40 do a better job than draft position at predicting future performance?.
My response post to your last question on my website goes up at midnight. The draft question might be a weekend piece.
sd,
They turned two significant assets (Posey and Collison) into a bad player (who will get a bunch of Collison’s minutes) and an ok player in Ariza. This is not the way to turn that team around.
Evan,
I blush but I’m just following some pretty large and awesome footsteps.
arturogalletti
August 11, 2010
And posted. Click my name if interested.
some dude
August 12, 2010
Arturo – I agree the trade doesn’t make a lot of sense for NOH. I’m not defending it. I just think you’re being overly harsh.
You mention a bunch of minutes eaten up by Collison’ replacement. True, but a lot of those minutes will be eaten up by CP3 as well, That could end up being a wash if CP3 stays healthy.
My issue is they got rid of their post-CP3 player if he were to leave (which seems likely) and that they didn’t get enough if they were going to trade him.
I just think your WP analysis was a bit unfair. more CP3 minutes and Ariza over Posey probably means close to status quo. And Collison could be a fluke, as well.
arturogalletti
August 12, 2010
sd,
The problem is that the status quo ends with that franchise in ruins. Collison had real value as a trade piece and I gotta think Posey, as a defensive wingman with playoff chops, had some value to a contender too (Posey for Rasheed’s contract who says no?). Yes the trade is a wash i nterms of wins but they really could have done better (and Hollinger actually agrees with me which is weird).
Xavier Q
August 12, 2010
Any chance of an analysis of the 4 team trade? Just wondering since the Pacers didn’t get much love on here during the season :) :) :)
nerdnumbers
August 12, 2010
Xaviar,
Long thread so easy to miss stuff:
http://arturogalletti.wordpress.com/2010/08/11/evaluating-a-four-and-a-half-team-trade/
“I’m pretty sure the rule has become if you have a question DJ hasn’t answered in the last 5 years that Arturo has answered it or it happens to be his next post. WoW Ninja!”
arturogalletti
August 12, 2010
Xavier Q,
http://wp.me/pYIAy-6Y
Posted yesterday :-)
Xavier Q
August 12, 2010
Haha, great, thanks NerdNumbers and Arturo!
khandor
August 12, 2010
———————————–
re: The frustration with players like Fazekas is that they get no chance at all. Teams happily recycle expensive garbage players like Nazr Muhammad and Glen Davis when there are numerous 2nd round picks or undrafted players who could step in and provide similar, possibly superior, production. – John G
———————————–
One of the problems associated with a strictly stats-based assessment of a player’s authentic basketball ability is the making of comparisons like this one, re: Nick Fazekas, in a French pro league vs Nazr Mohammed and Glen Davis, in the NBA.
Although Nazr [6-10, 250] and Glen [6-9, 289] are both less-than stellar Bigs in the NBA … the best league in the world … it is simply not a sound comparison to then suggest that a player like Nick [6-11, 235] might be capable of generating similar, or superior, production numbers, in the NBA, if afforded the chance … given his reduced level of overall size, strength, and relative explosive quickness, as a Big … based on how he’s performed in that French pro league.
[NOTE: Conversely, legitimate NBA-like Bigs, e.g. Arvydas Sabonis, Antonio Davis, Rasho Nesterovic, Fabricio Oberto, Luis Scola and Tiago Splitter, who toil successfully in Europe for several seasons are a sound comparison … given their [i] specific skill-sets, [ii] physical attributes, and [iii] Basketball IQ.]
todd2
August 12, 2010
I think Ariza might be a plus for NOH. If he can fill lanes and catch and finish, his productivity will jump. Unlike a newly acquired marquee player in MIA, CP3 is a leader. He’s constantly pushing team mates during games. Ariza’s shot selection should improve and he’s a more than adequate defender. Having said that, NOH is at best treading water in a brutal conference.
Italian Stallion
August 12, 2010
brgulker/some dude,
I wish you guys would get off the Bargnani discussion because this has almost nothing to do with his specific strengths or weakness even though he’s a misclassified player.
It may be more difficult to build a team of various hybrid players, but they exist at every position in fairly large quantities. Some are very productive and others are not. It’s when they are misclassified that they either look either more or less productive than they actually are. That’s because when you use one them, it doesn’t relive the NEED of the team to fill the gaps they don’t fill.
If you find a PG/SG that’s a great rebounder and slasher his rating is going to be off the chart. But how much is he REALLY adding if you can’t put a traditional SF that’s an excellent rebounder and slasher next to him because the team NEEDS a much better outside shooter and play maker at that position to create teh req
There are probably a couple of dozen combo guards in the NBA. You now what they are. They are SGs that aren’t good enough play makers to play the point but not tall enough to defend teh average SG/.
Italian Stallion
August 12, 2010
brgulker/some dude,
I wish you guys would get off the Bargnani discussion because this has almost nothing to do with his specific strengths or weakness even though he’s a misclassified player.
It may be more difficult to build a team of various hybrid players, but they exist at every position in fairly large quantities. Some are very productive and others are not. It’s when they are misclassified that they look either more or less productive than they actually are. That’s because when you use one them, it doesn’t fill the NEEDs of the team in other required areas.
That causes a problem in measuring their productivity.
If you find a PG that’s a great rebounder and slasher his rating is going to be off the chart. But how much is he REALLY adding if you can’t put a traditional SF that’s a good rebounder and slasher next to him because the team NEEDS a much better outside shooter and play maker at that position to create the required balance.
What you are going to show is that the PG is a super freak and the SF sucks, but what you should be doing is comparing the PG to the average SF and the SF to the average PG.
Italian Stallion
August 12, 2010
Sorry about the double post. I was in the middle of typing it and hit the enter button. Please delete the first one.
Italian Stallion
August 12, 2010
Maybe this will make it clearer.
We tend to define player’s by the position they defend, but then measure defense at the team level.
Why not define players by the individual skill set they bring to the table in terms of rebounds, assists, blocks, scoring efficiency, inside or perimeter shooting etc… and measure defense at the team level.
Tom Mandel
August 12, 2010
I notice that navigation to older and newer posts is now gone, and so is the list of recent posts.
Intentional? Maybe someone who communicates w/ Dave should mention this to him and find out.
arturogalletti
August 12, 2010
Tom,
I can still see that. Did you try refreshing?
Mike
August 12, 2010
IS,
I tend to think of it this way: if the number of “hybrid/non-traditional role” players are so high, then the average production of those positions would reflect that. WP calculates marginal value added by an individual vs the average player.
Basically you are placing the burden on team management to find players that will compliment the skill sets of a non-traditional player, whereas “traditional” is defined as the average production at a particular position, but isn’t it reasonable for management to assume that players will produce at least at the average level for their position?
Tom Mandel
August 12, 2010
Wow, Khandor — that is an extraordinary insight! You mean we can’t directly compare productivity in the French League with NBA productivity? Uh oh. Hey everybody — stop doing that right away!
Actually, Nick was more productive than Glen Davis in college (by far!). He had only 260 NBA minutes, obviously a very small sample; all the same, he was more productive than Glen Davis in those minutes too (by far!).
While preening, self-important pronunciamentos are certainly a valid activity, Khandor, reading carefully and thinking may also be worth the effort. The “Nick Fazekas problem” in WP48 is *not* that the methodology says a bad player is good. Nick is a good player. The problem is with the use of statistical knowledge as if it had direct import in judging a single instance.
Tom Mandel
August 12, 2010
Arturo — it’s always nice to have something one writes deemed “brilliant”, especially by a smart fellow like you! Thanks.
I’m not sure my comment qualifies, however. It’s a simple enough truth.
To provide what I hope will be an immediately graspable example of what I mean: to say that the set of people who smoke have a larger chance of getting cancer than the set of people who don’t smoke does *not* mean that every person in the former set has a larger chance — as if the statistical truth reflected adding up each person’s greater likelihood. What it means is that more people in the former set will get cancer than in the latter set. The statistical fact itself tells us literally *nothing* about any individual’s chance of getting cancer. (But don’t start smoking, ok?)
khandor
August 12, 2010
————————–
re: The “Nick Fazekas problem” in WP48 is *not* that the methodology says a bad player is good. Nick is a good player. – Tom Mandel
————————–
I suggest that you might be the one who could benefit from reading more carefully, on accasion … e.g. I did not say that Nick F. is a “bad” player.
In fact … I did not say anything about the specific “methodology of WP48”, at all.
————————–
re: Actually, Nick was more productive than Glen Davis in college (by far!). He had only 260 NBA minutes, obviously a very small sample; all the same, he was more productive than Glen Davis in those minutes too (by far!). – Tom Mandel
————————–
Nor, did I say that Nick F. failed to generate better “productivity numbers” than Glen Davis during their respective collegiate careers.
————————–
re: Wow, Khandor — that is an extraordinary insight! You mean we can’t directly compare productivity in the French League with NBA productivity? Uh oh. Hey everybody — stop doing that right away! – Tom Mandel
————————–
That which is profound … i.e. accurate in the real world in which we all live … does not necessarily have to be ‘old barrier’ shattering, in either style or content.
What I said concerned …
Projecting accurately how a player will perform in the NBA environment, when coming from high school, college, the D-League, or a non-US based pro league, etc., is best done without relying exclusively on a player’s individual “productivity numbers”, according to any of the already well-known metrics … while simultaneously disregarding a host of other important factors, as well – e.g. relative, positional size, strength, explosive quickness, specific skill-set, personal atrributes, Basketball IQ, etc. – some of which are easier to measure than others, and some of which are exceedingly difficult to quantify accurately, in the first place.
i.e. Scouting – young and old – basketball players properly and making accurate assessments of their basketball-related talents is far closer to being “an art” than it is to being “a science,” in spite of what certain stats gurus are continually trying to tell you.
khandor
August 12, 2010
—————————
re: Maybe this will make it clearer. We tend to define player’s by the position they defend, but then measure defense at the team level. Why not define players by the individual skill set they bring to the table in terms of rebounds, assists, blocks, scoring efficiency, inside or perimeter shooting etc… and measure defense at the team level. – IS
—————————
The problem you face with presenting Bargnani as a positionally “mis-classified” player is that, defensively, the only players he can guard effectively, in the NBA, on a consistent basis … i.e. without fouling and/or getting beat for high percentage shots galore … are power-based Centers, who set up shop offensively in the Low Post.
“Defense” is measured only at the “Team Level” by those persons without the actual ability to measure it accurately at the “Individual Level”, given what they know about the way the game is played, at this level of competition, from a technical Basketball standpoint.
[e.g. Go ahead and watch a Raptors game next season and track how many baskets they give up simply because Bargnani is not able to get back in defensive transition quickly enough, in order to stop easy hoops by, either, his own check, or one of his teammate’s, who Andrea should have been able to provide “help” for, if he’d been more aware and fully committed to doing this consistently.]
Tom Mandel
August 12, 2010
Khandor, no one in his right mind suggests “relying exclusively on… productivity numbers” in assessing a player’s ability to produce in the NBA (or in college before the NBA for that matter) or “disregarding …other important factors.”
At the same time, a 4 with “positional size, strength, explosive quickness…” ought to rebound quite well. If he has those but *doesn’t* rebound, I’m likely to be less interested. As to basketball IQ, that’s not measured with a “basketball IQ test” but by your play.
Of course, scouting is an art. For that matter, statistics is not “a science” it’s a set of mathematical techniques. None of this is about science — it’s about decision-making.
What made Wilt Chamberlain great was his ability to dominate the guys he played against and deliver wins for his teams. That “ability” showed itself in numbers, the number of rebounds, points, wins he delivered. Had he not delivered those results — those numbers — his “size, strength, quickness, specific skill-set” and so forth would have been a great deal less interesting.
khandor
August 12, 2010
… and, what made Bill Russell’s accomplishments dwarf those of Wilt Chamberlain’s – in a head-to-head comparison, when you compare their contributions to their teams, as individual players – are the plethora of basketball-related skills and attributes which Mr. Russell had that are simply not captured in box score data, whether measured in either “basic” or so-called “advanced” statistical metrics.
Yes, indeed, Misters Chamberlain and Russell can be effectively measured in terms of their individual Rebounds, Points Scored, Assists, Turnovers, Blocked Shots, Personal Fouls, and team wins, playoff series wins, championship wins, etc., but – unless, I’m mistaken – I do not see each of those specific categories … plus a number of others which also effect the outcome of games played in the NBA … included in a metric like “WP48”.
————————–
re: no one in his right mind suggests “relying exclusively on… productivity numbers” in assessing a player’s ability to produce in the NBA – Tom Mandel
————————–
I am sorry, but … when certain stats gurus go around saying that Kobe Bryant is not still the best player in the NBA today – and, by a fairly wide margin, to-boot – based on his inferior WinScore, or Wins Produced, or WP48, etc., it means that they are, in fact, using those specific metrics to determine who it is that they think qualifies properly as the best basketball player in the game today – based on the “available data” – while simultaneously disregarding a plethora of other factors.
And, unfortunately, the same thing holds true when comparing “average” players to other average players, or comparing other “top” players to one another, or to the prototypical/mythic “average” player, or when projecting which players are most likely to succeed/fail in the NBA, etc.
Chicago Tim
August 12, 2010
khandor
It’s one matter to scout players in college or foreign teams and try to predict how they will play in the NBA in the future. It’s even hard to predict after a rookie year how a player will develop in the NBA. But when it comes to assessing Kobe vs. other candidates for “best player in the NBA today,” we have a wealth of data, and we don’t have to predict anything. Therefore the arguments of those who wish to ignore the data become much more suspect.
some dude
August 12, 2010
“The problem is that the status quo ends with that franchise in ruins. Collison had real value as a trade piece and I gotta think Posey, as a defensive wingman with playoff chops, had some value to a contender too (Posey for Rasheed’s contract who says no?). Yes the trade is a wash i nterms of wins but they really could have done better (and Hollinger actually agrees with me which is weird).”
Arturo,
My argument isn’t so much with your grade of the trade. I’m fine with it being an F.
I just don’t agree on the WP analysis, because a lot of Collison’s minutes will be eaten up by CP3, not a worse player. And I think Posey’s 5WP guess is incorrect based on 3 year steady decline and age.
You can give it an F because of the considerations you mentioned (not getting more for Collison while also not slashing salary), but the argument shouldn’t be grounded on the WP you posted. I think it’s accurate.
some dude
August 12, 2010
Itallion Stallion,
Bargnani is anot a hybrid player. He’s a big man that doesn’t rebound, score in the post, defend the post, or help defend, but shoots and slashes.
Just because he’s 7 foot but plays like a SG/SF doesn’t make him hybrid. It makes him a SG/SF in a big man’s body.
For example, Magic Johnson was a hybrid. He could run the point, cut, shoot, defend guards, but also play in the post, score and defend down there. He wasn’t just a tall point guard. He was a hybrid because he could do multiple things. Pippen is another wonderful example. I think KG and Dirk are others.
If Bargs would be a hybrid, he’d be a great player. Look at KG in his prime. With his 1st step, range, ball handling skills and rebounding and defensive prowess, he was a clear cut top player in the game. Bargs does 1-2 things well and that’s it. And for a 7 footer to be terrible at rebounding and defense, you become a massive liability. You better be one of the best ever as a SG if you’re gonna be a 7 footer that can’t rebound and defend. He’s not close.
arturogalletti
August 12, 2010
SD,
That’s fair. I do think Posey can and would contribute on a good team (I’m still mad that the celtics let him go,I’m convinced he would have helped them get one more ring). I also think that wins wise the didn’t improve their roster and that’ll cost them CP3.
Tom,
I’m totally clear that no statistical model is perfect but this one allows me to make a reasonable guess at a high correlation.
Tom Mandel
August 13, 2010
Arturo — there’s no question that WP48 is the best statistical formula so far for assessing player contribution to team wins. Doesn’t mean someone won’t come up with something better, but for now it is the best by a tidy margin.
Because it gives you a single number, it is also incredibly convenient. Of course, the single number can hide anomalies and therefore lead a decision-maker into a mistake (i.e. preferring a guy whose foul rate is too high to be practical — no matter what his other numbers are), but that’s in the nature of things, and no one would fail to look at the numbers themselves in the decision-making process.
But the issue I raised has nothing to do with the quality of the statistical model — how “perfect” it is. It has to do with the difference between a statistical truth and the analysis of an individual case.
The logical conundrum I posed remains. If in fact WP48 (or, in this case, PAWS40) is an adequate measure of the likely success in the NBA of an *individual* player coming out of e.g. Division 1 NCAA, then you have to take Bryan Zoubek 6th in the draft; he had the 6th highest PAWS40 of any player in the draft.
If you won’t do that, then — without being willing, or maybe not able, to say why — you do grasp that predicting the future of a single young player is simply *not what WP48 is for.* It just can’t do it.
Tom Mandel
August 13, 2010
Aaah, I see that navigation forward/back among posts now appears *after* a post. That’s less useful than having it above the post, but at least it’s there.
But, I still don’t see a “recent posts” list in the right column as I used to.
The other way was better.
khandor
August 13, 2010
—————————
re: when it comes to assessing Kobe vs. other candidates for “best player in the NBA today,” we have a wealth of data, and we don’t have to predict anything. Therefore the arguments of those who wish to ignore the data become much more suspect. – Chicago Tim
—————————
When a supposed player evaluation metric fails to include such things as Team Win Totals, Playoff Series Victories, Championship Wins, etc., then it begs the question, “Who is it that is really excluding ALL of the available ‘data’ from an accurate evaluation of who is actually ‘the best’ player in the game today?” … as opposed to simply asking and then answering the question, “Which individual player has amassed the most prodigous ‘production numbers’, while participating in a game which fundamentally Team-oriented?”
—————————
re: Arturo — there’s no question that WP48 is the best statistical formula so far for assessing player contribution to team wins. Doesn’t mean someone won’t come up with something better, but for now it is the best by a tidy margin.
Because it gives you a single number, it is also incredibly convenient. Of course, the single number can hide anomalies and therefore lead a decision-maker into a mistake (i.e. preferring a guy whose foul rate is too high to be practical — no matter what his other numbers are), but that’s in the nature of things, and no one would fail to look at the numbers themselves in the decision-making process.
But the issue I raised has nothing to do with the quality of the statistical model — how “perfect” it is. It has to do with the difference between a statistical truth and the analysis of an individual case.
The logical conundrum I posed remains. If in fact WP48 (or, in this case, PAWS40) is an adequate measure of the likely success in the NBA of an *individual* player coming out of e.g. Division 1 NCAA, then you have to take Bryan Zoubek 6th in the draft; he had the 6th highest PAWS40 of any player in the draft.
If you won’t do that, then — without being willing, or maybe not able, to say why — you do grasp that predicting the future of a single young player is simply *not what WP48 is for.* It just can’t do it. – Tom Mandel
—————————
– Saying that someone specific is “by far the best criminal at-work today” does not make that person the equivalent of being a fine, upstanding individual.
– Saying that something is incredibly “convenient” does not mean that it, in fact, represents a real-life truth.
– Perhaps the flaw in your conundrum lies in your initial acceptance of any statistical-based metric – e.g. WP48, or PAWS40 – as “an adequate measure of an individual player’s ability,” coming out of a NCAA D1 program, relative to his peers and, therefore, also “as an accurate indicator of his likely future success in the NBA.”
– If you are able to see correctly how WP48, or PAWS40, etc., is truly incapable of making such ‘futures forecasts’ with a high degree of accuracy, then, the conundrum which I have concerning you is this: “What’s the reason you cannot see correctly how these same ‘measures’ also fail to paint an accurate picture when it comes to assessing, ‘Who the best player in the game actually is?’ vs ‘Who is the most statistically productive player in the game?’, since the highly specific answers to these two questions are not necessarily the same in the real world.”
nerdnumbers
August 13, 2010
“If you are able to see correctly how WP48, or PAWS40, etc., is truly incapable of making such ‘futures forecasts’ with a high degree of accuracy ”
Hey Khandor. You bring up an excellent point worth discussing. WP48 is very predictive of WP48. Or another way to phrase it is : Past performance in the NBA is a real good predictor of future performance in the NBA. Paws40 is a weak predictor, which is to say : Performance in college has some relationship on performance in the NBA, but don’t bet the farm on it. (There are tons of posts + comments on why this is a hard problem)
Now one thing missing from this is the last piece (Arturo and I have looked at this). The amount of minutes a player is given is a weak predictor of future minutes. Just because your player played 3000 minutes last year doesn’t mean they’ll be an ironman this year (Yao played a ton his first four years, look at him now). We also know a few players can hugely influence your teams performance. So if you can’t guarantee your best players will play, how can you know how well your team will do? Maybe Lebron gets hurt next year, maybe the Nuggets actually play Balkman. Maybe the entire Western Conference gets a Chinese superbug. No one can know these things. All we can know is how well the players will play. . . . if they play.
arturogalletti
August 13, 2010
Khandor,
So by your argument Kobe is a one man band. No one is arguing that the Lakers are not a great team. What the stats clearly state is that Kobe does not deserve as much credit for their success as is allotted to him. Gasol and Odom deserve similar credit. Switch Kobe for Lebron or Wade and the Lakers still win the title (even more handily) as they are both better basketball players by any reality based measurement.
This is all about qualitative vs. quantitative assessments of performance. At the end of the day you can argue potential, intangibles and physical measurements till you’re blue in the face but the true measure of a baller is on the court. Carmelo may have wonderful potential as a player but he doesn’t get it done on the court. What the stats tell us is how the player actually does empirically against NBA level talent and not how he should do based on the subjective judgement of a scout. Talent evaluation in the NBA can be statistically proven to be completely out of whack in the NBA, money does not correlate to wins.
As for the draft question, The problem isn’t with the model per se but with the sample itself. In he NCAA the level of talent from game to game fluctuates wildly. Imagine baseball players playing half their gaems in the majors and half in triple or even double A. The overall stats gathered are thus not an accurate reflection of the players talent (This is a problem that gotten worse as the level of talent in the NCAA has fallen). I can build a predictive model based on overall stats from the NCAA with close to 40% correlation but a more rational approach using a more representative subset of games is required.
some dude
August 14, 2010
Arturo, you had me til you mentioned Lamar. I’m fine with anyone arguing “Gasol deserves similar credit as Kobe,” but Lamar!?
It’s at this point where people question where those that do stats analysis actually watched games. There is no way any reasonable person could watch Lamar Odom and think he is half as important as Gasol or Kobe.
It’s amazing how this blog seems to absolutely love Lamar Odom and yet Laker fans want to tear their hair out watching him play.
And yes, if stats say he’s as valuable as gasol and/or Kobe, then I’m saying the stat isn’t measuring Lamar’s performance close to accurately. Note that I’m not saying Lamar is a scrub (far from it), but the guy plays once every 3 games and he far more inefficient than Kobe is offensively, but not in the traditionally measured way.
Tom Mandel
August 14, 2010
Khandor, there’s little point in dialogue with you, it seems. You didn’t take the time to *understand* the points I was making — which is of course your prerogative. Instead, you simply used my comment, as I note you doing quite often, as a platform to make a few more unsupported, and unsupportable, ex cathedra judgements.
As to: “‘Who the best player in the game actually is?’ vs ‘Who is the most statistically productive player in the game?’”, it is true by definition that a statistical measure of productivity tells you who is the most productive statistically. What else could it possibly tell you?
Given that a) the added-together WP48 scores of all players on a team across the season correlates to @ 94% with the team’s wins (as Dave claims) and b) statistically, players’ WP48 scores don’t vary meaningfully as they move from team to team, we can rely to a pretty high degree on the allocation of win-responsibility among a team’s players. That is not to say that we can rely on it *completely*, but it is all the same a rather good guide. A better guide certainly than the intuitions of a genius who simply “knows” who’s good and who’s not (or less so). Now, if greater or lesser responsibility for your team’s wins is *not* a measure of how “good” a player is, then of course WP48 is no help at all.
In any case, any statistical measure is only ‘good’ in the sense that a) it reliably aids decision-making and b) it is better than another measure in doing so. However “good” it is in these senses, it is not on its own a reliable predictor of a draftee’s success in the league if the draftee is a) young or b) provides a small dataset (two conditions that often go together obviously). But that is a *different* fact — a fact about *statistics* rather than a comment on the particular way of rolling up numbers.