The Warriors By Regularized Adjusted Plus-Minus
by onlxn on Mar 4, 2010 1:50 PM PST
Email
PrintThe results through February 25th, along with the data for every other team, courtesy of Joe Sill at Hoop Numbers. I will leave an explanation of the effects of regularization to those who are less ignorant than I, but supposedly Sill's method filters out teammate effects and other noise better than most. He lists the data in several ways... the link above points to the four-year time-weighted data, which gives a bigger sample size and hence more reliability. But the usual plus-minus caveats still apply, of course, and for low-minute guys like Hunter and Tolliver, grains of salt are especially recommended.
The numbers listed are the player's on-court effect on his team's point differential per 48 minutes... a +1.000 guy makes his team a net point better per 48, a -1.000 guy makes his team a net point worse. The vast majority of players lie somewhere between 3.000 and -3.000. LeBron (naturally) has the best four-year time-weighted RAPM of 6.046. (If that doesn't sound like a lot, consider that the difference between the Thunder and the Knicks is about six net points a game.) Helpfully, Sill separates results into offensive and defensive RAPM, giving a clearer picture of where guys help and where guys hurt.
Longer-winded thoughts on the Dubs' results can be found here, but a couple big takeaways...
1) Curry and Morrow are two of the biggest offensive assets in the league, and two of the biggest defensive detriments in the league.
2) Turiaf rates much better than Biedrins, but only because of the offensive gap between them... Biedrins's presence helps the defense a bit more than Ronny's does.
3) Randolph's somewhat subpar on both ends, and rates poorly overall. This is true of many young bigs -- Brook Lopez, Al Jefferson, Speights, Gortat, Hawes and Hickson rate worse than Randolph, among others -- but it's interesting.
4) Speaking of J.J. Hickson -- you know what he, Al Thornton, Sasha Pavlovic, Maurice Evans, Chris Wilcox and Ryan Hollins have in common? Of the 537 NBA players in these rankings, these are the only six guys that rate worse than Monta Ellis. Be afraid... be very afraid.
5) Let's end on a bright note: a young Warrior point guard rates as the 20th-most beneficial player in the entire league. Take a bow...
...C.J. Watson.
Tool through the numbers if you're bored. They're not the gospel, but they're damn interesting.
This FanPost is a submission from a member of the mighty Golden State of Mind community. While we're all here to throw up that W, these words do not necessarily reflect the views of the GSoM Crew. Still, chances are the preceding post is Unstoppable Baby!
110 comments
|
5 recs |
Do you like this story?
Comments
Do these numbers seem correct to everyone?
I don’t care what the numbers say personally, I would rather have Monta and Randolph out there then CJ and Hunter…
by freerandolph on Mar 4, 2010 2:59 PM PST reply actions 1 recs
Just some numbers from last years results of this stat.
Bet you guys didn’t know that
Dirk was the 50th best player in the league (only one slot behind Jamario Moon!) Duncan was 54th and Joe Johnson was 56.
Did you know
Matt Bonner was a top 10 player in the nba!
Ben Wallace was th 20th best player in the league last year, one behind Kobe! And to all you who think Kobe isn’t the best player on his team, this stat says that Gasol is the 45th best player, right after superstar Thaddeus Young.
I bet no one knew that Gerald Wallace sucked last year, even more then Derek Fisher.
And some of the free agents this year are overrated. Evidently Bosh isn’t as good as Fisher either. Bosh is also (evidently) considerably worse then Nate Robinson and former Warrior Al Harrington.
Why does everyone on this site want to trade for Bosh when we had a better player in Harrington!
Chauncy Billups and Paul Pierce occupy the two spots directely in front of (meaning that they are barely better then) none other then the beast that is known as… Ronny Turiaf! All three players rank around 100th best player.
I could easily go on showing how ridiculous this stat is, every time I glance at the list I see something new and utterly ridiculous, but I think this shows my point.
by freerandolph on Mar 4, 2010 3:11 PM PST reply actions 10 recs
You’re by no means required to take the stat seriously, and it’s just one part of a larger picture. But even the developers of the stat wouldn’t actually suggest that it can tell you who the “best” players are… what it tries to do is to tell you which players are most effective in their roles.
Matt Bonner was a top 10 player in the nba!
Nobody would seriously contend this, of course… Bonner is not one of the league’s 100 best players, let alone one of the 10 best. However, he is very good in his specialized role, as “big guy who stretches the defense by shooting threes”. He shoots extremely well and passes pretty well… at the other end, he can body up beefy guys pretty decently. There are a number of things he can’t do well at all, but Popovich doesn’t ask him to do anything of those things — he just makes Bonner do what he’s good at. As a result, the Spurs do incredibly well when he’s on the floor.
Matt Bonner’s not a great player, but he’s a great Matt Bonner… he is, while a limited weapon, a weapon that can be used to great effect in certain situations. The same goes for the other non-stars who rate highly here. Derek Fisher is not a star, but he’s good at being Derek Fisher… he still plays his Lakers role well. Al Harrington was never one of the three best players on the Warriors, but he was a guy that Nellie used intelligently and to good effect. And yes, Ben Wallace is still a really good Ben Wallace… an elite rebounder and defender can make a big difference, no matter how many free throws he airballs.
In the case of the Warriors, these numbers suggest that C.J. has been very effective in his role as a third guard who doesn’t dominate the ball, and that Monta has been very ineffective in his role as a superstar. That doesn’t mean that C.J. is more talented than Monta, or that C.J. would have better luck than Monta at dominating the ball. But it’s still interesting, and I don’t find it hard at all to believe.
Golden State Worriers: Angst & Analysis
This argument shows that the stat doesn't actually show what it is supposed to
The stat is supposed to show what players contribute to their teams regardless of context. The fact that Matt Bonner can be used well in the “certain situations”, which explains him being highly regarded by this stat, means that the stat doesn’t do a good job of taking situation out of analysis.
Also, to add to my previous list...
Even if you don’t think Ellis is very good, it is utterly ridiculous to say that he is one of the 10 worst players on his team, let alone in the league. Even if you do think he is one of the ten worst on out team, do you really think he is one of the worst players in the league!?!?!?
One more just for fun :)
Tyreke Evans this year has been the 504 best player! Right behind James Johnson! What!?!?! You don’t know who James Johnson is!?!?!
I don’t either…
Do I think he’s of the ten worst players in the league? No, of course not… again, that’s not what this measures.
Do I think he’s been one of the ten most ineffective players in the league? Sure, I could believe that. No problem at all. The Warriors have been completely horrible when Monta has been on the floor doing his thing. When he hasn’t played, they’ve been much, much, much better. And considering how often he misses or turns it over, and how much less our other, more efficient scorers see the ball when he plays, and the fact that he punts on the occasional defensive play to save energy… I don’t think it’s a coincidence in the slightest.
I believe that Monta Ellis is a very good basketball player. I believe Monta Ellis stinks in his current role. Both these things can be true.
Golden State Worriers: Angst & Analysis
Both those things can be (and in my opinion are) true.
But if this stat isn’t supposed to tell us how good players are, then whats the point of it?
It's supposed to tell us how effective they are in their role
it helps us figure out that Monta Ellis hasn’t been effective being the primary ballhandler/“The Man” this year.
by Reverend_Randy on Mar 5, 2010 3:45 AM PST up reply actions
The stat is supposed to show what players contribute to their teams regardless of context.
According to whom?
This stat does a pretty fair job of isolating a guy’s effect on a game from the effects of his teammates, but it can’t remove context altogether… how could it? How is any system supposed to account for the fact that, say, Corey Maggette has been playing out of position for most of his Warriors career? All the system can say is that Maggette was a pretty effective player as a Clipper, and has been less so, primarily for defensive reasons, as a Warrior. You’re always going to have to make some inferences on your own.
There will never be a basketball stat that gives you a full, clean picture of a guy’s ability… it’s just too interconnected a game for that. But this stat can show you part of the picture, and I think it’s a worthwhile part.
Golden State Worriers: Angst & Analysis
What part of the examples I game is worthwhile?
I agree with you in that I don’t think its possible to eliminate context, and this stat makes the claim that it does… Thats why I’m calling BS.
I don’t think its possible to eliminate context, and this stat makes the claim that it does
Where does it make this claim?
Golden State Worriers: Angst & Analysis
Meh...
supposedly Sill’s method filters out teammate effects and other noise
I you want to quibble semantics… If “teammate effects and other noise” don’t count as context, what are they? Sure, it doesn’t claim to eliminate noise completely, but it does claim to do it better than other statistics… yet its findings are not particularly trustworthy, as FR has pointed out.
And you know what, maybe teammate effects should factor into how a player is rated (i.e. good players make others around them better).
Trust me, learning english isn’t a waste of time. It is actually sort of useful.
-randolphforpresident
by Dubs fan in Boston on Mar 4, 2010 5:09 PM PST up reply actions
I wasn’t sufficiently clear. I actually wrote"supposedly Sill’s method filters out teammate effects and other noise better than most," intending “most” to be read as “most other plus-minus systems.”
it does claim to do it better than other statistics… yet its findings are not particularly trustworthy, as FR has pointed out.
Well, first of all, FR’s list of goofy rankings comes from a one-year sample, which is much noisier and much less credible than a four-year sample. By the equally-weighted four-year sample, the most effective players in the league over that span have been LeBron, Garnett, Nash, Dirk, Kobe, Duncan, Wade, Chauncey, Rashard Lewis, CP3, Howard, Odom… you get the idea. There are still some showings that might surprise people — this system finds Rashard Lewis, like Bonner and Harrington and several other stretch fours, to be extremely effective — but they’re not nearly as wacky in aggregate as the data from the one-year samples. Note that above, I was using four-year data.
But more to the point, why would these rankings not be trustworthy? Sill’s methodology has been much discussed and pored over, and most ABPRmetricians see a lot of merit in his approach (discussion can be some found here)… his paper that explores this method is a finalist for the award for the best sports-related statistical research paper at the Sloan Analytics Conference this weekend. We’re supposed to ignore the system because it sports some small-sample results it disagrees with?
maybe teammate effects should factor into how a player is rated (i.e. good players make others around them better).
Those do factor in — LeBron’s +6.0 offensive RAPM reflects the change in the overall effectiveness of the Cavs’ offense when he plays, not just his personal production. A lot of that jump does come from making his teammates better. What RAPM tries to do, like all plus-minus systems, is to give that credit to LeBron, not to the teammates. You don’t get credit for having good teammates, but you do get credit for making your teammates better. That strikes me as the right balance.
Golden State Worriers: Angst & Analysis
by onlxn on Mar 4, 2010 6:37 PM PST up reply actions 1 recs
This stat does a pretty fair job of isolating a guy’s effect on a game from the effects of his teammates
As FR pointed out, there are way too many outliers for it to be “doing a pretty fair job” of anything involving rating players against each other. It’s one thing to have Troy Murphy in the top 20, but have everybody else make sense, but it’s another thing entirely to vastly underrate 2/3 of the top players in the league.
There will never be a basketball stat that gives you a full, clean picture of a guy’s ability
Absolutely, but every metric should be validated before it’s trusted. This one doesn’t validate well (i.e. far too often, players who are good are rated poorly and players who are bad are rated well).
Trust me, learning english isn’t a waste of time. It is actually sort of useful.
-randolphforpresident
by Dubs fan in Boston on Mar 4, 2010 5:03 PM PST up reply actions
I think the problems with APM come when people compare players across teams.
It’s an interesting metric for looking at an individual in the context of their team, but any meaning it has breaks down when you look at the entire league. These numbers have a ton of noise, and any league-wide assessments are fun to look at, but have very little use, it would seem.
by Spider Jerusalem on Mar 4, 2010 5:30 PM PST up reply actions
This is definitely true… the numbers make more sense in a team context than in a leaguewide context. Still, I’d say the league-wide assessments aren’t wildly off, given a big enough sample size. The guy with the 20th-best RAPM in the league is not definitely less effective than #19 or more effective than #21, but it’s pretty safe to say he’s a very effective player.
Golden State Worriers: Angst & Analysis
True, but players still need to be looked at within the context of their team.
So instead of seeing Matt Bonner at #17 and declaring he’s the 17th best player in the NBA, I see Bonner at #17 and say: well, it seems Matt Bonner was very effective for his team in his prescribed role.
by Spider Jerusalem on Mar 4, 2010 11:52 PM PST up reply actions
I think it was best said in a previous thread where APM was used:
APM is a great tool for coaches, not so much for GMs (I’m paraphrasing). In that it shows how effectively a certain player fulfilled his given role, and doesn’t really comment too much on his talent level (though there are obvious correlations).
by Spider Jerusalem on Mar 4, 2010 11:54 PM PST up reply actions
That’s a good way of looking at it, and I think it’s particularly germane when considering several Warriors. Monta, Biedrins and Maggette all rated as solidly effective players two years ago, so it’s not like these guys inherently stink… in their current roles, however, they’re not very effective.
Golden State Worriers: Angst & Analysis
This stat does a pretty fair job of isolating a guy’s effect on a game from the effects of his teammates
Pretty good job? More than half of the players on the Warriors have std errors for the measure greater than the absolute value of the measure. Standard errors that large leave a lot to be desired.
by jae on Mar 5, 2010 2:38 AM PST up reply actions
So it goes with plus-minus statistics… there’s always going to be some murk you have to deal with. That doesn’t mean you have to just throw the metric out. Even if you can only peg a guy in a 2.2-point range with confidence, it can still tell you a good bit. For instance, based on this data, you can make the following assumptions about the Warriors’ guards:
Stephen Curry
1) has been an effective offensive player.
2) has been an ineffective defensive player.
3) hasn’t made an enormous impact on the team’s effectiveness overall.
4) has more likely helped than hurt.
C.J. Watson
1) has almost certainly had some amount of positive impact on the Warriors’ offense.
2) has almost certainly had some amount of positive impact on the Warriors’ defense.
3) has been a significantly effective player overall.
Monta Ellis
1) has been an extremely ineffective offensive player.
2) has not drastically changed the Warriors’ defensive effectiveness in either direction, but has probably hurt it.
3) has been an extremely ineffective player overall.
Precise? No. But these are useful takeaways all the same.
Golden State Worriers: Angst & Analysis
Do you really think CJ has had that positive of an impact on the team when he has played?
by freerandolph on Mar 5, 2010 11:29 AM PST up reply actions
Do I personally think that? Yeah, I do, actually. I think CJ’s a massively underrated guy. He’s not a star, but he makes very few mistakes… he almost never turns it over, he grabs a lot of steals without drawing a lot of fouls, he doesn’t force shots he can’t make.
CJ’s played in 47% of the Warriors’ minutes so far this season. In those minutes, the Warriors have outscored their opponents at a 1.6-points-per-48-minute pace. The other 53% of the time, they’ve been outscored by 8.6 points per 48 minutes. With CJ on the floor, the Warriors have played like an above-.500 team; without him, they’ve played about as well as the Timberwolves.
Now, the sample size there is small. And a disparity like that is not always proof that a guy’s making a big difference — a Cav who subbed in and out when LeBron did, for instance, would look great for reasons having nothing to do with him. Sometimes team patterns can create illusions.
But there are no patterns on these Warriors — the team is in total chaos, And CJ has started, he’s come off the bench… he’s played three minutes, he’s played forty-eight minutes… he’s defended ones, he’s defended twos, he’s defended threes. He’s spent a lot of time playing alongside Curry, Morrow, Tolliver and Turiaf… he’s spent a fair amount of time playing alongside Monta, Maggette, Vlad and Randolph, an entirely different set of teammates. CJ has been bounced all over the place.
Through it all, there’s been one constant: the Warriors have been a lot better with him on the floor. And while I don’t think his presence makes a ten-point difference or anything like it, I do think his presence helps a good bit. The Warriors were better with CJ on the floor last year, as well. They were worse with him on the floor in his rookie season, but only barely, and considering that he was usually subbing in for Baron, that’s actually pretty impressive.
Good things happen for the Warriors when CJ Watson’s on the court. I think he’s a real asset.
Golden State Worriers: Angst & Analysis
So you don't believe in stats
why comment in a stats thread?
by Reverend_Randy on Mar 5, 2010 5:03 PM PST up reply actions
I don't believe in a stat the first time I ever hear of it right away...
And since even in this thread it has been pointed out that stats can conflict each other (someone wrote about a stat that says that out of CJ Monta and Curryburry, CJ fairs the worst in scoring compared the scoring of the player he guarding) you question doesn’t make sense.
Do I believe in stats? Are you talking about the ones that say CJ is better then Monta and Curry or the ones that say CJ is worse the Monta and Curry?
I get you. You just have to find reliable stats in your opinion. This stat isn’t too reliable in my opinion, but i would definitely not call it meaningless.
The stats i like to use are TS%, WP by Dave Berri, and 82games.com PER comparison. The last two aren’t completely right all the time, but i’ll take my chances.
banned like chris andersen
by MDB on Mar 5, 2010 7:53 PM PST up reply actions
Which stats say that CJ is better?
Certainly not this one.
by Reverend_Randy on Mar 5, 2010 11:05 PM PST up reply actions
It ranks CJ as being more effective
in his role, not a better player.
by Reverend_Randy on Mar 6, 2010 4:31 PM PST up reply actions
CJ is also better than Monta by WP48.
After 41 games, at least, monta was .049, Watson was .103.
Again, not surprising. Monta has been bad this year.
Basketball Reference’s Win Shares posts an even bigger disparity - they credit CJ with 2.9 Win Shares to Monta’s 1.0, and that’s despite CJ getting less than two-thirds as many minutes as Monta. By Basketball Prospectus’s Win Percentage, Monta has a slim .482.478 lead, but they haven’t updated their stats in two weeks… I wouldn’t be surprised if CJ had nosed ahead in that time.
Golden State Worriers: Angst & Analysis
…dang cross-out. ’Twas unintended.
Golden State Worriers: Angst & Analysis
What do you mean by "better"?
Do you mean more effective or more talented? CJ is definitely one of those things.
by Reverend_Randy on Mar 6, 2010 10:08 PM PST up reply actions
All this says is that CJ this year has done a much better job of being a 3rd guard than Monta has done being a ball-handling, high minute PG.
Yes
He’s been more effective than Monta this season.
by Reverend_Randy on Mar 6, 2010 11:38 PM PST up reply actions
You guys are ignoring the question that I asked, or at least interpreting it in a way that is obviously different then what I meant.
Whatever you think “better” means, (probably something like if Monta and CJ switched places in how much the team relied on them, how well do you think they would do) do you think CJ is better then Monta?
I guess you guys are saying that this stat shows that CJ is better at filling his role then Monta is at filling his role. To that, I respond, who cares if a player who doesn’t have an extremely important role on their team fits that role better then a player with a very important role on a their team fits their role.
by freerandolph on Mar 7, 2010 12:47 AM PST up reply actions
Well, the last time you tried to ask
you made it more like an accusation so I responded it to it like it was an accusation.
None of us think that CJ is a better/more talented player than Monta. I don’t know why you thought anyone was saying that, because no one was.
The important part isn’t about CJ being really effective. The important part is about how Monta has been really ineffective this year. It’s also kind of interesting because it correlates with what Homer Fitz says all the time about CJ being one of the best backup guards in the NBA.
by Reverend_Randy on Mar 7, 2010 2:02 AM PST up reply actions
What the numbers show is that the Warriors have performed better as a group when Monta has been on the bench and Watson has been on the court. Is that entirely those two variables? The plus minus doesn’t say that, but other numbers suggest that Monta has been overmatched and that his individual play suggests that he’s part of the problem more than part of the solution.
Many people use measures of “better” that remain subjective messes. There’s clearly aesthetic values that go into it and there’s clearly some some of “good player” inertia where someone who has been considered “good” (largely by having scored a lot in the past) will for quite some time be considered better than the previously less heralded player. As a NBDL player, there’s some inertia against Watson, fair or not, that goes beyond any actual evaluation of what he does on the court. Monta faced that as a 2nd round choice once upon a time, but he was younger, was out of HS so there was a thought that he might have been the ‘steal’ and he’s flashier, so that tends to trump “quietly effective”.
Still the only measure that’s worth while for “better” is one that figures how much more likely a team is to win with a player. And this year, Monta hasn’t been ‘better’ by that measure. Does that mean it’s likely to be that way in the future? Not necessarily, though it’s useful to be open to the possibility that the evaluation of Monta as ‘better’ has been in error. Such errors do happen and they happen with rather alarming regularity.
(fyi, it’s “better than”, not “better then”.)
by jae on Mar 7, 2010 3:44 AM PST up reply actions
I can’t dispute this about C.J. I’m not 100% sure about this take on him, but I can’t dispute it. This season, C.J. has been very solid. There is no denying that.
Welcome to the Pit of Despair! Don't even think about trying to escape.
I really haven't noticed him being very solid. Its definitely possible that I just haven't noticed, but I thought he wasn't haveing that great of a year.
It has seemed like he started doing better recently then at the begining of the season.
by freerandolph on Mar 11, 2010 10:33 AM PST up reply actions
He had a good november and december
then a crumby January. He’s been good ever since.
by Reverend_Randy on Mar 11, 2010 3:19 PM PST up reply actions
The stat is supposed to show what players contribute to their teams regardless of context.
No. You have that exactly backwards.
The stat is supposed to show what players have contributed to their teams regardless of in the context they were actually used.
Something like WP48 is supposed to show what players contribute regardless of context.
where is the stat that adjusts for madman coaches who think actual positions don’t matter?
Sittin in my scraper watchin Oakland goin wild, ta-dow!
…I wish I was a better writer/more clear thinker than I am. There’s just so much wrong with that phrase, and I only bring it up because I actually see similar things tossed around pretty often. It’s just not a correct mindset/criticism.
by Missing Barry on Mar 5, 2010 1:43 PM PST up reply actions
huh? the take home message of this whole APM thing is that the way players are used is just as important, if not more important, than their actual talent level. Having a coach who doesn’t understand that (as we currently do) is clearly a “detriment to winning.” Maybe you just didn’t understand my point…?
Sittin in my scraper watchin Oakland goin wild, ta-dow!
It had nothing to do with the actual point, which is certainly a valid one. It’s the way you phrased it I object to. The “Where is the stat that adjusts for…” part.
by Missing Barry on Mar 5, 2010 4:33 PM PST up reply actions
No, as I’ve said, I do statistical research, and it’s a conceptually flawed thought process. Yes, pointing out Nellie’s lineups is fair, but there’s obviously not a stat for it and probably not a practical way to measure/adjust properly for that kind of thing. It’s a discussion point – and there are some things you can do, like look at how much more effective a guy like Maggette was when he was playing the 3 full time and point that kind of thing out, but that phrase just….irks me.
by Missing Barry on Mar 5, 2010 4:47 PM PST up reply actions
Not sure what’s so conceptually flawed about it. You said it yourself, there are ways to measure these things (namely by looking at what happens when guys are forced to play out of position), but there may not be a catch-all metric for it. More than actually looking for a given statistic, I was pointing out that APM tells you more about coaching decisions then it does about an individual player.
And no, I am not trying to resurrect the retarded “stats vs gametape” debate that dominates this board ever so frequently.
Sittin in my scraper watchin Oakland goin wild, ta-dow!
It’s really just the phrase itself that bothers me, not the point you’re making.
by Missing Barry on Mar 8, 2010 2:53 PM PST up reply actions
Can't believe you missed this one...
Kevin Durant is the 62nd best offensive player in the NBA. Right between Al Harrington and Aaron Brooks… and worse than Andrea Bargnani, Marco Belinelli, Derek Fisher, Beno Udrih, Goran Dragic, Jon Brockman, Matt Barnes (the 11th most lethal offensive weapon an NBA team could have employed over the last 4 years), and Jamal Crawfor- (#6).
Also notable, Cartier Martin, Nick Collison, Eric Dampier, Hilton Armstrong, and Jermaine O’Neal are all better defenders than Tim Duncan, Marcus Camby, and Dwight Howard (76th!).
Suffice to say, just because it looks all scientificy doesn’t mean it can pass the most basic common sense tests.
Owen, it’s an interesting idea trying to filter out teammate effects and noise, but the metric is clearly not well thought out, as it validates very, very poorly in so, so many cases. Just because it sorta fits with how you see the Warriors doesn’t mean it’s actually worth anything or even if it should be trusted. I’m not saying you’re wrong on the Warriors, just that this statistical metric doesn’t add any credence to any analysis of the NBA in any fashion whatsoever.
Trust me, learning english isn’t a waste of time. It is actually sort of useful.
-randolphforpresident
by Dubs fan in Boston on Mar 4, 2010 5:40 PM PST up reply actions
Suffice to say, just because it looks all scientificy doesn’t mean it can pass the most basic common sense tests.
Here’s a common sense test for you: if a guy’s on a team with a horrible offense, and it gets even more horrible when he takes the floor, is it likely that he’s playing effective offensive basketball?
That was the case with Durant in his first two seasons… his gunning didn’t get the Thunder anywhere. This year, he’s upped his game (on both ends), and his numbers in this system, as in every other, are rising.
Just because it sorta fits with how you see the Warriors doesn’t mean it’s actually worth anything or even if it should be trusted.
Randolph and Biedrins, the two guys I spend most of my time defending, fare horribly here… I don’t find this data worthwhile because it fits my preconceived notions. I find this data worthwhile because APBRmetricians I respect find it worthwhile. You can discount it if you like, but this is not just some goofy thing some guy made up. Several teams have hired analysts because of their work in adjusted plus-minus.
Golden State Worriers: Angst & Analysis
Dubs fan...
I know it may seem crazy that I missed what you pointed out, but there were just too many ridiculous things to point out and I didn’t want to waste too much of my time pointing them out.
Going through the list and pointing out examples that you don’t think are correct, especially if you don’t have a good feel for what the statistic is actually telling you, isn’t really a meaningful way to discuss whether it’s a useful statistic or not….
by Missing Barry on Mar 5, 2010 5:42 AM PST up reply actions
I don't get how that isn't a fair thing to do...
From what I read from Onix about the stat, it seemed like it was trying to tease out a players comparison for their team. If that is what the stat really does, then the best players in the league should be ranked at the top using this stat. Am I wrong about that? If so, explain please.
by freerandolph on Mar 5, 2010 11:31 AM PST up reply actions
The stat is meant to identify how a player being on the court has affected the teams performance when he’s on the court. It depends on how the player is being used, and it’s not a measure of “true talent level”.
Also, it’s just not an objective or meaningful evaluation to go through the list and say “Player X is ranked here and my opinion I formed before seeing this list is that he’s a better/worse player so the list is wrong”. Even if you find a bunch of examples, that’s not actual analysis. That’s just rejecting something that doens’t fit what you expected. You have to come up with reasons why the stat does or doesn’t make sense. Anecdotes are not evidence. Plus, who knows, maybe these guys are just excelling in the roles they’re put into? That doesn’t make them good players, just means they’re productive at what they’re expected to do.
by Missing Barry on Mar 5, 2010 11:51 AM PST up reply actions
You have to check a stat with you anecdotal opinions!
Otherwise I could throw up a stat like, “Monta Ellis is sixth in the league in scoring so he is the sixth best player” and if you just trust the stat and don’t compare it to your anecdotal opinion, then you have nothing to debunk a stat with…
I see you point, and maybe the stat can be used for something in a valuable way, but I just don’t see how a stat that rates Bosh as less valuable to his team then Fisher (and I could list a BUNCH of other examples as ridiculous as that) can be taken to mean much.
by freerandolph on Mar 5, 2010 12:58 PM PST up reply actions
Otherwise I could throw up a stat like, "Monta Ellis is sixth in the league in scoring so he is the sixth best player" and if you just trust the stat and don’t compare it to your anecdotal opinion, then you have nothing to debunk a stat with…
No. No, No, No, No. :(
You do actual, objective research into what wins games, and then you look to test whether PPG is actually a good indicator towards that. Turns out, PPG is meaningless and has no context – you need to know how efficient it comes, how many minutes they’re playing, etc – and we find that out by doing real, objective research. Subjective opinions/anecdotes have no place in this research. You need populations/objective random samples with large enough sample sizes.
(Sorry for the attitdue, I do statistical research for a living)
by Missing Barry on Mar 5, 2010 1:37 PM PST up reply actions
What kind of statistical research do you do for a living? (just because I'm curious)
If this stat really shows what contributes to winning, why are players like Pierce Duncan Howard Kobe, who have consistently contributed to winning ranked so low on last years stats?
Huh?
Duncan is +2, second-best on his team and 28th in the league. Kobe is +3.5, which is a pretty fantastic number, #7 in the league. Dwight Howard is over four, and has the second-best number in the league. Pierce is over 1, which is good, but certainly not spectacular … which makes sense. He’s been on the downslope for several years now.
I’m literally baffled here by how you think these players look bad according to these stats.
The stats linked at the top of this page are a 4-year weighted average
Since everybody acknowledges that +- based stats are very noisy, this seems like a more logical way to look at these numbers than looking at any one year’s numbers.
That was just the first link I looked at, but it seemed to me that since last year was the freshest in my memory
that it would be interesting to see how players were ranked compared to how I remember them performing last year.
Why?
A season is an arbitrary amount of time. In and of itself, there’s no reason why one should assume that a season is enough time to quiet the noise in the data.
When you do statistics, you do it for a certain level of reliability in the results, and then you get as much data as you need to get that.
The challenge with RAPM is that you’re trying to tease out a lot of information. Yes, it’s great that you have 82 games … but how many minutes do you have with any given player on the court and NOT certain other given players on the court?
For example, if I’m looking at Varajao’s performance, I may have a lot of Varajao minutes … but most of them are also LeBron James minutes. To make make RAPM work, you need enough Varajao-but-no-James minutes. I also need Varajao-but-no-Paker minutes, and Varajao-but-no-Williams minutes – enough of ’em to get me through small sample size issues.
A stat like WP48 doesn’t have those problems, so you would expect it to be reliable with a much smaller number of games.
Oh goody, An adjusted plus/minus post. Time for people to freak out because they don't get it.
Thanks for posting this onlxn. I’ve tried using APM in the past around here to little affect.
The main problem is that people see the list and think it’s an absolute ranking of true talent. It’s not. It’s simply a estimated measure of how well the team did when a player was on the court, adjusted for the players he played with and against. Naturally there is noise, but as far as measuring a players impact on the game, and how much a player has impacted his team winning, it’s a pretty solid measure.
RAPM tells us that the Warriors play about 2 points better than their opponents when CJ Watson plays, and a little over 3 points worse when Monta plays. If you’ve been watching the Warriors this season, this should make a lot of sense. It’s NOT saying that CJ is definitely a better player than Monta. It’s saying that the Warriors play better when CJ is in.
CJ is better
Than Monta at certain things. CJ can shadow Ray Allen entire quarter and get 6 steals! I don’t think Monta can play defense like that. CJ can shoot spot up 3s very effectively as well as push the ball. He cant finish like Monta in traffic, but thats not end of all things. Its just casual fans think that flashy finishes win games. They don’t. Effectively executed plans win games. And CJ is very good at executing game plan effectively.
They why isn't he a starting guard on our team.
The last time I checked we were trying to win games, so if CJ is the best chance at doing that don’t you think he would have been utilized more this season???
Maybe the coaches don’t think he’d be as effective in a bigger role as he is in the current role he’s in. Just because he’s effective in a limited role doesn’t mean he can be effective trying to do more than he’s currently doing.
by Missing Barry on Mar 5, 2010 5:44 AM PST up reply actions
1) CJ is being utilized more and more… since coming back from his hand injury, he’s averaged 33 minutes a game. It may have more to do with Monta’s injuries than anything else, but CJ is getting starter’s minutes right now.
2) As Barry points out, the fact that CJ is excelling in one role doesn’t necessarily mean that he’d excel in another.
3) Maybe Nellie decided it wasn’t worth giving bigger minutes to a guy who might not be here next year.
4) Maybe Nellie decided it wasn’t worth ruffling Monta’s feathers to win a few extra games in a lost season.
5) Maybe Nellie’s just making an evaluative mistake.
Golden State Worriers: Angst & Analysis
So the conclusion to take from CJ's stats is...?
That he should be starting?
That he is good in a limited role?
That he is the best player on our team?
or what?
I just don’t get what the point of this metric is if its not telling you how a good a player is. You keep saying that it tells you how a good a player is in their role… So does that mean that a generally bad player who plays their role well will be ranked high on this stat?
by freerandolph on Mar 5, 2010 11:35 AM PST up reply actions
So does that mean that a generally bad player who plays their role well will be ranked high on this stat?
If their role helps their team and they play it well, sure. See: Bonner, Matt.
by Missing Barry on Mar 5, 2010 11:52 AM PST up reply actions
So what conclusion are you supposed to come to from the stat????
It seems from what everyone is saying the stat isn’t telling you how good a player is. It’s telling you how much a player is contributing to winning on a team. But how can you say that Bonner is contributing to his team more then Duncan is? Is the stat saying that if you took each player of the team for a season that they would do better with Bonner and no Duncan then the would do with Duncan and no Bonner? I just don’t get what this stat is trying to show. It seems like stats are usually used to rate players, but I guess thats not what this stat is supposed to do. But everyone keeps saying that this stat is meant to show a players impact on a team, which doesn’t seem too terribly different (although slightly different) then showing how good a player is…
Also you never answered my question in the post that you replied to.
I figure I should probably pass on a lot of your questions to people with more knowledge and less skepticism about this particualr stat than myself, so I’ll let them chime in. I’m just here to try to make sure we’re all using sound statistical principles/theories to support opinions. :)
by Missing Barry on Mar 5, 2010 1:39 PM PST up reply actions
So the conclusion to take from CJ’s stats is…?
If I were the Warriors, here are some conclusions I’d draw from CJ’s good showing. These are, obviously, just my opinions.
1) CJ is probably the best defensive guard on the roster. The fact that he has the best defensive RAPM is not necessarily proof of that… sometimes a smart team can cover up a defender’s weaknesses in certain schemes. (The much-discussed Matt Bonner, for example, would probably rate as a less effective defensive presence on a team that didn’t use him as carefully.) But there are no aspects to the Warriors’ defense that would make me think CJ’s numbers are inflated… like Curry and Monta, he’s basically just sent out there to cover a dude and play the passing lanes. He has seemingly done that a lot better than they have. So if you are looking to compete better on the defensive end, it’d probably be a good idea to have CJ on the floor.
2) Some of Monta’s minutes should probably go to CJ. Obviously a non-issue while Monta’s hurt, but this is a tweak worth making even when Monta’s healthy. Even if CJ’s effectiveness went down a little due to a larger role (hard to say if it would), he’d still be a net positive. And who knows — Monta’s effectiveness might improve if he didn’t play so many minutes. We can’t know if that’s true, but there’d be no real reason not to find out.
3) It may be worth spending a little to keep CJ in Oakland. A good third guard is a nice thing to have, and CJ rates as one of the best around, if not the best. If another team offers him big money for a backup — say, $11 million total over three years — the Warriors should at least consider matching, even with Curry and Monta aboard. CJ’s not a world-beater, but he’s a useful piece of the puzzle… he’s very good at what he does. And a team this bad should not give a guy like that up lightly.
Golden State Worriers: Angst & Analysis
I understand that reading the stats make CJ seem to be solid defensively, but the eye site test has shown many times over the last 2 years him getting burned with ease by many lesser players. He also has a tendency to miss rotations badly. (Not that this is a strength the rest of the team has) Maybe it’s just me, but it has been fairly obvious many times.
Fully agree that 5-7 minutes of Monta ball should be given to CJ. Seeing how Curry’s efficiency has somewhat slipped while playing 40+ seems to lean towards evidence that Monta would be more efficient with more rest.
I’ll agree that you try to bring back Watson IF they don’t land Wall or Turner. The problem is he wanted to go to the Magic this last year, so why wouldn’t he want out again?
I would agree that CJ gets burned a fair amount. On the other hand, though, he is an incredibly effective gambler — when he goes for a steal, he’s very likely to get it. Gambling for steals is not generally a good defensive formula, but CJ seems to make it work better than most. He also doesn’t foul guys in the act of shooting very often. These are little things, but they add up.
I don’t think he’s a big-impact defensive player, but I do think he’s the best defender of our guards… he handles the chaos better than the other guys.
The problem is he wanted to go to the Magic this last year, so why wouldn’t he want out again?
He’s a restricted free agent again, so the Warriors are free to match any offer. If he’d clearly just hate to be here, maybe that changes their thinking. But they’ll be able to keep him if they’re willing to pay.
Golden State Worriers: Angst & Analysis
So just conceptually, a positive adjusted +/- on defense for Watson (if we assume it’s accurate) doesn’t mean he’s a good defender, just better than the average of the other guards we play when he’s not on the court, right?
by Missing Barry on Mar 5, 2010 4:34 PM PST up reply actions
Conceptually, it means that Watson is a good defender, period (again, assuming it’s accurate)… adjusted plus-minus tries to measure a player’s contributions in absolute terms, not simply relative to those of his teammates.
An easy shorthand way to look at this would be to look at the RAPMs of this year’s Nets. (This is single-year data, so there’s tons of noise here, but we’re not worried about the individual ratings right now). In a simple net plus-minus, as listed at 82games, the players’ results would (when factoring in playing time) add up to zero… they’re only compared to one another. In an adjusted plus-minus system like this one, though, the Nets are compared to everyone in the league… in that context, just about all of them rate as ineffective players. The sum of their RAPMs — again, corrected for playing time — should approximate the team’s efficiency differential. In the Nets’ case, that number is currently -11.9 (98.6 offensive efficiency – 110.5 defensive efficiency), though they were surely a bit different when these RAPM numbers were calculated last week.
So theoretically, C.J. Watson has been an effective defender period, not simply a less crappy one than Monta and Curry. He doesn’t rate as a hugely effective defender, but he rates as an effective one.
Golden State Worriers: Angst & Analysis
Hmmm, I don’t know anything really about how the stat is constructed, but it seems to me it would really be difficult to rate these players in a context that isn’t related to their teammates….
by Missing Barry on Mar 7, 2010 2:22 PM PST up reply actions
Very argumentative, as that claim was never made.
But, since you opened the door, I would say Monta has been the best of the 3 (not saying much) at staying in front of his guy. This can probably best be attributed to his quickness. I agree with onixn that Watson has developed a good knack (as Monta has had and Curry is learning) for timing steals.
Interesting stat: On 82games.com, they show stats for production per 48 minutes against opponent counterpart production per 48 minutes for each player. Of the 3 players, CJ has by far the worst differential.
This stat, like most all other stats, has flaws, but is at least interesting in comparing how the opposing player at that position plays when a certain player is in.
"The main problem is that people see the list and think it’s an absolute ranking of true talent. It’s not. It’s simply a estimated measure of how well the team did when a player was on the court, adjusted for the players he played with and against."
“Naturally there is noise, but as far as measuring a players impact on the game, and how much a player has impacted his team winning, it’s a pretty solid measure.”
So first you say the problem is that people think its a ranking of true talent. Then you say its a pretty solid measure of how much a player helps a team win… That sounds an awful lot like what I would call talent. There aren’t very many players who I would say are very good or talented if they can’t help a team win…
How is saying that the Dubs play better with CJ in then Monta saying anything other then CJ is better then Monta.
Look, I’m not saying that Michael Jordan is better then Turiaf, I’m just saying that the Dubs would play better if they had MJ in his prime then having Turiaf in his prime.
No
you’re confusing talent and effectiveness in a given role. Completely different things. CJ isn’t a hugely talented player, but he’s a very effective backup guard. Monta is a very talented player, but isn’t an effective ballhandler or primary offensive option.
by Reverend_Randy on Mar 5, 2010 3:47 AM PST up reply actions
Not at all.
I actually think Monta is a great example of the conceptual problem you’re having here.
Monta is clearly an extremely talented basketball player. He has a skill set that few other players have.
However, this season, he has not translated that talent into performance on the court. Whether that fault is his, his coaches, or whatever … well, that’s an interesting discussion. But I think Monta, this season, is a great example of how a player can be both talented and ineffective.
Let’s use your Turiaf-for-MJ example. Let’s say that we had MJ on this team, in his prime, but instead of letting him be MJ, we asked him to be Turiaf. He was asked to guard the opposing center. He would door very poorly at that. Jordan was an excellent defensive player but even mediocre centers would score easily over him. Despite being a perfectly reasonable rebounder for a guard, Jordan would be vastly outrebounded playing center. And his offensive game based on his explosiveness would’t work so much if you just threw it to him in the low post, and asked him to score using low-post moves. (Even that late-career, turnaround/fadeaway jumper would be fairly ineffective when he was guarded by a center.)
Michael Jordan, in those circumstances, would be every bit as talented as he was … but he wouldn’t be very effective.
Now asking Jordan to play that way would be mind-numbingly stupid. If an opposing team put a center on him, nobody in their right mind would expect Jordan to beat him in the low post … they would isolate him on the wing.
The Dubs play better with CJ than with Monta – and it’s really not disputable at this point – not because CJ is better than Monta, but because CJ is a point guard, and Monta isn’t, and yet the team keeps asking him to play point guard. Just like asking Jordan to play center, asking Monta to play point guard (and even when Curry’s on the floor, Monta is often the de facto point guard because the offense runs through him) takes a talented player and puts him in a position where he can not use his talents to the benefit of the team.
Asking Monta to play point guard, the way he’s playing now, isn’t quite as dumb as asking Jordan to play center, but it is, conceptually, the exact same mistake. You a putting a good player in a position where he does not have the skills to do the job.
by Ronaldinho on Mar 5, 2010 10:48 AM PST up reply actions 1 recs
Ok, so I understand what you guys are saying now. Thanks, that was a good explaination.
I still don’t see how this stat is useful then. Did Matt Bonner have a top ten season last year in fitting into the role he was supposed to? How could players that are discussed as top 10 players in the league be rated between 50-150 using this stat? Does it mean that they didn’t play well in the role that they were supposed to?
by freerandolph on Mar 5, 2010 11:40 AM PST up reply actions
Who are these "top ten players in the league?
I mean, if we used Simmon’s trade value column (not perfect, but a place to start) the top ten players in the league are:
LeBron, Howard, Durant, Wade, Byrant, Anthony, Paul, Rose, Roy, and Williams.
Of those guys, only Rose and Roy have bad scores in the rating … but they have something in common. They are players who have been in the league short enough that this stat – which is based on the previous four years – includes their rookie seasons. So you’re still capturing some of the learning curve.
I also suspect – although I haven’t crunched the math – that this stat is imperfect in filtering out the net effects on the team – in other words, a player on a winning team will tend to look slightly better than a player on a losing team. So now you’ve got the 4-year average capturing negative past effects on their team.
Another issue is that it is possible, with a role player like Bonner, to very clearly define their role and have them do nothing but that role. On the other hand, a player who is the central focus of their team is going to be on the court more, and by necessity be used in positions other than the ideal position for them, because you still gain from having them on the court … just not as much.
Maybe an analogy to football will help. Imagine the third receiver on a team, who only gets on the field in passing downs. His average yards per play will be higher than the receiver who plays every down because he’s not on the field for all those two-yard-dives into the line or those short little possession passes. If you’re throwing to him, it’s because it’s a play where you need a lot of yards. He’s only on the field in situations where you want a third receiver. Basketball isn’t quite so clear-cut (because you don’t shuffle personelle quite so much) but the same principle applies.
Another big factor in Roy’s bad rating is that his defense has really suffered this season due to injuries… it’s something the Blazers have been pretty open about. He rated as the 16th best player a year ago, so the system does acknowledge him as a greatly effective player when healthy.
Golden State Worriers: Angst & Analysis
ok you guys have some good points, and now that I understand the stat more I can see it probably does have useful features.
Still seems like it has a lot of problems though.
It’s definitely a weird stat, and it requires a lot of context to be useful… the fact that Drew Gooden and Monta Ellis have similarly bad RAPMs, for example, does not suggest any actual similarities between them in talent, role, etc. But it has its merits, as a way to help you see what’s working in your game plan and what isn’t. C.J. Watson, Third Guard, is working… Monta Ellis, Superstar, is not.
Golden State Worriers: Angst & Analysis
"C.J. Watson, Third Guard, is working… Monta Ellis, Superstar, is not."
I just feel like people will see stuff like this and think “trade Monta! CJ is obviously better, look!” Which I just don’t think is true. So its just weird when a stat rates CJ so much higher then Monta, because it seems like its trying to show that CJ is actually better then Monta.
Its seems that is not actually the point you are trying to make, but it seemed like thats what you were trying to say at first.
The problem is "actually better" is a tricky concept.
Monta is pretty bad this year by every reasonable statistic. We all believe, based on what we’ve seen, that he’s capable of contributing more, but when you say “he’s actually better” is that what you mean?
It is a very reasonable conclusion, based on looking at these stats, that the Warriors would win more games if they shifted some of Monta’s minutes to Watson.
How many minutes? Well, if you were asking me, I would shift some of the minutes, and see how the numbers changed. And if they didn’t change, I’d shift more minutes. And if they still didn’t change, I’d shift more minutes.
Because it may well be that Monta will never be an efficient player for the Warriors again.
This follows the basic precept: when a player does well in a limited role, expand his role to see if he can do well in a bigger one. If a player does poorly in a big role, contract his role to see if he does better in a smaller one.
I gotcha. No, I’m not even necessarily in the “trade Monta” camp… I like the guy, and much like Biedrins, I think there’s a pretty decent chance that he’ll get back on track in a more structured environment. But Monta’s current approach is not a good one. He’s very good, but not good enough to dominate the ball this much. It’s hurting us.
Golden State Worriers: Angst & Analysis
How could players that are discussed as top 10 players in the league be rated between 50-150 using this stat? Does it mean that they didn’t play well in the role that they were supposed to?
When looking at a top guy, it’s not all that important to measure whether he fills his big role as effectively as Matt Bonner fills his small one… it’s more important to gauge whether he’s effective, period. The fact that Durant ranks a tad lower than Anthony Parker is not as relevant as the fact that Durant’s RAPM is +1.5, which is really, really good… he’s clearly very effective in his superstar role.
So let’s look at the showings of big-role, big-scoring, big-name guys… the 20-point scorer types. (As above, we’re looking at the four-year time-weighted data here, which gives more credence to more recent seasons.)
1) LeBron, Dwight Howard, Dirk, Wade, Kobe, CP3, Nash, Garnett, Deron Williams, Pau, Carmelo, Chauncey and Duncan have RAPMs of 2.000 or higher. They all rate as hugely effective in their big roles.
2) Yao, Granger, Vince Carter, Manu, Durant, Aldridge, Tony Parker, Rashard Lewis, Pierce and Gerald Wallace have RAPMs between 1.000 and 1.999. They all rate as very effective in their big roles. Given the standard error, it’s quite possible that a couple of these guys are actually more effective than the guys in the list above them. In the aggregate, however, these guys are probably not quite as effective… still very effective, but less effective.
3) Rondo, Bosh, Z-Bo, Joe Johnson, Roy, Kidd, Boozer, Chris Kaman and Caron Butler have RAPMs between 0 and 0.999… they all rate as effective in their big roles. Not hugely so, but effective… their teams are better for their presences. Bosh, Johnson, Roy and Boozer are dinged for poor D (Roy’s bad showing there very likely has something to do with his hamstring issues, as he was a top-tier guy last season)… Rondo and Kaman haven’t made positive effects on their respective offenses… Kidd, Boozer and Butler rate pretty close to average on both ends. Still and all, this data suggests that their teams are right to give them big roles.
4) Baron, J-Rich, Ben Gordon, Shaq, Jamal Crawford, Andrew Bynum, Derrick Rose, Amar’e, Iguodala, Rip Hamilton, David Lee and Arenas all have RAPM between -0.999 and 0. They rate as somewhat ineffective in their roles. Again, it’s possible that some of them should be a bit above zero, instead; the margin of error here is not tiny. But on average, the system finds these guys to be slight negatives. All of these guys, except for Shaq and Rip, rate as defensive detriments to their teams.
5) Jamison, Maggette, Mayo, and Darren Collison all have RAPM between -1.999 and -1.000. They rate as pretty ineffective in their roles. Jamison, Maggette and Mayo all rate horribly on defense… Jamison actually has the worst defensive RAPM of any player. Collison actually rates as a big offensive detriment. Sounds surprising, but if you look at the numbers, he’s scoring pretty inefficiently and turning it over a lot… he hasn’t been quite efficient enough to justify taking the offense the way he sometimes does. It’s pretty comparable to Monta’s situation, actually.
6) Brook Lopez, Kevin Martin, Al Jefferson, Tyreke, Troy Murphy and Monta have RAPMs of -2.000 or worse… they all rate as extremely ineffective in their roles. Lopez rates as an enormous defensive detriment… that might sound excessive to some, but to me, if you’re playing 37 minutes a game and your team in 6-54, it seems pretty likely that you’re doing something wrong. Martin’s also been a monstrously ineffective presence on D. Jefferson’s not as big of a defensive detriment as Lopez, but he also rates as a slight negative on offense, as well. Again, that makes sense, as I see it… an inefficient post scorer who doesn’t pass out of double teams just isn’t going to help you much. Tyreke’s poor showing is all due to the offensive end, and strongly suggests that point guard is not his best role. He’s moved off the ball a bit more since Kevin Martin left, and the Kings have been doing better… my guess is that he won’t rate this low by the end of the season. Troy Murphy, as always, rates as a harmful addition to a defense. Monta’s struggles have been discussed, to say the least.
There really aren’t many players I’d consider to be really “good” who rate poorly here… there are a lot of 20-point scorers who rate below zero, but most of them are guys who score a lot on bad teams. I think this system sorts players pretty effectively.
Golden State Worriers: Angst & Analysis
The thing about this is that no two roles on two teams are the same, so comparing them seems wrong and inaccurate to me
For example, Tyreke Evans plays on a bad team. Dirk plays on a good team. They both are asked to carry the biggest out of players on their teams, but Dirk has help from Howard/Butler, Kidd, Marion, and some other decent players. Tyreke has Nocioni… So even though you could try to compare them because they both have big roles on their respective teams, it is an unfair comparison because they aren’t really in the same position. I would still rather have Tyreke be the first option on my team then a lot of the players ranked ahead of him.
This stat is interesting, but it just seems like there are a lot of arbitrary things factored in.
Tyreke Evans plays on a bad team. Dirk plays on a good team. They both are asked to carry the biggest out of players on their teams, but Dirk has help from Howard/Butler, Kidd, Marion, and some other decent players. Tyreke has Nocioni…
The Mavs’ supporting cast is definitely better, but it’s not like Dirk has the best supporting cast in the league, and it’s not like Tyreke has the worst. Isn’t it possible that part of the big gap between the teams is that Dirk’s ball-domination is more effective than Tyreke’s?
I would still rather have Tyreke be the first option on my team then a lot of the players ranked ahead of him.
I’d rather have Tyreke than almost everybody ranked ahead of him. I think he’s going to be a hell of a player, and these numbers don’t make me think otherwise. They do make me think that it’s worth moving him off the ball more, to see if it helps the offense. And indeed, the Kings seem to be doing that… Udrih’s now playing big minutes alongside Tyreke, and the Kings have won three out of four.
RAPM isn’t a final verdict on a player or anything like it. But I think it can document some interesting things. And in Tyreke’s case, I think it’s on to something. Really effective star players don’t tend to be on teams as bad as the Kings, just as really effective star players don’t tend to be on teams as bad as last year’s Thunder. Tyreke, like Durant, is a star, and like Durant, he probably will be a damn effective one very soon. But I could definitely believe that he hasn’t been an effective one yet. He’s still figuring some things out.
Golden State Worriers: Angst & Analysis
If there’s a pattern I can believe for these Warriors, it seems to be that playing big men helps the team. Ignoring the ever-problematic standard error for a moment (not something I advise doing as a rule, but just for sake of argument here) it looks like playing Biedrins or Turiaf or Hunter or Moore has helped the Warriors defense some. Shocking? Not at all. Even ineffective defensive bigs who either can’t guard their man (Biedrins) or can’t rebound up to the standards of an NBA big (Turiaf, Hunter, Moore) is better than letting a 2/3 like Maggette play center.
Maggette’s tumbling defensive numbers since coming here might lend credence to that idea, as well.
Golden State Worriers: Angst & Analysis
Speaking of big men & defense - "Let's talk about statistics"
People didn’t give us credit for being as good as we were last season. Personally, I think we won because we had the best team in the league. Some guys talked about all the stars on the other teams, and they quote statistics to show other teams were better. Let’s talk about statistics. The important statistics in basketball are supposed to be points scored, rebounds and assists. But nobody keeps statistics on other important things—the good fake you make that helps your teammate score; the bad pass you force the other team to make; the good long pass you make that sets up another pass that sets up another pass that leads to a score; the way you recognize when one of your teammates has a hot hand that night and you give up your own shot so he can take it. All of those things. Those were some of the things we excelled in that you won’t find in the statistics. There was only one statistic that was important to us—won and lost.
Bill Russell, 1969 – following his championship and retirement
For more, see a post at : http://www.brightsideofthesun.com/2010/3/5/1359473/bill-russells-1969-retirement
Just to give a slightly different angle on things: here’s
The standard errors are a bit bigger here, especially for guys who’ve only been in the league for a year or two. Grains of salt, natch. But a couple interesting takeaways about the guys with larger samples:
- Andris has a +0.617 RAPM (-0.638 offensive, -1.255 defensive; in defensive RAPM, negative numbers are good). Even with his struggles this season and a below-par showing last season, Andris rates as having been a solidly effective player from 2006-10, with his strong defensive presence outweighing his somewhat counterproductive offensive presence.
- Maggette has a -0.098 RAPM (0.641 offensive, 0.738 defensive). He rated as very effective in his two years as a Clipper and very ineffective in his two years as a Warrior; over the four-year span, the two basically canceled each other out, leaving him right around average for the time period as a whole.
- Monta has a
0.970 RAPM (-0.639 offensive, 0.332 defensive)… he rated as a slight detriment on both ends overall from 2006-10. As with Beans and Maggette, the system sees Monta as having had effective years in ’06‘07 and ’07-’08. His current season rates as so monstrously ineffective that it greatly outweighs his earlier work.
For the time period, Biedrins has a decent showing, Maggette a mediocre one, Monta a poor one. But broadly speaking, the system comes to the same conclusion about all three guys: they were effective players through April of 2008, they’ve been ineffective since. Most of Monta’s and Biedrins’s deterioration has come on the offensive side of the ledger, while Maggette’s taken something of a hit on both ends. This suggests two explanations, both of them pretty obvious: Monta and Biedrins miss Baron, Maggette misses being a small forward.
Golden State Worriers: Angst & Analysis
…whoops! First paragraph should read:
Just to give a slightly different angle on things: here’s team-by-team RAPM data with a four-year equally-weighted sample. The data we’ve been discussing is time-weighted with an emphasis on more recent seasons, to try to give you a picture of how effective the player currently is. In this data, each of the last four seasons is weighted equally… this gives you more of a picture of how effective they were from 2006-10.
Golden State Worriers: Angst & Analysis
Comments For This Post Are Closed
Secondary Sidebar
User Tools
"UNSTOPPABLE BABY!"
Golden State Warriors rookie Marc Jackson to the Mavericks' bench, after hitting a lay-up during a 29-point loss (2000)
FanPosts
Community blog posts and discussion.
Recommended FanPosts
Recent FanPosts
Featured Stories
Most Commented
In Association With
Ball Don't Lie - NBA - Yahoo! Sport
- Days of NBA Lives: Wherein Roy Hibbert tap-dances to Maine
- Grizzlies Valentine’s Day cards are the key to love
- Kevin Love would like the NBA to take the MLB’s approach on All-Star games
- Kobe Bryant fires a few hilarious salvos at Jeremy Lin
- Brandon Jennings states the obvious regarding his 2014 free agency, everyone goes nuts
Certain photos copyright © 2012 by Associated Press or Getty Images. Any commercial use or distribution without the express written consent of Associated Press and Getty Images is strictly prohibited.
