APBRmetrics

bchaikin wrote: If I tell you exactly the "things" you want to know, can you tell me exactly how much each "thing" is worth?

i'll certainly try - what exactly did either vince carter or matt bonner do specifically such that they rate better than many players most would think are better?...

My question was hypothetical. I obviously don't know what answer you're looking for. The point is, let's say I did know exactly, could you tell me more precisely than the +/- rating?

Bonner spreads the floor with his 3-pt shooting. How many points per 100 possessions is that worth aside from the points he actually scores himself?

Player X sets "solid screens". How many points is that worth? And so on...

Can you quantify all these skills? If so, please share some of your data.

bchaikin wrote:It's not necessarily sampling error. I've calculated individual year RAPM ratings for the last several seasons and Bonner has had a consistently high rating every year. In Bonner's case, it's likely the case that Pop uses him in exactly the right way to maximize his 3-pt shooting.

so then what are you saying? are you saying matt bonner this season was better than all PFs but dirk nowitzki in terms of helping his team win games (on a per minute basis)?...

Is that so surprising?

if - and i say if - that is what you are saying, then yes it's very surprising...

No, I'm not saying Bonner is "better" than Dirk. What I'm saying is that Pop seems to know precisely how to use someone of Bonner's talents on that team. On another team playing for another coach, I can imagine Bonner having a much lower rating.

YaoPau wrote:I'd like to look at those individual year ratings.

You could be right about Bonner, who knows, nothing is for sure, and there aren't many PFs making 42%+ of their 3s and playing good D. But #2 overall would be very surprising to me... being rated consistently positive makes more sense, and that's where I'd expect him to be.

Players with > 1000 possessions in each of the last 4 seasons (this was done around the ASG, so the average possession count for 2012 is lower ~1800):

One more thing about Carter. In terms of his +/- when playing with the same 4 teammates, he was positive across the board this season:

http://stats-for-the-nba.appspot.com/pm/224.html

Not many players have such consistently high S4PM ratings like that.

"you are focusing on the methodology - i am focusing on it's results."

Did you say or ask anything about the A4PM factor level results?

No, you questioned the overall methodology and the overall A4PM results (because you appear to wish to focus on and criticize the "blackbox") of a few carefully selected players. Nothing at all on the 8 Factor level results of any player, including Carter.


"are detailed estimates the same as player specifics?"

No, detailed estimates in this context, in my view, when they have been provided, are the factor level estimates which you entirely ignored.


"is asking for player specifics not the same as trying to find detailed estimates?..."

No, in my perspective as explained above.

While acknowledging they are estimates and a modest number of them across the entire league may be outliers, which of Carter's factor level estimates do you reject? On what basis?

The estimates or really the ranges around them, with a confidence interval determined by the sample size are somewhat helpful new information to consider, more likely to be close to the center point of the range than far from it. Helpful new information especially for the Factors that the boxscore is entirely silent on (shot defense) but also for the indirect impacts of every Factor.


The rest of your post is a unenhanced simple repeat of your first post, and since I've already responded several times to the extent I am able and fairly represented what the numbers are and to no affect on your posts or posture, I may not going to add anything further on that.


My posts were from the very beginning intended mainly for the other readers. I've said what I wanted to say for their consideration.

I obviously don't know what answer you're looking for...

again, the answer to this question - what exactly did either vince carter or matt bonner do specifically such that they rate better than many players most would think are better?...

Bonner spreads the floor with his 3-pt shooting...

great, that's a start...

but why does bonner spreading the floor with his 3pt shooting make him the 2nd best PF in the league? a number of players took a similar amount of 3s in similar minutes that shot just as well if not better, but this version of APM does not list them as the 2nd best in the league at their position. for example PF steve novak took more 3pters in less minutes than bonner did, shot better on them, scored at a much better per minute rate, and was more efficient on offense, even with a higher rate of turnovers per touch. if spreading the floor is so important, why is novak rated so much lower in this version of APM?...

what is it that PF matt bonner did that PF steve novak did not do to rate so much higher?...

What I'm saying is that Pop seems to know precisely how to use someone of Bonner's talents on that team. On another team playing for another coach, I can imagine Bonner having a much lower rating.

are you saying that if bonner had the identical stats on another team he might not be the 2nd best PF in the league?...

The rest of your post is a complete unenhanced repeat of your first pos,

yes i've asked the same questions repeatedly about the same specific players that you have not answered - what exactly did either vince carter or matt bonner do specifically such that they rate better than many players most would think are better?...

and since I've already responded several times to no affect on your posts or posture, I may not going to add anything further on that.

did you somewhere here say specifically what carter and bonner did to rate so high in this version of APM, compared to what other players rated lower perhaps did not do?...

why is it when someone questions the results of APM and it's numerous renditions all it's proponents all of a sudden clam up?...

again -

why does vince carter rate so high in this version, or any version, of APM?...

why does matt bonner rate so high in this version, or any version, of APM?...

are these not legitimate questions to ask?...

"why does vince carter rate so high in this version, or any version, of APM?...

why does matt bonner rate so high in this version, or any version, of APM?...

are these not legitimate questions to ask?..."


They are legitimate questions and I have never said that aren't.

"Why" in great detail can't be stated with certainty, from the overall or factor level APM model, and no one claims (and I am not) that they can. You knew that from the start, so you are confident in your posture. It is the "where" and how much at each of the factor levels that is being estimated by A4PM. A4PM is designed to rate him with moderately good accuracy in most cases in 8 specific areas that are accepted as the cornerstone elements of the game. You continue to show absolutely no interest in the where or the how much at the factor level, again refusing to offer any analysis of those detailed results.


If you want to understand Bonner frame by frame or coach others to try to achieve his estimated impact in the exact same ways, A4PM won't do that all by itself. Criticize the metric for not doing more than it was designed to do if you want. It is unfortunate but not surprising that the metric by itself is not that powerful, but nothing else is fully satisfactory on its own either. As one extra tool among many it can be generally helpful over not considering it, IMO.

Stick to incomplete and often misleading boxscore stats that everyone else has if you want. Count discrete non-boxscore actions that most teams don't have data on, as much as the team will spend on it. That is probably worthwhile (when used as a building block alongside other information), but of undetermined value. Without an APM-type approach (and SPM also uses an APM approach) you don't "know" the specific value of any those non-boxscore actions, those possible "whys", so what have you specifically and concretely achieved? You haven't even tried to estimate those values by any analytic approach. You can assigned subjective values if you wish but that seems fraught with as much or higher probability of imprecision than APM.

Ignore the APM class of tools if you want or feel you should or have to. Free to use and not use what you want. But nothing you have offered changes my view that it is better to look and consider APM data alongside everything else. Of course I have made that points a few times, enough or too much I think.

I'll close by saying that if a team had complete, accurate non-boxscore information for a list of such actions thought to be important or possibly important, it probably could build that additional detail into an APM model (or some other approach)and find estimated value for those whys. The expertise it out there. If a team really wants to know the value of the "whys", hire lots of videotape counters of non-boxscore actions and then pay Jeremias or Evan or somebody else to build those things into an even more complicated model than the factor level model and find their estimated values at a finer degree of detail. Split the value of an action into credits for 1 or more preceding actions and the final action. And then if you want to analyze the data, call me or somebody else who does it and thinks they can find useful stuff in it when considering it alongside everything else. And then decide if anything useful has been produced, or stumbled upon, or hinted at and see what doing that yields overall vs the teams that don't do that kind and level of work.

EvanZ wrote:One more thing about Carter. In terms of his +/- when playing with the same 4 teammates, he was positive across the board this season:

http://stats-for-the-nba.appspot.com/pm/224.html

Not many players have such consistently high S4PM ratings like that.

Interesting stuff, Evan. I did a little filter of the players who consistently ranked high, and Bonner's name could not have stuck out any more haha. Actually... a weighted average of those RAPMs since 2009 produces a pretty impressive ranking of players, Bonner aside. (1) LeBron (2) Bonner (3) Garnett (4) Wade (5) Howard (6) Dirk (7) Duncan (8) Paul

I like your theory behind it, that he's really effective in certain situations. But I'm not sure what those situations are. Here's his list of lineups this year ... he always plays PF, but he seems to play with all his teammates. He also played about 40% of the season, so it's not like Pop was being too choosy. Hmm..........

Evan, what's your minutes cutoff for keeping players in the RAPM model?

YaoPau wrote:Evan, what's your minutes cutoff for keeping players in the RAPM model?

No cutoff. Maybe I could reduce the error by introducing that, but I haven't bothered to do it yet.

Interestingly (maybe), the average A4PM rating of all players with <500 possessions this season is -2. So even with few possessions, it seems to be picking up close to the replacement value that would be used. This is to say, I'm not sure it would make a significant difference.

bchaikin wrote:I obviously don't know what answer you're looking for...

again, the answer to this question - what exactly did either vince carter or matt bonner do specifically such that they rate better than many players most would think are better?...

Bonner spreads the floor with his 3-pt shooting...

great, that's a start...

but why does bonner spreading the floor with his 3pt shooting make him the 2nd best PF in the league? a number of players took a similar amount of 3s in similar minutes that shot just as well if not better, but this version of APM does not list them as the 2nd best in the league at their position. for example PF steve novak took more 3pters in less minutes than bonner did, shot better on them, scored at a much better per minute rate, and was more efficient on offense, even with a higher rate of turnovers per touch. if spreading the floor is so important, why is novak rated so much lower in this version of APM?...

what is it that PF matt bonner did that PF steve novak did not do to rate so much higher?...

Steve Novak also had a very high rating, did he not (btw, so did Anderson, Frye, Curry, Turkoglu, Radmanovic, etc.)? If you expect that every player that shoots 3's at a high rate should have exactly the same rating and they do everything else exactly the same, I'm not sure I could give you a great answer.

What I'm saying is that Pop seems to know precisely how to use someone of Bonner's talents on that team. On another team playing for another coach, I can imagine Bonner having a much lower rating.

are you saying that if bonner had the identical stats on another team he might not be the 2nd best PF in the league?...

I told you already I didn't think Bonner was "the 2nd best PF in the league". Why would you keep insisting that is the case?

I'll give you the same answer I gave Mike earlier: Bonner has the 2nd highest rating according to this metric. That is a fact. Take it FWIW. YMMV. It's not my responsibility to convince everyone that every rating perfectly measures player value.

Would you advise Pop to play him less? Or pay him less? He makes $4M/yr. Seems like a good deal to me.

This is a good thread, and I don't think BobC is trying to denigrate advanced models so much as to play devil's advocate and ask some searing questions.

The list of consistently great RAPM over the last 4 seasons --

(1) LeBron (2) Bonner (3) Garnett (4) Wade (5) Howard (6) Dirk (7) Duncan (8) Paul

-- provided by YaoPau, is pretty good, but lacking Kobe, Gasol, Durant, while including Bonner at #2 ... it's indeed a head scratcher.

Bonner only plays 15-20 minutes a game. It's likely that Pop uses him as a matchup dilemma for opponents. It's reasonable to wonder how his RAPM would change if he played 30 or more mpg, with the Spurs or another team.
Conversely, one could use DWade or any of the others quite selectively, and more effectively on a per-minute basis. Kobe could be the ultimate "microwave" type.

Still, it's almost disingenuous to insist on knowing the "why" of an APM rating. It is what it is, like the price of gas. You can discuss "the why" out of interest, but that's just the price, today. Take it or leave it.

Unfortunately, by the time a 5-year history of a player's APM has given anyone a clear picture of his effectiveness, he's often not the same player that the interval has measured.

bchaikin wrote: why does vince carter rate so high in this version, or any version, of APM?...

why does matt bonner rate so high in this version, or any version, of APM?...

are these not legitimate questions to ask?...

Those are legitimate questons. And that has indeed not much to do with sampling errors.

Let us take a look at Carter. Carter is big for a SG, he does not have problems to compete with any kind of other SG. That helps his defense, especially when guards are trying to post him up. Carter is one of the best in terms of defending the post-up. Carter has still long arms and is pretty quick on his feet. He can cover ground and that helps to defend shots at the perimeter. His opponents does not have an easy way to get by Carter, Carter stays in front of him and that forces more contested shots. Compare Carter and Bryant for example in terms of closing out on a 3pt shooter, Carter is better at that. It is also the case that Carter can defend the p&r pretty well. In fact the ball handler is more often in trouble. Overall Carter's defensive abilities and his size is causing a clear positive impact on the defensive end. Tougher shots for the opponents with a clearly lower conversation rate. That explains Carter's positive defensive RAPM. The value comes mainly from him being matched up with SG. When he plays SF, his quickness in comparison helps to limit drives to the basket. But he also doesn't have the size advantage anymore, which takes value away.

His value on offense comes from shooting, passing and from attacking the basket. His shooting range helps spreading the floor. Especially in comparison to Marion at SF, the Mavericks with Carter are becoming a much better offensive team while Carter plays the SF spot. The floor spreading is not just seen by clearly better 3pt shooting, but opens up the middle also for Nowitzki and makes it easier to create those midrange shots. Carter's passing helps to limit turnovers, provide better ball movement, which overall leads to a better team offensive efficiency.

No idea, but those things seem to me pretty obvious when watching the games. RAPM now can quantify the effect and we can also check the various other On/Off Court numbers in order to learn from them or backing up things we see.

Also, one of the biggest mistakes by Carlisle in the playoff series was not using Carter as SG and instead trying to let Terry/West defend Harden. Harden has much bigger problems with bigger guards, while Carter is also capable of defending the p&r ball handler really well. No idea what the thought process was behind all that, but I'm pretty sure that this playoff series could have ended much different, if the Mavericks would have used their most successful lineup. West-Carter-Marion-Nowitzki-Haywood would have been the best defensive lineup against Westbrook-Harden-Durant-Ibaka-Perkins. Offensively it would have meant going through Nowitzki in each possession, but he can handle that. Using lineups with Terry/West against Westbrook/Harden took the advantage of the Mavericks (size) completely out. Not a good decision at all, and I would really like to get an explanation why Carlisle didn't use his size advantage.


For Bonner we see another effect by floor spreading and 3pt shooting from a big. Such stretch PF are messing up the opponents defensive rotations and help defense, because very often the help is provided by the bigger and more agile players like Garnett, Ibaka, etc. pp. Those players can now choose to stay on the perimeter with a guy like Bonner and thus are not available for their usual help defense or they can leave Bonner open and risk a open 3pt, because the rotation/help brought them too far away from Bonner in order to close out in time.
Players like Bonner will also not become defensive liabilities that easily, because such players tend to reduce the turnover rate for the team, which means less fast break opportunities for the opponents, and they also are set to go back on defense quickly, instead of crashing the offensive board. Such behaviour leads again to less "easy basket" opportunties, either due to the first or secondary breaks. And while they are not good in terms of halfcourt defense, they are still big enough to bother their opponents. Bonner for example was still a pretty good defender against Dirk Nowitzki, because he does not go for every fake, stays on the ground and tries to defend the position much more than the shot. With the given set of rules, that is the best way to defend. The shot defense becomes less important, because challenging the shot can too easily result into a foul and free throws. When a player defends the position alone, he not only forces his opponent to a location from which a successful shot is less likely, but is also in a better position for the defensive rebound. Well, while Bonner is not great, his way to act on defense does not become a negative impact per possession. And as we all know basketball is decided by the average efficiency, not by the way the efficiency is achieved.

I hope that helps to clear things up in regard to those two players.

Regarding the floor spreading and the effect, two impressive examples we can use: Vladimir Radmanovic instead of Ariza playing for the Lakers in 2008. The spacing with Radmanovic in the TPO and in p&r situations is so much better and opens up opportunties for Bryant and Gasol.
Last season on the Mavericks, watch how Barea plays completely different with and without Nowitzki on the court. With Nowitzki he has wasy more room to go to the basket, without him he takes far more contested long shots.

Those are things not captured by the boxscore, but can be seen in the +/- based analysis.

Mike G wrote: The list of consistently great RAPM over the last 4 seasons --

(1) LeBron (2) Bonner (3) Garnett (4) Wade (5) Howard (6) Dirk (7) Duncan (8) Paul

-- provided by YaoPau, is pretty good, but lacking Kobe, Gasol, Durant, while including Bonner at #2 ... it's indeed a head scratcher.

No head scratcher at all, because that is not a list of players perceived as the "best", but players helping the most per possession in their respective roles. Something most people gets confused, because they think a ranking is always meant to say who is the overall best.

You're right, it was just a bug.
And that was a really great overview, assuming you've actually observed these tendencies.
Carter played 25 mpg, coming off the bench about 1/3 of the time. He's still only on the court half the time, so we can't say whether he's as good for 30 or 40 minutes.

Mike G wrote:You're right, it was just a bug.
And that was a really great overview, assuming you've actually observed these tendencies.
Carter played 25 mpg, coming off the bench about 1/3 of the time. He's still only on the court half the time, so we can't say whether he's as good for 30 or 40 minutes.

I'm not suggesting taht he should play 40 minutes per game, just 30 minutes while mostly used as SG instead of using him against Durant as SF. Carlisle should have trusted the size advantage and play Haywood more. Yeah, the guy has bad hands and makes questionable decisions on offense, but that just means that he can't be used like Tyson Chandler on offense. Also, losing Odom meant that the PF backup was Marion again, which is a pretty bad idea. Marion can't compete with those bigger guys and his real strength (defend at the perimeter) is taken out. Offensively Marion depends much more on open lanes to the basket, using him as a post-up guy is not working well. Maybe use Haywood/Mahinmi more as real C, and bring Wright as a PF backup for the few minutes Nowitzki is out. If that doesn't work, give Cardinal the backup minutes on PF instead. You can still use Cardinal as the decoy like Nowitzki, then, if the opponents using Cardinal's defender to help, move the ball quickly to give Cardinal the 3pt shot. Marion shouldn't have played a single second as something else than SF.


Edit: Btw, just that nobody thinks the stuff is just said in hindsight, here is my post on RealGM before the series started:

he Thunder over the course of the season were pretty clearly working better as a team than the Mavericks. If we go by the average performance level of each team against all teams, that series should be won by the Thunder. Two things are speaking against the Thunder here: First, the matchups are in favor of the Mavericks. With West against Westbrook and Marion against Durant, the Mavericks have the right players to keep them in check. That means Harden must beat his opponent constantly. But in this season he will not go up against Terry and take him to school with his height and strength advantage, but against Carter, who can match the size and strength. The frontcourt of the Thunder does not scare the Mavericks at all. Neither Perkins nor Ibaka nor Collison can make something happen offensively. The Mavericks will be fine defensively with all of their frontcourt players they have next to Nowitzki (Haywood, Mahinmi, Wright). The defense of the Thunder is questionable, and as soon as the Mavericks can have their ball movement going, they will score their points. Too illustrate that, we can use some data from this season. During the time Durant and Nowitzki were on the court at the same time, the Mavericks outscored the Thunder by 7 points per 48 minutes. The lineups with 5+ minutes over the last 2 seasons against the Thunder outscored them in average by 7.7 points per 100 possessions (only possible lineups in the playoffs this season). The Thunder on the other end for all those lineups were outscored by 1.4 per 100 possessions. So, the most used lineups against each other as well as the time when both teams best players were on the court at the same time really speaks for the Mavericks.

When the Mavericks are staying healthy and play to their strength, they can beat the Thunder. In that case the Mavericks will likely win in 6.

http://forums.realgm.com/boards/viewtop ... #p31539612

The Mavericks stayed healthy, but they did not play to their strength, but rather let Scott Brooks dictate the matchups. The Mavericks once did the same mistake in 2007, when they decided to go away from their best lineups and play small, instead of using their size advantage to beat the Warriors.