A Substitute for War

Basketball philosophy

Posts Tagged ‘+/- Statistics

Kevin Love, Dirk Nowitzki, Spencer Hawes, Andrea Bargnani and “Vertical” vs “Horizontal” rebounding

with 4 comments

Kevin Love, the 5th pick

Image via Wikipedia

Measuring an individual’s impact on team rebounding is a tricky, mystifying subject. In a way, rebounding is analogous to scoring in that it’s not as simple as measuring points per game and assigning impact from that alone. When a player takes a shot, he may be infringing on another player’s effectiveness or drawing enough defensive attention to improve it. Corey Maggette appears to get in the way of teammates by selfishly stopping the ball, Dirk Nowitzki appears to help teammates greatly by drawing defenders out of the paint. Even staticians favored stats like True Shooting %, measuring points per shot, can miss the whole picture.

Rebounding has even less statistics differentiating it. We have rebounds per game, Rebound %, plus/minus stats… and that’s about it. The rest we have to judge ourselves.

There are rebounders who have higher numbers on the boards who draw skepticism. Among those are Kevin Love, David Lee and Marcus Camby. The reason for this is that it looks like that instead of boxing out an opponent to get a rebound, they prefer to chase the ball to where it’s going to go. This is fine, but if they guess wrong, the man they’re not boxing out is open to get it. If they do get it, they may be taking a rebound a teammate was already going to get – thus despite being credited with the rebound, they are giving their team no additional value on the play. Thus the defensive rebound on this play is a misleading stat. On the other hand, if a player like Dirk Nowitzki shuts down his opposing PF’s bid to grab a rebound by boxing out and the ball is going towards him, but at the last moment a horizontally moving teammate grabs it, the player who grabbed it gets the rebounding credit, but it’s likely Dirk who deserved it.

Read the rest of this entry »

The Nash Disequilibrium, or Why I Use +/- Statistics

with 9 comments

Image by OakleyOriginals via Flickr

I felt the need to write this as a result of the article I wrote on Kobe Bryant and his adjusted +/- statistics this season. That article showed my perspective as someone who uses these stats – this one gets into why one should use them.

I’m a math kind a guy. I’ve been making statistical rankings of basketball players and other such trivia for forever. When the internet was first reaching prominence, many did see how they would use it, though they actually did end up using it obviously. I was dying for it though from the start. To have access to data like basketball-reference.com has is like a geek nirvana for me.

Now, I always knew that in basketball, the stats didn’t cover everything, but I always figured that what they missed was relatively small and not ridiculously biased. And then in ’04-05, I found myself utterly fascinated by the Phoenix Suns and Steve Nash. Every metric I’d ever come up with or ever seen said that Nash wasn’t the best player on that team, but my common sense just found this absurd. He was the one directing that offense, not the scorers. The team had launched forward far beyond what anyone expected because of an improvement in team offense that was completely unbelievable, and the team had made but one major change and one other major decision: Sign Nash, and put the ball & decision making in his hands.

And Now for Something Completely Different

Read the rest of this entry »

Written by Matt Johnson

March 26, 2011 at 12:04 am

Letting the Lamppost Illuminate

with 22 comments

Image by motiqua via Flickr

He uses statistics as a drunken man uses lamp-posts…

For support rather than illumination

Andrew Lang (1844-1912)

I’ve been rather crucified lately for having Kobe Bryant so low on my MVP list, and I completely understand why. He’s the player many consider to be the best in the game, his team has an elite record, and as we approach the final stretch of the regular season, both he and his team appear to be entering back into their familiar, championship-winning groove. How can he be not anywhere near the top 5 in the MVP race?

And I answer, knowing that my response hurts my credibility among quite a few not simply because they disagree with me, but because a significant influence on where I place Kobe is his poor performance in adjusted +/- statistics (APM).

Remaining Sober

Read the rest of this entry »

Written by Matt Johnson

March 19, 2011 at 9:52 am

Kobe Theory: Adventures in Distorted Probability

with 13 comments

kb_0563cf

Image via Wikipedia

We saw some fantastic, hard-hitting articles this week about Kobe Bryant‘s clutch reputation. I’ll go over them briefly, and then just talk about what people’s perceptions say about people in general, and running a basketball team specifically.

Henry Abbott at ESPN’s TrueHoop does a great job of just summarizing the fact that despite Kobe‘s reputation as the ultimate clutch performer, all the evidence says this is not the case.

Kelly Dwyer at Yahoo’s Ball Don’t Lie shows some moxie in making clear that he’s quite comfortable saying that if NBA GM’s don’t see the problem with Kobe’s stats, then the GM’s are in the wrong.

Zach Lowe at SI’s The Point Forward chimes in, but also emphasizes the larger trend that NBA offenses in general do terrible in the clutch. Scoring at far lower rates than they do in the rest of the game.

All very cool stuff. Here’s the most telling fact as I see it: People who reject the numbers here do it by dismissing statistics as not being as valid as what they see, which is an argument that often has merit, but is not valid at all here. So, Why isn’t it valid? and Why are people like this? Read the rest of this entry »

The Carmelo Conundrum, how good is he?

with 7 comments

Carmelo Anthony during an NBA preseason game i...

Image via Wikipedia

Carmelo Anthony has been a lightning rod for debate for a long time, but with his recent decision that he doesn’t want to play for the perpetual playoff team Denver Nuggets any more, we’ve reached a local peak in activity.

A few days back political statistical superstar Nate Silver wrote an article about Melo.  I’m a big fan of Nate’s – but it really was a terrible article.  First off the title was “Why Carmelo Anthony is the Ultimate Team Player”.  Aside from the fact that “ultimate team player” isn’t something that can be measured by statistics completely, it’s not like Silver actually went about comparing him to other players and showing Melo’s superiority.  I really hope that the title choice was done by someone other than the statistician.

I focused though on Silver’s specific analysis.  He’s saying that teammates do better at shooting efficiently with Melo.  Taken at broad strokes, this is essentially a +/- argument using shooting efficiency instead of the scoreboard.  While such an analysis can be useful because it is specific enough to suggest a particular means of impact, if we’re talking about a player’s overall impact it’s inherently weaker than what we call +/- statistics because it factors in only one part of the game, and so I said as much:  Whatever Melo’s factor on his teammates shooting, his total impact based on the +/- stat of greater scope and at least as much credibility is not anything like the true superstars of the world like draftmates LeBron James and Dwyane Wade.

Today Henry Abbott over at TrueHoop has an article about the debate about how good Melo is, and it’s pretty good.  It mentions things from both sides.  Among other things it mentions a more clean take down of Silver’s analysis:  Silver analyzed this by measuring efficiency of players when they were on Melo’s team compared to on other teams, but if you actually look at his on/off numbers on 82games.com, you see that for both this year and last year, his team shoots better when he’s on the bench.

I want to respond to some specific points from Henry’s article:

Read the rest of this entry »

Written by Matt Johnson

January 18, 2011 at 11:01 am

Why the faith in OPS?

leave a comment »

First off, I’m not a sabermatrician. I won’t pretend to be in their league in advanced baseball knowledge.  But I have a problem with their most widely used advanced batting stat: OPS.

OPS stands for On-Base Plus Slugging. In short, On-Base Percentage measures how often a player gets on base and Slugging Percentage accounts for the value of extra bases. Both these stats are fine. But I don’t see the basis for simply adding them together 1 to 1. Since they clearly aren’t worth the same, this immediately makes the stat flawed.

The consensus is OBP is worth more. It’s more important to not waste one of the precious 3 outs than advance more bases when you get on. The number I’ve heard is OBP is worth 1.8x more, though some have estimated as high as 3x and up. If true, OPS is very off. Furthermore, SLG% itself is also flawed because it weighs singles and walks the same. Singles are worth more because they advance players on 2nd and 3rd without a force from 1st, making them much more potent for scoring runs. Once again this just trips up any pretensions of accuracy for OPS.

The real basis for it is adding them together happens to coorelate pretty well with offensive production. So under the guise of “it works”, it’s stuck. I’m not buying it. Stats should be equated for reasons making sense on their own, not just because they give us the good looking answer. We shouldn’t accept flawed stats because it gives us an answer we want. Especially in baseball, a sport where eventual statistical exactness is not only possible, but realistically attainable.

In truth, OPS is a great ballpark stat. The basketball equivalent is John Hollinger’s Player Efficiency Rating (PER) or Win Shares (WS). In one number it can tell you who the MVP candidates, all-stars, and mediocore players are. It should not be treated as more.

Recently OPS+ was introduced, adjusting OPS for ballparks and giving OBP 1.2x weight. It’s not enough. It still doesn’t make sense without relying on the conclusion looking right. To me the answer is a multiplication equation involving OBP and SLG%, not addition.

My quick glance across the internet has shown me fringe sabermatricians have realized the faults of OPS and tried to develop better, multiplication based batting stats for years. Much credit to them. I’d only say not to treat OPS as gospel or the end of the line because it’s clearly not. Baseball is a sport where we can truly exact offensive value. It’s all on the paper. To treat simply adding together OBP and SLG% together 1 to 1 as enough, is selling ourselves short.

Looking at Nash’s Crazy +/- Numbers

leave a comment »

With Phoenix Suns making a huge trade I wanted to take a moment to talk a bit about Steve Nash in terms of +/- statistics.  The Suns made the trade because of dissatisfaction with the way the team is playing, and with a sub-.500 record that’s understandable.  While I’ve yet to hear many people talking about the Suns’ struggles as a reflection on Nash, neither am I hearing Nash getting much attention in the MVP-sense.  I understand that to some degree as I currently don’t have him in my POY top 10, but it should be a hard decision for people.

According to basketballvalue.com, Nash is currently #1 in the league in Adjusted +/- for the year (and for the last two years combined actually).  His rating is +29.14, there are only 5 guys total in the league with a rating north of +20, and the other guys (Chris Paul, Dirk Nowitzki, Chris Bosh, Pau Gasol) are all on much more successful teams.  Most people reading this blog probably have the general understanding that adjusted +/- means roughly how much better the team does with you compared to without you after making some statistical adjustments to reduce noise.  But what does it mean when a guy on a mediocre team does well by this metric?  How do you compare him to guys on great teams?

There’s no clear cut answer to this, but I think it’s helpful to look at some raw +/- to get a more concrete feel for things.  nba.com has raw +/- numbers, i.e. how much in total your team has outscored opponents with you on the court over the season, and here’s how the Phoenix Suns look right now:

1. So Nash is at over 100, and 8 of the 9 other rotation players have a negative.  Stunning these are the guys Nash is out there with, and when he’s with them the team is still doing great, but virtually to a man, things are going terribly when Nash is out.

2. The only other positive rotation player is Frye, and he’s just barely above zero.  Nash’s lead over his nearest teammate is 95 points, which is the biggest gap between a team leader and his teammates of anyone in the league (Dirk is 2nd with an 81 point lead).

3. In case you’re wonder how a +109 net stands up, well to some degree it’s modest.  Dirk’s at +255 for example.  But check out some elite point guards on more successful teams:

Deron Williams, +51

Derrick Rose, +106

Russell Westbrook, +21

Chris Paul, +117

It becomes pretty easy to make the case that Nash is only ranked below these other point guards in MVP races because of how bad his teammates are.

Now, with all this said, I don’t have Nash in my top 10 right now.  The truth is I’m not entirely convinced by the metrics I’m showing here myself, with one major reason being that Nash has missed some time, and that doesn’t come into the above line of reasoning at all (and clearly it needs to fit in somewhere).  What I am adamant about is simply that when you look at these results, it should make your jaw drop.  You should be dang impressed with Nash, and you should consider that truly, the only thing keeping Nash from leading a contender is a decent supporting cast.  I don’t know whether the Carter & Gortat trade will do the trick, but I think it’s worth a shot.

Written by Matt Johnson

December 19, 2010 at 1:40 pm

Gasol, Bosh, and +/- Statistics

leave a comment »

I’m getting a lot of Heat for continuing to have Pau Gasol ahead of Kobe Bryant in my POY/MVP rankings now that Gasol is clearly in an extended slump.  There are multiple factors involved in my decision making process, but something I haven’t heard any one deal with regarding this debate are the +/- statistics.

For those unfamiliar, +/- statistics rate players based on how well their team does when they are on the court compared to when they are off the court.  It was invented in raw form in hockey, but it’s basketball that has really gotten sophisticated with it.  The great thing about +/- stats is that they literally catch everything relevant that happens on the court.  That great screen a guy set that let to the open shot?  Counts.  That great pass before the pass that got the assist (aka the hockey assist, so named because they actually give that guy an assist in hockey)?  Counts.  Box score based stats are forever at the mercy of what scorekeepers do and don’t track.  The gambling thief who keeps trying and failing to get and assist, leading to the opponent getting an easy buck is doing nothing wrong according to traditional box score stats, but +/- stats catch what he’s doing.

Of course there are weaknesses to the stat as well.  First, the very fact that it catches everything when we don’t have box score stats for everything means that it’s advantage in coverage is a disadvantage in explanation.  This opens the door for skeptics to say that any number of factors could be involved in stat’s results, and they’ve got a point.  If a stat says I did something good just by being on the court when my team’s star gets into a zone, that stat is certainly not perfect.  Second, +/- stats tend to be noisier than traditional box score stats.  The guy who scored 30 points efficiently in a game can be said with almost complete confidence to have had a great game.  You can’t say a guy had a great game simply because +/- said he did well, and when we get into the most sophisticated +/- stats, statisticians strongly prefer to use 2 or more years as the sample size.

Also, related to the “black box” aspect of these stats, is the fact that if you point to them without giving an explanation for the cause of the results, you’re admitting that you don’t really know what’s going on.  This is something that bothers everyone.  It’s hard to stick your neck out there just based on a possibly unexplainable number.  It’s easy to chafe against something that essentially calls your understanding of the game into question.

So, this year, Pau Gasol’s raw +/- (how much the Lakers have outscored opponent’s while he’s on the floor) is +265, while Kobe Bryant is only +148.  Every more advanced metric we have along these lines says the same thing – the Lakers are more dependent on Gasol this year than Kobe, and this is not something we’ve seen in previous year.

Also of note, Gasol’s raw +/- is 2nd in the league.  Who is first?  The often mocked Chris Bosh who is well ahead of his more respected teammates LeBron James and Dwyane Wade.

People who aren’t sold on these stats use this as ammunition against the stats, “See, it’s meaningless I tell you!”.  I can’t help but marvel though and how those numbers actually fit with other things people are saying.

The Lakers, everyone agrees, lack depth at big man.  Bynum’s been injured.  They signed an old Theo Ratliff out of desperation.  They’ve just yesterday traded for an old Joe Smith out of desperation.  Do we really think it’s a coincidence that when the Lakers are desperate for big man, the team appears to be extremely dependent on their star big man (Gasol), who just happens to be playing by far the most minutes of any of their players?

The Heat, everyone agrees, have 2 superstar perimeter players capable of being elite scoring option and offensive decision makers who have had some trouble blending their talents together.  We also know that the Heat are struggling to find quality big men.  Do we really think it’s a coincidence that the guy who the team falls off when he’s not on the court, is the quality big man they do have?

Now if you want to make the argument that there should be more to accolades than a statistic that can be so shaped by the depth of the team the player plays on, I think that’s a perfectly reasonable point to make.  I do think though, that it’s unreasonable to dismiss what +/- stats have to say here.

Written by Matt Johnson

December 15, 2010 at 11:23 am

2010-11 NBA Predictions: ROY

leave a comment »

This is a tougher award to judge than I think anyone realizes.  Here’s the thing, if you asked most people how they judge the ROY compared to the MVP, I think they’d probably say they think about them similarly.  The MVP of the rookies if you will.  However, if you actually look at ROY’s from a +/- perspective, you start seeing some major problems.

Now let me elaborate for those of you not as stat-obsessed as I am.  +/- statistics simply measure how well how many points more than your opponent are scored while you’re on the court versus when you aren’t on the court.  It’s something that came from hockey, but in the last decade basketball statisticians have really taken it to the next level.

Analyze +/- data, and what you’ll find that pretty much any guy considered a strong candidate for the MVP does really well in the stat.  However, if you apply the same stat to ROY candidates, you’ll find chaos, and if you think about it, that makes perfect sense.  Rookies typically are not guys who completely turn around their team so much as they are guys considered to have great upside that the team decides to build around.  They’ve earned their primacy based on future value rather than present value.

If you don’t believe me, let’s consider LeBron James as a rookie.  Read the rest of this entry »