Thursday, October 14, 2010

Is Bargnani a bust?

This article uses the Wins Produced metric to discuss the productivity of NBA players. This metric is based on box score statistics that are adjusted for other factors including pace, position and team. A general scale is given for these, and links to advanced explanations are listed at the bottom of the page.

Wins Produced per 48 Minutes (WP48) – The number of wins a player produces in 48 minutes of play. 0.100 is average and 0.250 is considered the “superstar threshold”. A player a WP48 of 0.000 produces no wins, and any player with a negative WP48 produces a negative number a wins (or, if you prefer, a positive number of losses)

Over on TrueHoop today, Zach Harper writes that we shouldn't call Andrea Bargnani a bust just yet:
In the last 10 drafts, Oden, Andrea Bargnani and Kwame Brown are probably the only players considered a bust. While Oden’s injury issues and Kwame’s issues with being able to play NBA basketball at a high level are the reasons for their bust label, trying to determine why Bargnani is a bust might be as simple as figuring out if he’s even been playing the correct position.

The reason I bring this up is because of the reputation Bargs has garnered in his short career. Is he simply a case of Kwame Brown, in which he’s just not good enough? Or is he more like an Andrew Bogut-type of first pick that was playing out of position early on and needed the right fit to start to blossom? 

And additionally, Zach Lowe of The Point Forward, after talking at length about how poorly Bargnani has played thus far in his career, writes:
Still, through four seasons, Bargnani has proved to have a single valuable skill: an ability to score from the perimeter, both off the catch and off the dribble. That’s a great and unique thing for a 7-footer to be able to do, but it will not be enough anymore. The Raptors need more from Bargnani, and it’s unclear if he can give it to them.

I have already taken a look at Bargnani's production before, but sure, let's go ahead and revisit this issue.

"Bargnani is not a bust"

A first overall pick is expected to not only be a good player, but a great player, and, hopefully, better than any other player in that year's draft. Although NBA teams don't always get it right, the productivity of first overall picks is much higher than the productivity of any other pick. Arturo Galletti has done a lot of work on the productivity of draft picks, and he determined that the average WP48 of the first pick (for the years 1977-2006) is 0.169. The next most valuable pick is the third pick (yes, ahead of the second) at a WP48 of 0.133. The best player in the draft usually has a WP48 around 0.278, and the 20th best player usually has a WP48 of around 0.048. With that in mind, who were the last 12 first overall picks in the NBA draft (12 because John Wall and Blake Griffin have yet to play in a regular season game), and how productive have they been over the course of their careers?

Hmm. So Bargnani doesn't really compare very well to those players, or to what we should come to expect from a first overall pick? Who would've known. Bargnani and Bogut are not even remotely comparable (Bogut has almost 34 wins on him). In his first four years, he actually managed to produce a number of losses even greater than the number of wins that Kwame (who is popularly - and deservedly - regarded as a bust) produced in his first four seasons. You think that's bad? It gets even worse for Bargnani when you consider that:
  • Bogut, Howard, and James all improved significantly in the years following their 4th season
  • Rose and Oden improved on their rookie years during their second seasons
  • Blake Griffin is on the cusp of what appears to be an excellent rookie season
  • Without playing a single game, Griffin and Wall have been more productive than him
Maybe it was a weak draft year? Well yes, 2006 was a weak draft class (thanks again to Arturo), but there were still plenty of productive players available in that draft.

Nope, I think I'm going to have to pull a Bill Walton again: Andrea Bargnani was a terrrrrrible first overall pick. In fact, the expectation to be productive magnifies his lack of production, and so Arturo ranks him the tenth worst pick - at any draft position - of the last 30 years.

Let me let reiterate: the choice to select Bargnani as the first overall pick was worse than every other draft decision made in the last 30 years, with the exception of the drafting of the following players:
Not exactly elite company, is it? I think it's safe to say that - by any stretch of the imagination - Bargnani's been a bust of colossal proportions.

"Bargnani has a single valuable skill"

Basketball is about scoring more points than the other team, something I've written about before (and you should probably read that if you haven't already). In order to win, you have to be good at both scoring and preventing your opponent from scoring. Andrea is not good at preventing others from scoring; he doesn't rebound well, pass the ball well, or steal well, and is around average when it comes to getting blocks. He doesn't turn the ball over. His defense is below average; he's improved over the years and isn't as bad as he used to be, but no one is going to mistake him for Dwight Howard. The one advantage Bargnani has over other bigs is his shooting, as he is above average with regards to Adj.FG%, FT%, FG attempts, and points scored.

The problem is that Bargnani's "unique and valuable skill" of perimeter shooting...isn't all that valuable. If you want to score points, there are two areas on the court you should be shooting from: 1) the immediate basket area and 2) the three-point line. All those deep twos and midrange shots? Inefficient. That's because the farther you get from the basket, the harder is is to make your shots, and the lower your percentages get. The reason why three-point shots are useful is because of the extra point. If you are shooting from somewhere other than the immediate basket area, it should be from the three-point line.

How have Bargnani's shot attempts changed over the years?

As you can see, shot attempts increased everywhere on the court, except for 3pt shots. The shots that increased the most were <10ft and 16-23ft shots (also of note: his percentages for shots other than threes and layups were abnormally high. I expect that, while part of the increase may be due to true improvement, most of it was due to random luck). He takes deep twos at almost the same rate that he takes threes; if you are going to be taking long-range shots, why not take a few steps back and go for the extra point? So, while Bargnani is a good shooter, he'd be even better if he started taking more threes and driving to the basket more often, and cut back on the other types of shots as much as possible.

And I think that's the reason why deep-shooting big men are so rare: big men are the tallest players on the court, and as such, are able to park themselves right next to the basket and score a bunch of easy points. A big man who doesn't do this is wasting his key talent - being tall. The big guys on your team must be able to rebound and get easy points next to the basket and Andrea doesn't do either of those things. Until he learns how to do those things, he will forever be an inefficient, unproductive player.

 - Devin


  1. Not arguing with your evaluation of Bargnani but I've begun to wonder if the Raptors compounded the mistake of choosing him by trying to turn him into a centre. If you treat him as a SF and his stats stay the same (which, of course, they wouldn't if he were a 3), how does he look? Still costing his team games or just a below-average player?

  2. Nice post,

    Nice questions posed as well,

    How would Bargs numbers look if here were compared to PFs and SFs?

  3. yep really want to know where he compares to sf to (always thought he should have played there... still do)

    Two things when looking at Wins Produced, as is stated they "adjusted for other factors including pace, position and team." If I'm not mistaken FG% and Rebounds have a very big influence at the C... one which Bargs completely sucks at (rebounding) but his 'style' will always result in a lower number (both rebs and %).

    Would still take Bargs over Kwame Brown any day of the week.

  4. It doesn't surprise me that Berri's metric has Mahmoud Abdul-Rauf (nee Chris Jackson) as a bad player. Prime scorers on bad teams tend to be overguarded and shoot for a lower percentage. Jackson was not an overwhelming physical talent like LBJ or Jordan. He could not overwhelm and devastate his opponents. But no one in their right mind would suggest that he belongs on a list with Joe Wolf and Brad Sellers. Not even close.

  5. New post with Bargnani moved to SF is up.

    Anonymous: the draft scorecard - created by Arturo Galletti - ranks picks compared to the average production level offered by players taken at that pick. Abdul-Rauf was the 3rd pick in the 1990 draft, and remember, after the 1st pick, the 3rd pick is expected to be the most productive. His WP48, or per-minute production, was slightly larger than both Joe Wolf and Brad Sellers, but also way below what we should expect from a 3rd pick. So yes, you are partly correct: Abdul-Rauf was a better player than Wolf and Sellers.

    But a better comparison for Abdul-Rauf would be Adam Morrison - both were taken at the 3rd pick. It just so happens that Morrison was almost exactly twice as unproductive as Abdul-Rauf, and Arturo ranks him as the worst draft decision of the last 30 years.

  6. Thanks for the SF, and PF numbers Devin,

    I know Bargs is a defensive 5, I just thought it would be cool to see.

  7. One, Chris Jackson was a whole lot more productive Joe Wolf and Brad Sellers, and if you have a statistic that tells you otherwise, stop fucking using it, because it sucks like Paris Hilton after a fifth of Black Death vodka. Two, Galletti's list is fucked nine ways from nowhere because the one thing you can't do is claim that a player is a bust by comparing his production to other players selected in other years at a similar spot. For one, in 1984 you had some really great choices at #3. In 1990 there was Jackson and pretty much nothing else. mean, sure, there were roleplayers like Ty Hill & Toni Kukoc, but if the Nuggets had drafted either of them they'd be even higher on that bust list.

  8. ehmunro:

    So much to talk about, and so many ways I could respond. We'll go point by point, and I'll pretend that you weren't so vulgar.

    Why do you say Abdul-Rauf was more productive than Joe Wolf and Brad Sellers? As I said above, Wins Produced actually tells us that Abdul-Rauf was more productive than those two, but I want to hear your explanation.

    We'll address those other points after this one.

  9. The win shares metric is as bad, and generally useless, as any other magic number. Galletti's list is case in point. Win shares tells him that there were nearly 160 worse picks than Darko Milicic. And Darko is the biggest bust of the last decade. Yes, even bigger than Kwame Brown as Washington didn't pass up Carmelo Anthony, Dwyane Wade, and Chris Bosh to draft him. (There was one player as good as Bosh available in 2001.)

    As for Chris Jackson, he was the primary scorer on a poor team. Sellers & Wolf were bench fodder. And in Sellers case short term bench fodder. If they were capable of being NBA starters, they would have been. And you can't compare Jackson to third picks in good drafts and call him a bust because you have to show that there were much better options available. And in 1990 there weren't.

  10. 1)First of all, we don't use win shares. We used Wins Produced. There is a difference.

    2)You haven't told me why Abdul-Rauf was better than Sellers and Wolf. The old "if they were capable of being NBA starters, the would have been" argument doesn't work. We all know that coaches and GMs don't always realize who the best players are, and thus who should be starting.

    So again, tell me why Abdul-Rauf was better than Wolf and Sellers? What numbers can you use to back up your case? Give me some tangible, non-subjective reason.

  11. It is your argument that is subjective. You are subjectively judging, in the absence of any evidence except your metric of choice, that Abdul-Rauf was comparable to flotsam like Sellers and Wolf. By Dean Oliver's and John Hollinger's metrics, however, it's not even close. Abdul-Rauf accounted for 25.2 career Oliverian Win Shares compared with Wolf's 3.5. Wolf's Hollingerian career Player Efficiency Rating: 7.8. Abdul-Rauf's: 15.4

    Those are two metrics that say the opposite of what Berri's says. I am aware that Berri has criticisms of PER, but they're incorrect. Shooting percentage is not as important as Berri thinks. Abdul-Rauf is a perfect example. He was the leading scorer and first offensive option on a playoff team: He was not a stretch shooter who is often open because a teammate is drawing the defense away. We can expect prime scorers like Abdul-Rauf to shoot worse than specialists for this reason.

    I think that you believe in wp48 so blindly that you have allowed it to obscure your ability to see the obvious. I mean this isn't even close. You might as well argue that Troy Murphy is the best player in the NBA (oh wait, Berri did say that at one point).

  12. Anonymous #23475291 and 1/2:

    My argument is objectively based on WP, which is an objective stat. PER is highly flawed, as Berri has noted in the past. Win Shares is better, but not as good at explaining productivity as Wins Produced. That is an objective observation no matter how you want to slice it. Berri, Galletti, and Alvarez (among others) have written about this before, and the same tired criticisms have been around since the beginning.

    WP correlates very highly with win production in the NBA, much higher than PER or Win Shares. It is the best way to evaluate player productivity in the NBA.

    In the past I was skeptical of WP and treated it like any other metric. "It over-values rebounds, and treats OReb and DReb the same," I thought. But the truth is that rebounding is valuable, and missing shots and TOs are very costly. I've learned to trust WP over every other metric, and that is because it has a proven track record.

    Because of people like you, I will be creating another page on this site. From now on, if anyone raises one of the same old tired questions, I'll simply point to that page. I won't be wasting my time on these already settled questions anymore. If you really care to understand WP and player productivity, you will read everything I put up. Otherwise, feel free to live in ignorance.

  13. Wins Produced is still a fairly useless number, it has no understanding of individual roles on offense and defense (and this is a huge weakness of all the magic numbers, Hollinger at the least admits this by attempting to adjust the value of a missed field goal). Primary scorers draw extra defensive coverage, and tend to be less efficient, but their inefficiency creates greater efficiency for the other players on the floor. (Yes, LeBron is an exception given his high efficiency, but can we grant that he's a freak?)

    And, no, you don't get to cite Wolf & Sellers as two of the ten worst picks in NBA history out of one side of your mouth and then out of the other call them hidden gems who were relegated to waiving towels by stupid coaches out of the other. Wolf couldn't score efficiently as a fifth offensive option, wasn't a terribly effective rebounder, even in his limited minutes (RebRates for part time 4/5s tend to be slightly inflated). By contrast Jackson was the primary scorer for the Nuggets. Was he a great primary scorer? No. Overall he was probably an above average player.

    And none of this changes the fact that Denver drafted the third best player with the third pick in the 1990 draft. And Detroit drafted about the 20th best player in the 2003 draft with the #2 pick. And yet Jackson was one of the ten worst picks in NBA history and Darko can't even make the bottom 100? If that's what your "objective numbers" are telling you, your methodology is wrong.

  14. "It has no understanding of individual roles...."

    Really, you need to understand what WP is about. WP is about measuring how productive a player is, and how many wins they create for their teams. It's not about who would win one-on-one, who looks the best, etc.

    Ah, the old usage rate argument. It'll be up on Required Readings. Other high efficiency scorers? Nash, Durant, plenty of others.

    Way to make shit up. When did I say that Wolf and Sellers were hidden gems? I said that coaches don't always know which players are the best players on their teams - I never said that all those two were "hidden gems". Good work. Off the top of my head: Kevin Love, DeJuan Blair, and Balkman are all examples of good players stuck on the bench, and there are more.

    The list of players from the 1990 draft I'd rather have (for their first 4 seasons) over Abdul-Rauf: Cedric Ceballos, Toni Kukoc, Lionel Simmons, Derrick Coleman, Loy Vaught, Tyrone Hill, Dee Brown, Antonio Davis, Kenny Williams, Gary Payton, Derek Strong, Duane Causwell, Elden Campbell, Bimbo Coles, Felton Spencer, Jud Buechler, Kendall Gill, Anthony Bonner, Terry Mills, Tony Smith, Marcus Liberty, Steve Henson, Alaa Abdelnaby, Gerald Glass, Carl Herrera, Travis Mays, Rumeal Robinson, Willie Burton, and Dennis Scott. But I wouldn't want anyone after Gill. So the Nuggets selected (at best) the 30th most productive player at #3 in the 1990 draft.

    And still you haven't cited any numbers. I've asked you three times and you have failed to do so, which tells me that either you can't or you won't. If you'd like me to read any subsequent comments, I would suggest you start using some numbers.

  15. I'll take it that since you're now deleting responses that you're admitting that you're wrong and are simply going to keep deleting. I'll expect this one to be removed forthwith as well.

  16. Nope, just the insulting tired arguments.

  17. If WP is such a useless number then what would you use that's so much better? Your gut feeling doesn't count for much in the world of stats.