Adjusted +/- of your Chicago Bulls

Those who know me, know I'm a stat geek, so advanced statistics always interest me.   I saw this linked over at

Stephen Ilardi of recently posted on the APBRmetrics board with updated 6 year adjusted plus minus numbers (70%
weighted towards this past season and weighted greater for the most
recent years of the remaining 30%). I decided to look through the
current Bulls remaining on the team to get a sense of what we might be able to expect from this team next year.

Player, 2008-09 Minutes, Off. APM, Def. APM, Std. Error, Total APM

1. Brad Miller, 2100, 1.96, 3.00, .98,  4.96

2.  Joakim Noah, 1911, .89, 2.87, 1.1, 3.76

3. Kirk Hinrich, 1338, 1.12, 2.38, .96, 3.50

4. Luol Deng, 1627, -.48, 2.35, 1.03, 1.87

5. Ben Gordon, 2974, 3.02, -1.53, .99, +1.49(d'oh!)

5. Aaron Gray, 693, -1.74, 2.7, 1.56, .96

6. Derrick Rose, 2977, 2.03, -1.70, 1.13, .33

7.  Tyrus Thomas, 2175, -5.05, -.53, 1.02, -5.58

8.  John Salmons, 2959, -1.8, -5.35, .91, -7.14

On the roster but missing in the numbers along with reasons they didn't make it: Jannero Pargo (Europe), James Johnson (rookie), Taj Gibson (rookie), Lindsey Hunter (not enough minutes), Jerome James (corpse).

The longer I've seen use of the +/- stat the less validity I think it has.   First, you have to be dealing with the adjusted +/- and not the standard yahoo/nba/boxscore +/-.  Adjusted +/- can be found on, and it adjusts the +/- to a per minute basis relative to how the team performs with and without you on the court.

The point being that a bad team would give all players hugely negative scores and a good team would give all good players very positive scores regardless of the players impact.   The Cavs won the vast majority of their minutes on the floor, does that mean all of their players deserve a huge +/-?   Looking at the difference between when they are on and off the floor improves the statistic 10 fold.   These numbers do use adjusted +/-, so this is more of a note for those playing at home and doing their own studies.

It still falls short of being all that useful though, because you have things like the LeBron factor.  Using the Cavs as an example, if a player plays basically all of his minutes with LeBron and another guy plays half with and half without then guess who's +/- is better?   Caliber of competition makes a difference too.   A deep team's second unit may fair much better in the stat than it's first unit because teh players are similar in caliber but consistently get more minutes against weaker competition.

Anyway, disclaimer's aside, what would these numbers tell us if we believed them.   The most eyepopping number might be that John Salmons was pretty scary bad in +/-.  However, it's worth noting that his +/- as a Bull was a + .9 rather than a negative.  Our first clue these numbers may not mean much and mitigate the damage.

Next, you'd notice that Derrick Rose would have been 7th on the team last year, not what you'd expect given the media hype.   This one doesn't surprise me as I've often felt that over 36 minutes Kirk Hinrich was giving the Bulls as much quality as Derrick Rose was, especially if you factor Ben Gordon being on the court which removes some of the value of Roes's shot creation ability (since Gordon can be a primary shot creator).  Hinrich also had an excellent year while healthy and after being overrated two seasons ago appears to be underrated to me again.  The value of Rose's replacement probably hurt his numbers in this example.

Our big men, at least two of them, played pretty darn well.  Noah and Miller were our two best players in the statistic which is interesting given that we feel the team's biggest problems are in the frontcourt.   I think Noah benefited from playing on the bench for much of the year and not playing big minutes early in the season, especially as a starter, when things were going bad.  Plus he was a huge +/- player his rookie year. 

Unfortunately, I don't think this is a correlation proves casuality thing going on here.  The high marks for these two don't mean that they're our most important players or close to it.  Tyrus Thomas always did fairly well in +/- in the past, but his numbers fell off a cliff this season.   Again, this is likely related to the Bulls traditionally having a very deep team and their bench players getting better numbers than their starters as described above.

What is there to ultimately take from this?   Probably not a whole lot.   There are too many facters influencing +/-, and any stat that doesn't directly measure what a player does has to be a bit fishy to begin with.  


Leave a comment
  • No comments...I wonder why?

    Well, we know why. 80% of the world doesn't do match and 50% of the people who do match fib about their skills.

    I'm good at math but not Ph.D. good as the originator of this rating system. How good can Stephen Ilardi's rating system be when 14 people are better than Kobe Bryant (including Ron Artest and Yao Ming)?

    I use this rating system.

    Points + rebounds + assists + blocks + steals - turnovers = Production

    Production / minutes per game = Player Value

    When you do this with a spreadsheet on a team's individual stats, things become clear. If your Player Value > 1, you're a man.

    Lebron was 1.14 for last year. Dwight Howard, Chris Paul, Kobe Bryant were all just barely a 1.0. I didn't do other big names.

    The amazing Ben Gordon was .72 using this sytem last year. 3/10 of a point doesn't sound like much, but it's the difference between a Ben Gordon and a Kobe Bryant.

    My rating method allows "pretenders" to look good. For example, last year's "best Bull" was the amazing Anthony Roberson with a rating of .84 followed by Brad Miller at .79. So my rating system needed work. Some guys are efficient ballplayers for the "mop up" time that they play, perhaps that was Roberson.

    So before you use my rating system, pick a points per game minimum and eliminate all players from consideration if they didn't score points. How do you eliminate the boys from the men in the NBA? Is it 15 points, 18 points, 20 points?

    Let's make it 16 points since Derrick Rose averaged 16.8 points last season. How do the Bulls look now?

    Only Ben Gordon, Derrick Rose, and John Salmons made the cut as serious players. They averaged more than 16 ppg and their ratings were .72, .68, and .64 respectively. Derrick Rose in his rookie year had a rating of .68 with 16 ppg. A very good rookie season.

    Rose often gets compared to Chris Paul. Reasonable.

    Derrick Rose in his rookie year had a player value of .68 and 16.8 ppg while Chris Paul in his rookie year had a player value of .80 and 16.1 ppg.

    Last year, Chris Paul's player value was 1.01 with 22.8 ppg compared to Derrick Rose's .68 and 16.8 PPG. You might argue that Rose was a rookie and Chris Paul a young veteran.

    But that's the standard I set. Derrick Rose wasn't as good as Chris Paul in his rookie year. And Chris Paul was a man last year.

    How will the Chicago Bulls be this year based on previous ratings? It's very hard to say, since Ben Gordon is gone. The loss of Ben Gordon gives various players and the entire team a chance to become statistically better, or much worse. It should be a fun year.

  • In reply to rkraneis:

    Too bad blogs don't have spell check...

    I see I can't spell math, since I'm spelling it match twice in my reply.

    So I'm forced to reply to my reply. My spelling skills are actually better than a 3rd grader, I just can't prove it today.

  • In reply to rkraneis:

    Firefox, my friend, Firefox. It has built in spell check.

  • In reply to rkraneis:

    Speaking of spell check, there is a good program Spell Check Anywhere (SpellCheckAnywhere.Com) it adds spell check to all programs. It works everywere including blogs (it's great).

  • In reply to rkraneis:

    Doug, let's test your theory...

    Give me a "big man chucker" who gets lots of rebounds or a PG ball hog who gets lots of points and I'll run my "player value" on them. Don't forget to set your points minimum for the player to be considered at all.

    Dennis Rodman and Paul Silas had value of .69 and .74 but points per game under 10. Karl Malone, who some say would miss a shot just to get a rebound sometimes, was a man at player value lifetime of 1.01 with 25 ppg lifetime.

    In 2000 - 2001, Stephon Marbury had a player value of .86 but .74 lifetime. Ben Gordon's best player value was .79 with a lifetime player value of .73.

    Inefficient players are "punished" in my rating system since they can't accumulate all around points. If you contribute in many categories you are rewarded with points. Setting a points minimum helps to eliminate rating anomalies like Anthony Roberson on Bulls last year (best rating, .84).

    Great ball players score out at 1.0 or greater, it's an elite club.

    I don't know how you define "pace" but I suspect it's running teams. A running team will score more points but all players will suffer a bit more from turnovers. Give me a great running team and I'll run their numbers for last year.

    My rating system is weak (for now) because it doesn't de-value inconsistent players. Carlos Boozer's lifetime player value is .88 but he misses lots of games, just ask Utah Jazz fans. Perhaps I should just toss out players who miss 25% of their games. That would probably eliminate Boozer and Luol Deng over their player lifetimes so far.

    Players on the rise are identified after 3 years. Paul Millsap is a good example going from .74 to .8 with increasing points per game. Tyrus Thomas is another good example of a player losing value over time from .73 to .71 even though his points per game average is rising. Not a good sign for Tyrus.

    So it's fun Doug. My player value identifies the great ball players, tosses out the wannabees based on their low ppg, and identifies young players on the rise and fall like Paul Millsap and Tyrus Thomas.

    I hope I'm wrong about Tyrus, but I don't think so.

  • In reply to rkraneis:

    I could be totally wrong, but I thought this version of adjusted +/- attempted to control for the quality of teammates and opponents?

    I agree it's not a good stat to use for comparing between two different players straight-up, but I like APM for what it tries to do with quantifying the contributions of non-big-stat producers. It's far from perfect, but at least it's something that looks toward the non-stat contributions of guys on the team.

  • Well Doug, it's been fun.

    No rating system is perfect. Mine is simple enough for fans to use.

    Efficiency? Zach Randolph had a player value of .89 for the Clippers last season with a field goal percentage better than his team's average. Chris Kaman and Marcus Camby shot better than Randolph at over 50% but took a whole lot less shots, even though they played similar minutes.

    Randolph took a shot every 2 minutes of play, Kaman took a shot every 3 minutes of play, and Camby took a shot every 4 minutes.

    An interesting twist on efficiency. Why would slightly better shooters take significantly less shots in an offense, when they're close to the bucket?

    So, these rating systems are fun. Just wanted to spice up your blog today since there weren't any other takers on basketball statistics.

Leave a comment