You Are Here > > BBR Blog > NBA and College Basketball Analysis

SITE NEWS: We are moving all of our site and company news into a single blog for We'll tag all Basketball-Reference content, so you can quickly and easily find the content you want.

Also, our existing Basketball-Reference blog rss feed will be redirected to the new site's feed. // Sports Reference

For more from Neil, check out his new work at

Gilbert Arenas

Posted by Neil Paine on December 20, 2010

So, I was thinking now would be a good time to talk about Gilbert Arenas.

First of all, I'm an unabashed Gilbert fan; I've always found him to be one of the NBA's most interesting people, in addition to one of its most gifted players. And after everything that's happened over the past few years, I'm glad he finally has an opportunity to make a fresh start in Orlando.

That said, I'm not sure he can help the Magic very much at this stage of his career.

Before we talk about the present, though, let's go back in time a bit and take a look at Gil's game in better days. At his best, Gilbert was capable of making a major positive impact on his team. From 2005-09, the Wizards were 127-116 (.523) in the games Arenas started and 63-104 (.377) in the games he didn't. In 2007, Gil was my dark-horse MVP candidate because Washington was a staggering 14.1 pts/100 poss. of efficiency differential better when he was in the game vs. when he was on the bench. And the Hibachi's adjusted +/- scores from 2006 + 07 ranked him among the top difference-makers in basketball.

Looking at the box score data, it's not hard to see why Arenas' presence on the court made Washington so much better. In 2006 & 07, he was the most efficient high-usage player in the game -- better than Kobe Bryant, LeBron James, or Dwyane Wade. He was a stellar initiator who could create for himself (57 TS% while taking 30% of the Wizards' shots when in the game) or others (he assisted on 27.1% of his teammates' buckets as well). His .471 FTA/FGA was absurdly high for a guard (meaning he was getting to the rim with impunity), he was deadly on pick-&-rolls, and he was an expert at making plays in buzzer-beating situations. Even his much-maligned defense deserved a second look in 2007, with a -2.1 on/off-court split. Simply put, Gilbert Arenas was one of the NBA's elite players in 2006 and 2007.

Unfortunately, a torn MCL at the end of the '07 campaign started Arenas on a downward spiral that culminated in his exit from Washington this weekend. He played only 15 games in 2008 and 2009 combined, then was markedly less effective in his 2010 comeback before the infamous gun incident shut his season down. And ostensibly playing second fiddle to John Wall in 2011, Arenas has shown little of the dynamic Gilbert we saw in 2006 & '07.

Which brings us to today. With the Heat & Celtics rolling and the Magic struggling to keep pace, Orlando clearly needed to upgrade their offense, the 15th-ranked weak link that all too often betrayed their 4th-ranked defense. To that end, Orlando dealt a slumping Rashard Lewis for Arenas, plus shipped Vince Carter, Marcin Gortat, & Mickael Pietrus to Phoenix for Hedo Turkoglu, Jason Richardson, & Earl Clark. Obviously, this dramatically changes the way Orlando will operate on offense, and it also creates a backcourt logjam with Jameer Nelson, Jason Richardson, & Arenas all vying for playing time (and FWIW, Stan Van Gundy says Nelson's job as starter is guaranteed).

If Arenas was still the Gilbert of old, this would have been just the kickstart Orlando's offense needed, but unfortunately he has given few indications that he's still capable of that kind of impact. Arenas needs the ball in his hands to facilitate himself and others, and was actually leading the Wizards in possession usage (26.2%) despite professing to be Wall's sidekick. When he's in the game for the Magic, he will almost certainly be their #1 usage man on the perimeter. Trouble is, Arenas has produced just 1.02 pts/possession since 2008, and is sitting at 0.98 so far this season (compare to Carter's 1.11 and the league average of 1.07). Unless Gilbert can somehow channel his 2006/07 self again in a new environment, Orlando will be allocating a quarter of their possessions to a player who's instantly their least-efficient option, plus damaging their defense at the other end.

Now, maybe Gilbert can still create opportunities for others by shaking up Orlando's increasingly stagnant offense. According to SPM, he's still worth +1.13 pts/100 poss. above average offensively (and was worth +4.06 last season), despite the poor efficiency stats. But the most troubling fact of all is that Arenas' overall on-court impact, once profoundly positive, is now subpar. His adjusted +/- for 2010 & 2011 combined is -1.46, a steep drop for a player who had never been below-average from 2003-2007.

Based on the numbers, one has to conclude (sadly) that the Gilbert Arenas Orlando just acquired is not the same version we saw before his injuries and personal travails. He's settling for too many jumpers, no longer drawing fouls, no longer avoiding turnovers, no longer scoring efficiently, and consequently he's not having the same positive impact on his team. As one of my favorite players, I'm rooting for Agent Zero to buck these trends and rediscover his game in Florida -- but as it stands now, I'm not sure he can make the kind of difference Orlando is counting on.


64 Responses to “Gilbert Arenas”

  1. Neil Paine Says:

    I will say that no matter what happens, he's automatically going to be better than Chris Duhon. So that in and of itself is an upgrade, if they relegate Arenas to backup PG duties.

  2. Jason J Says:

    Except that Duhon has no problem being a good soldier coming off the job and doing what he's told. Can the same be said for Arenas? We'll see. He was pouty in Washington starting next to Wall. How's he going to feel coming off the pine for Nelson and / or JRich both of whom he probably considers himself to be superior to? And why do I still call it "the pine" when it's actually a row of folding chairs?

  3. Neil Paine Says:

    Right, the Magic now have 3 starting-caliber guards, and the most naturally talented one (aka the one least inclined to accept a reduced role) is going to be coming off the bench.

  4. P Middy Says:

    Mm, I'd still start Nelson and Arenas in the backcourt and bring Hedu and Richardson off the bench. Or heck, put Hedo out there at SF and start Ryan Anderson at PF with Gilbo and Nelson. Whatever is up with Gilbert, he is still 10,000 times better Rashard "i'm a hipster now, so no more of that try-hard stuff" Lewis. That guy was garbage.

  5. JTaylor21 Says:

    How wonder how three players that need the ball in their hand to be succesful are going to perform playing around a DHow that demands and needs the ball in his hands. This is not the same DHow that Turk played with the last time in Orl, so it will be interesting to see how him and the other new players respond and perform around him. I do know that Orl will miss the defense Petrius and Gorat brought.

  6. taheati Says:

    If the Magic are done dealing, I see more mins for Brandon Bass than Ryan Anderson despite Anderson's length advantage.

    Bass has always played bigger than his size, is a competent rebounder and more aggressive on the offensive glass (than Anderson), defends better than his DRtg, is less foul prone (than Anderson), has superior on/off court numbers (including defense)than Anderson *or* Lewis, finds open teammates *and* will pass the ball.

    Gilbert did pretty well with Brendan Haywood/Etan Thomas "defending" the paint. Bass is at least the equal of Thomas, so the Magic should be covered (w/Howard/Bass) whenever Arenas (or JRich) matadors on d.

    But it's still an "if" whether Orlando is actually "done dealing." JJ Redick suddenly becomes a trade chip which the Magic could conceivably package for another big.

  7. AHL Says:


    a DHow that demands and needs the ball in his hands

    ...what? Demand...maybe situationally when he's only getting 10 shots or 23% usg or something. Needs... uh, I think he does kinda okay without the ball.

  8. P Middy Says:

    He really does seem to put better moves on his guys off the ball for o-rebounds and alley oops than when has the ball, doesn't he? At least he did last year. Haven't seen the Magic play except for that blowout loss to the Heat.

  9. Greyberger Says:

    The real danger is this cat gets a starting role or 30+ minutes a game and starts taking possessions from Dwight Howard. Here's the new math on Howard:

    60% TS, 17% TOV, and 23% usage for career
    59% TS, 16% TOV, and 29% usage this year

    He's taking a lot more possessions than in his early years without much drop in his efficiency. He's a legit MVP candidate for them... but gets visibly frustrated when his teammates treat him like they're Kobe, he's Gasol, and he needs to wait his turn.

  10. Jared Says:

    I loved Arenas and think he's one of the biggest tragedies of the last decade. He's so similar to Sprewell to me in that he had all the talent, but it never was put all together for a sustained period of time.

  11. Neil Paine Says:

    I would say Arenas at his apex (2005-07) was pretty significantly better than Sprewell ever was. Sprewell was better defensively, but even there the numerical evidence suggests his defensive ability may have been overstated.

    Sprewell had a 15.1 career PER, and I think that pretty much reflects his actual value, even taking defense into account. He wasn't very efficient and was an above-average -- but not huge -- possession user on a series of teams that ranged from mediocre to barely-average (2004 T-Wolves excepted).

    John Hollinger used to joke that, after joining the Knicks, Spree and Allan Houston used to play tug-of-war with the title of "NBA's Most Overrated Player". There's something about playing in NY and tasting a modicum of success that automatically makes the media overinflate your worth (check out this year's Knicks, owners of a -1.03 SRS despite the effusive praise heaped on them during the 8-game W streak).

  12. Joe Schaller Says:

    Sprewell had one of the great scowls of all time though, even better than Chris Webber.

    The Magic may improve by subtraction of Lewis allowing both Bass and Ryan Anderson more minutes. Bass has been huge and the sole bright spot for the Magic. Jason Richardson is the big acquisition of this deal, spreads the floor better than Carter, is younger and more athletic also. Arenas could be an excellent sixth man if he accepts the role and should be a huge improvement over Duhon.

  13. DSMok1 Says:

    Advanced SPM still likes Arenas quite a bit. He's projected to be +2.4 still, quite good (ASPM had him as a top 5-10 player at his apex).

    The numbers for the 6 players traded are here:

  14. Elmo Says:

    It's funny that when the Magic signed Gortat to his current deal they were widely criticized for overpaying and creating a redundancy at the five. Now that Gortat is gone, people are worried about the depth there? Between Anderson, Bass, and the best defensive big man in the league, I think Van Gundy is OK with his bigs.

    As mentioned above, J-Rich is far from ball dominant. See 2007 Barron Davis led Warriors, or the last few years of the Nash led Suns, for reference. From my limited viewing, seems more comfortable off the ball than on it.

    All in all, I think Smith showed some big ol' balls with this move. The Magic might not even leave the east, but I absolutely believe that they are a better team now than they were before the trade.

  15. David Friedman Says:

    The 2005-2009 time frame is a very interesting choice. The 2009 Wizards were not only without Arenas' services but Caron Butler missed 15 games, Brendan Haywood essentially missed the entire season and Antonio Daniels--who filled in very capably for Arenas in previous seasons--was on his last legs before being traded to the Hornets.

    The period when Arenas was allegedly an elite player had already ended by 2009, so let's just look at 2004-05 through 2007-08; during those four seasons Arenas made his only three All-Star appearances and earned his only All-NBA selections (Third Team in 2005 and 2006, Second Team in 2007). The Wizards won 45, 42, 41 and 43 games during those seasons. Keep in mind that in each of those seasons Arenas had at least one other All-Star or All-Star caliber player by his side. Here are the Wizards' records with and without Arenas in each of those seasons:

    2005: 44-36 with, 1-1 without
    2006: 40-40 with, 2-0 without
    2007: 39-35 with, 2-6 without
    2008: 6-7 with, 37-32 without

    The Wizards were 129-118 with Arenas during his prime (.522) and 42-39 without Arenas during his prime (.519).

    It would also be very interesting if you shared with your readers the Wizards' record in games that both Arenas and Butler missed as opposed to the games that only Arenas missed. I am surprised that someone who is trying to look at things analytically would not want to factor in the impact of another All-Star being out of the lineup at the same time that Arenas was out.

    The reality is that even when Arenas was at his best he was on the fringe of being elite (I consider elite to be top five to top 10, but some people throw that term around so loosely it seems like there are supposedly 20 or 30 "elite" players at any given time) and his team only performed slightly better with him than it did without him. The Wizards were slightly above .500 during Arenas' prime when he played and they were slightly above .500 during Arenas' prime when he did not play.

  16. Jason J Says:

    DMosk is 2.4 SPM better or worse than Jameer?

  17. Neil Paine Says:

    Arenas still had the benefit of the doubt through 2009. It wasn't until 2010 that he definitively proved he was no longer his old self.

    Re: DSMok, did you project regularized APM?

  18. DSMok1 Says:

    Drat. I read off the wrong number. Arenas was projected at +2.4 BEFORE this season, but has played so bad this year (-0.7) that his projection is now down to +1.1.

    Jameer, on the other hand, started the season projected at 1.7, but has been so GOOD this year (+4.0) that his updated projection is now +2.4.

    I'm sorry I quoted the wrong number, there!

    I updated the sheet on the trades with some explanations:

  19. DSMok1 Says:


    No, that would be unnecessary. Regularized APM is a method of enforcing regression to the mean within the APM framework (which is really useful for 1 and 2 year APM's).

    However, when constructing my ASPM model, I regressed the statistics onto Ilardi's 6-year average APM's (that was the primary source, at least). The 6-Year APM gave a reasonable, stable APM--but didn't tell you anything but how good a player was on average. But it also is the perfect source to regress an SPM model onto.

    I then have constructed methods to regress my ASPM based on the error, and based on ASPM's out of sample performance. I basically determined how to run a projection for the next year (how much regression to use) and then subtracted back out the aging.

    What I end up using for regression is 2 different things: first I regress to a player's expected talent levels based on the player's MPG and quality of team. That's the best estimate from a single year. The other way incorporates regressing based on MPG and quality of team, but it is used for a Bayesian updating of the pre-season projection for the player.

  20. Neil Paine Says:

    OK, cool. I was just wondering because I saw a column for regularized APM, and I didn't know if you had run a separate regression against Joe Sill's data.

  21. DSMok1 Says:

    I think you saw an abbreviation of Regressed APM, not Regularized... sorry for the confusion!

  22. DSMok1 Says:

    Here's a single-game analysis of Orlando's first game:

  23. DSMok1 Says:

    Sorry, wrong link: just click open, here:

  24. Basketball Goal Says:

    I think Gilberts blind confidence in his basketball abilities will help the magic. A team that lives by outside shooters and the rebounding ability of Dwight.

  25. David Friedman Says:


    You do not find it statistically significant that throughout the period when Arenas was an All-Star/All-NBA player (1) his team was barely above .500 when he played and (2) his team essentially posted the same winning percentage whether or not he played? How many "elite" players barely led their teams to .500 records during their primes over a period of four years? How many teams performed essentially the same without an "elite" player even when that "elite" player missed a substantial number of games?

    It is instructive to compare Arenas' impact on the Wizards to Tracy McGrady's impact on the Rockets during a similar time period, a subject that I discussed in March 2008 when the Rockets posted the second best regular season winning streak in NBA history:

    "There is a stark and dramatic contrast between the Rockets’ record when McGrady plays (162-83, a .661 winning percentage) versus their record when he is not in the lineup (19-46, a .292 winning percentage). Prorated over 82 games, the Rockets have essentially performed like a 54 win team with McGrady and a 24 win team without him. This year, the numbers read 36-13 (.735) with McGrady and 8-7 (.533) without him, which prorates to 60 wins and 44 wins respectively."

    From 2005-08, McGrady made the All-Star team three times and made the All-NBA team three times (Third Team twice, Second Team once) but he had a much greater impact on his team's won-loss record than Arenas did during that same time frame. Overall, McGrady made the All-NBA First Team twice and the All-NBA Second Team three times and he was clearly an elite player (when healthy) for an extended period of time, whether one looks at his skill set, his individual numbers or the striking impact that he had on his team's won-loss record--when he played the Rockets were a contending team (projecting to 54 regular season wins) but when he did not play the Rockets performed like a lottery team.

    The burden of proof is squarely in the corner of anyone who suggests that Arenas ever was an "elite" player, because the evidence strongly suggests that this is not true--unless you define "elite" to be top 20 or top 30; I am defining "elite" to mean someone is one of the top five to 10 players in the NBA, which usually corresponds to making the All-NBA First or Second Teams. Granted, sometimes there can be mistakes in the voting and a player could theoretically be "elite" without making those teams if there is a glut of talent at his position (voting is done by position, so in theory a non-elite player could make the All-NBA team at, say, center, while an elite level guard fails to make the team).

    Since you elected not to share the Butler splits with your readers, permit me to quote from some research that I did two years ago:

    "In 2007-08, the Wizards essentially replaced Arenas with career journeyman Antonio Daniels--a solid pro who has played with five teams in his 11 year NBA career--and not only did not miss a beat, they actually performed better. It is important to remember that Butler missed 24 games last season; the Wizards went 33-25 (.569) with Butler and 10-14 (.417) without him--and five of the losses with Butler also came with Arenas in the starting lineup. Washington's best starting lineup last season (by winning percentage, with a minimum of 10 games) was Butler, Daniels, Antawn Jamison, Brendan Haywood and DeShawn Stevenson. That group went 23-16 (.590) for nearly half a season without Arenas, which projects to a 48-34 record, a mark that would exceed the Wizards' best season since acquiring Arenas."

  26. Neil Paine Says:

    Since you enjoy "with you vs. without you" stats so much, the Wizards were 8.2 pts of efficiency differential better when Arenas was on the floor vs. off from 2005-2009, peaking at +14.1 in 2007 (a year when Butler was only +6.1). It's tough to say the Wizards weren't significantly better when Arenas was on the floor.

  27. David Friedman Says:


    Actually, I believe that you made "with or without you" stats a key part of your argument that Arenas was an elite player--in fact, those were the very first stats that you cited and I am assuming that you put forth what you considered to be your strongest evidence at the beginning of your article; I simply provided some additional data in order to put Arenas' impact in proper context. I could add that without Haywood in 2009 the Wizards slipped from being a decent rebounding team (15th in rebounding differential in 2008) to being a very bad rebounding team (24th in rebounding differential in 2009). Lack of interior presence at both ends of the court had a lot to do with Washington's poor record in 2009. Also, in previous seasons the Wizards adequately replaced Arenas with Antonio Daniels but Daniels was no longer the same player by 2009.

    It is interesting that you assume that Arenas' absence in 2009 is the main reason that the Wizards were terrible but you neglected to mention that prior to that season the Wizards were an average team whether or not Arenas played. It certainly seems more statistically significant to cite what happened over a four year period than to just lump 2009's games into the mix and act like Arenas had such a huge impact on winning--and in doing so you make a huge assumption that the 2009 version of Arenas was as good as the earlier version even though all of the evidence (Arenas' subsequent performance, the way that knee injuries have hindered the careers of many other guards) suggests otherwise.

    No matter how you try to frame the discussion, the bottom line is that during the four seasons of Arenas' prime the Wizards were essentially a .500 team when he played and a .500 team when he didn't play. That suggests (1) that he did not have a major impact on whether they won or lost and (2) that it is therefore suspect to call him an "elite" player. It would be very interesting if you could cite any examples of elite players who had such little impact on winning over such a large sample size of games. Specifically, is there an elite player whose team was roughly a .500 team with or without him for a four year period during which he missed a statistically significant amount of games? Generally, when a team adds an elite player they gain at least 15-20 wins and when they lose an elite player they lose 15-20 wins (I am thinking of Larry Bird with the Celtics and David Robinson with the Spurs but there are other examples, too); of course, there are exceptions and there are also cases in which a team lost an elite player but added enough depth elsewhere to somewhat soften the blow but I find it remarkable that someone would try to make the case that Arenas was an elite player during a four year period in which he led a team with multiple All-Stars to a record a little above .500, particularly when that same franchise posted a nearly identical winning percentage over a substantial number of games that Arenas missed.

    I already demonstrated that during the same period of time that Arenas had no discernible impact on winning McGrady's presence or absence made a significant difference in Houston.

    Older fans might cite the fact that Pete Maravich's Jazz teams never made the playoffs but they were an expansion team that was largely bereft of talent and they had a much better record when Maravich played than when he didn't play.

  28. Neil Paine Says:

    You know what they say happens when you assume...

    There's a great deal of evidence for Arenas if you throw the 2005-09 W-L record as a starter out. In fact, that's probably the weakest part of the argument, because to get any kind of sample you have to extend it to 2008/09 -- he only missed 12 games during his very best years (05-07).

    The strongest part of the case: His on/off-court impact was huge, especially when you adjust for the strength of teammates and the opposition. When he was on the court, he made them significantly better. And he was definitely elite (top 10-15) by the advanced box-score based metrics.

  29. Greyberger Says:

    Re: #27 I don't understand how you can write many paragraphs, ostensibly on Gil Arenas's production between 2004 and 2009, and not once mention Gil Arenas's production. Win-Loss record with/without him seems like a very roundabout way to measure a player, star or otherwise. And if you do think it's a pertinent part of the puzzle, he still stands out looking at box-score-derived metrics and on-court/off-court numbers.

  30. David Friedman Says:


    With all due respect, you are making a lot more assumptions here than I am.

    I already stated why I disagree with your premise in terms of lumping 2009 in with the rest of Arenas' career but it is clear that you have little to no interest in even acknowledging--let alone substantively addressing--the various questions that I raised. Rather than getting involved in an endless back and forth that will lead to nowhere, I will "fight on your turf" so to speak. Let's assume that you are right that Arenas was an elite player from 2005-07. How do you explain (1) that despite having a solid supporting cast (including other All-Stars) the Wizards were never much better than .500 during his prime and (2) that when Arenas essentially missed an entire season (2007-08) right after having arguably the best season of his career the Wizards went 37-32 without him?

    My point is two-fold:

    1) For someone who was supposedly elite Arenas had very little apparent impact in the won-loss column when he played.

    2) The Wizards had little trouble replacing the supposedly elite Arenas in 2007-08 even though the rest of their core rotation was largely the same as it had been the previous season: the top players in minutes played other than Arenas in both 2007 and 2008 were Butler, Jamison, Haywood, Stevenson and Daniels. The only difference in the rotation (other than Arenas missing most of the 2008 season) was that Roger Mason essentially took over Jarvis Hayes' minutes.

    Do you really think that if the Lakers lost Kobe, the Heat lost LeBron, the Magic lost Howard or the Mavs lost Dirk and those teams made no other change to their rotation other than the equivalent of swapping Mason for Hayes their winning percentages would remain exactly the same? Do you think that if the 2007 Wizards had Kobe, LeBron, Howard or Dirk that they would have been a .500 team?

    Efficiency differential and other "advanced stats" may be very useful in certain situations but that particular stat sure does not seem to explain this particular situation: your number is saying that Arenas made his team much more efficient but the bottom line is that when he was deleted from the lineup for an entire season the team experienced little if any decline in performance.

    Put it this way: if I had said to you before the 2008 season that the Wizards would be just as good without Arenas as they had been with him because I consider Arenas to be somewhat overrated, you would have retorted that I am not being objective and that you have objective, scientific numbers that "prove" how valuable Arenas is. Those are two completely conflicting hypotheses and the only way to find out which hypothesis is most accurate would be to remove Arenas from the lineup while making no other changes (or at least as few changes as realistically possible) and then see what happens. Basically, that is exactly what took place: Arenas missed almost the entire season, the Wizards did not make any other major changes to their rotation and they kept right on rolling along at slightly above .500 without Arenas' great "efficiency differential." It would seem logical for to therefore question the value of a statistic that does not seem to accurately predict/describe reality but you still insist that "efficiency differential" proves that Arenas was an elite player during the time in question.

    I recently read an article about Galileo which stated that during his lifetime some of his critics contended that telescopes did a good job of describing "things below" (i.e., on Earth) but they were useless to describe "things above" (i.e., in outer space). That seems to resemble your contention about "efficiency differential": in cases where "efficiency differential" says that a player is elite and the won-loss results support this contention we have proof of the value of "efficiency differential" but in cases (like Arenas) in which "efficiency differential" is not supported by the facts we are supposed to believe in "efficiency differential." Galileo's critics said that the Earth does not orbit the Sun but Galileo replied "It still moves." How is this different from saying that Arenas' efficiency led to wins even though this team won at the same rate without him?

  31. David Friedman Says:


    I was addressing the very first argument Neil made regarding Arenas' alleged elite status, namely that the Wizards were supposedly much better with him than they were without him. In my opinion, it is misleading to lump 2009 in with the other seasons because other factors besides Arenas' absence contributed to the Wizards being lousy in 2009 and because it is far from clear that Arenas was still an elite player in 2009. Arenas missed virtually all of the 2008 season without affecting the Wizards' winning percentage very much so I think that it is reasonable to suggest that if Arenas had played in 2009 the Wizards would have been about as lousy as they were without him. There is certainly no evidence that Arenas contributed very many wins so far this season, either.

    Since you asked my opinion of Arenas' productivity, I would say that during the period in question (2005-08) he was an All-Star level player--i.e., one of the top 24 or so players in the NBA. You could make a case that he was briefly on the fringe of being elite (i.e., top 10) but he certainly did not have the staying power at that level that players like Kobe, LeBron, Howard, Nowitzki, etc. have displayed. At that time, Arenas was a combo guard who could both shoot from long range and drive to the hoop but he had poor shot selection, did not play much defense and he largely disregarded his coach's game plans. Bad shot selection is important regardless of field goal percentages/true shooting percentages because when a player takes a lot of bad shots his team is not in position to offensive rebound and the other team may get easy fastbreak opportunities due to bad floor balance. When Arenas got hot he could carry the team for a quarter or even for a game but when he was cold he could also doom his team. He was productive but with very high variance--kind of like J.R. Smith is now, though of course Arenas was more skilled and more productive than Smith. I wrote at that time that I did not think that any team with Arenas as its best player would ever advance past the second round of the playoffs.

    I think that Arenas got a lot of publicity because he is so gregarious and colorful but he is not more productive or valuable than a guy like Joe Johnson, another All-Star caliber player who is not elite and who likely is not good enough to carry a team past the second round without a lot of help.

  32. Greyberger Says:

    Re: #31 specifically, since you were kind enough to reply fully where I was perhaps a bit glib.

    As for the natural experiment you describe, it is quite damning. Agent Zero's defenders do have to explain why the Wizards underperformed expectations with him leading the team and didn't disintegrate when he was injured. I think of Vince Carter in Toronto, and the Knicks with Patrick Ewing, which everybody's probably heard about enough.

    I'm not sure trying to tease out a ton of conclusions from the record with him and without him (Really, we owe U2 royalties by this point) is the best way of measuring how bright his flame burned. It's a great perspective for writing the narrative, but team wins is not a good way to compare or evaluate players. Staying power and durability is important, but the point is how good he was in his 163 games between '05 and '09 and the shame it is there won't be more games like it.

    And trying to walk him down from the top 24 is selling him short! In the 'most efficient' link, Arenas compares (on offense) to Kobe, Lebron and Wade. That'd be somewhere in the top four. By Win Shares (which neither gives him much credit for individual defensive stats nor the team defense component) it's a club of nine to eleven. I can see that sliding overall thanks to weak defense, but that's where the conversation should start.

    The team was average, but the offense was good, and the numbers suggest Arenas was the main reason. Arenas in '06 and '07 compares favorably to Joe Johnson, any year - more efficient, more possessions, more assists, more steals, same rebounds (!). If the Arenas defense lawyers have to explain the team continuing to be average when he was injured, his prosecution has to explain his superstar rating in advanced stats and on/off splits when healthy.

  33. David Friedman Says:


    My main objection is to any Arenas narrative that suggests that he was ever a credible MVP candidate and/or that he was ever much better than (briefly) the 10th (or so) best player in the NBA (by definition, a credible MVP candidate should be a top five player). I am not attempting to "walk him down from the top 24"--he was in the top 24 during the period in question, he just was not in the top 10 (except for a very brief period) and he was never in the top five: for instance, in 2006-07 (the only year that Arenas made the All-NBA Second Team), the All-NBA First Team was Duncan, Nowitzki, Amare, Kobe and Nash, while the other Second Team members were Bosh, LeBron, Yao and T-Mac. The Third Team that year was Melo, KG, Dwight Howard, Billups and Wade. I can perhaps go along with putting Arenas 10th among those 15 players for that season but I really don't believe that he was a top 10 player at any other time.

    Another way to frame this is to say that an elite player--by which I mean a player who is in the top five or top 10--is someone who is good enough to be the best player on a championship team. Most NBA championship teams have been led by a player who just about anyone would consider to be "elite." I don't think that Arenas was ever good enough to be the best player on a championship team.

    I agree with you that it is not proper to judge a player's value solely by wins and I would extend that point to say that it is not proper to judge a player's value by any one stat or factor--and I am glad that you agree with me that Arenas' advocates should have to explain why Arenas' teams were so mediocre during his prime and why the Wizards fared just fine without Arenas even when he was at his peak. I also agree with you that it is a shame that Arenas did not stay healthy (the same applies to T-Mac and, more recently, to Yao as well).

  34. Neil Paine Says:

    OK, I'll acknowledge that 2008 represents an interesting situation. The Wizards were essentially just as good (40-42 pythagorean record) both years, despite losing Arenas in '08. But like Grayberger mentioned, how they got to 40-42 pyth was different -- the Wizards were better defensively in '08 but worse on offense by almost a full point per possession. And that's with Butler and Jamison having the best seasons of their careers (plus Daniels, who was actually a highly efficient & underrated player, getting a bump in PT, and Roger Mason being far more efficient than Jarvis Hayes ever was).

    So perhaps Arenas' on/off impact was overstated by the +/- data, just like it seems understated by the apparent lack of change in Washington's W-L. It's certainly plausible that Arenas was "elite" offensively but his defense -- as I said, much-maligned throughout his career -- was easy to replace. Then again, just as including 2009 may not say much about 2005-07 Gil, using the 2008 Wizards is a roundabout way to determine Gil's value in his prime, especially when the ultimate "with or without you" (thanks U2!) stats from those seasons say Washington totally fell apart when he wasn't in the game.

    At any rate, why did we just devote so many words to a semantic argument about the word "elite"?

  35. Joe Schaller Says:

    I would draw comparisons of Arenas to Iverson. Streaky with mediocre shooting percentages, way too many turnovers, lazy defense and a primary value of an ability to get to the free throw line and convert at a high percentage. Elite? Not in my book.

  36. David Friedman Says:


    I think that it is important to define terminology, because otherwise words become meaningless. You asserted that Gilbert Arenas was "one of the NBA's elite players in 2006 and 2007" and the first piece of evidence that you cited was a comparison of the Wizards' won-loss records with and without Arenas from 2005-09. Whether or not your assertion is correct depends first on how one defines "elite" and then also on if the evidence supports including Arenas within the "elite" category. As I explained, I limit "elite" to roughly five to 10 players in a given season (i.e., All-NBA First Team and Second Team level, though we may disagree with some of the "official" selections in that regard) and I think that, at most, one could say that Arenas reached the fringe of that territory for one season. This is not just a matter of semantics but rather an important part of communicating clearly: if I say that someone is an "elite" player it is quite clear what I mean (that is not to say that you must agree with me or that I am always right but simply that I have established exactly what I mean when I use that term). Part of the reason that I commented on your article was to receive some clarification regarding what you mean by "elite"; another reason that I commented is that citing all of the Wizards' games from 2005-09 provides a distorted picture of Arenas' impact: the 2009 Wizards were probably going to be bad with or without Arenas and the 2008 season provided strong evidence that, regardless of what some stats may seem to suggest about Arenas' productivity, it was not that difficult to replace him. I simply cannot imagine that a truly "elite" player could be so easily replaced.

    There are a lot of factors involved in winning and losing that are difficult to precisely measure. Almost as a throwaway line you mentioned that Butler and Jamison had career years when Arenas missed almost all of the 2008 season--but isn't that very significant? Great players supposedly "make their teammates better" (I prefer to say that "Great players create openings and opportunities for their lesser talented teammates to do what they do well"), so why did Butler and Jamison perform so much better without Arenas? That is something worth thinking about/researching. Arenas' impact is not accurately and completely described simply by looking at his "efficiency differential" or any one statistic/fact/observation; his impact can only be assessed by looking at a variety of factors. I think that when considering the big picture--his skill set, his teams' mediocre performance during his prime even when he played, his poor defense, his reluctance to follow his coach's game plans--Arenas during his prime was an All-Star level player but he fell short of truly reaching elite status. Arenas' lack of attention to detail defensively is not just important regarding his defensive assignment in a given game but also because of the impact that Arenas' attitude about defense had on the team overall: the team's best player sets a tone for the rest of the squad and if the best player does not focus on defense then the other players are going to feel like they don't have to focus on defense either. That may not be something that can be quantified in "efficiency differential" but it is nevertheless a factor that must be considered when deciding whether or not a player is "elite." I would have no objection if someone just said, "Arenas was one of the 20 best players in the NBA during his prime and he was really fun and exciting to watch even though his team never won anything and even though his team hardly missed a beat when he was hurt for virtually an entire season"--but it is another thing entirely to say that the best player on a mediocre team was as good/almost as good as guys like Kobe, LeBron and DWade.

  37. David Friedman Says:


    Isn't it inconsistent for you to say on the one hand that the games from the 2009 season that Arenas missed matter because we did not yet know if he was still in his prime but on the other hand "he only missed 12 games during his very best years (05-07)."? It is more likely that 2007-08 would have/could have been a prime year for Arenas (coming off of his selection to the All-NBA Second Team) then it is that 2009 would have/could have been a prime year: by the 2009 season it was pretty apparent that Arenas would never be the same player but coming into the 2007-08 season there was every reason to think that Arenas could perform at the same level that he did in 2007. This is significant because the 2008 Wizards lost an All-NBA Second Team player and yet their record did not get worse, while the 2009 Wizards underwent a lot of turmoil before losing a player who had already missed almost an entire season. In other words, by 2009 Arenas was already out of his prime and many other factors had an obvious effect on the Wizards' record but that was not the case in 2008.

    Also, I would be interested to know what you would have said back in November 2007 if someone would have predicted that even if Arenas missed virtually the entire season that the Wizards would have basically the same winning percentage despite not making any other significant rotation changes.

  38. Neil Paine Says:

    My final comment: You seem to be very fixated on the Wizards being just as good record-wise without Arenas in 2008 vs. with him in 2007. That's certainly one data point to consider, but I've detailed many others where Arenas does fit even your strict definition of "elite". If we are deciding to give extra weight to team performance the year after a player departed, though, recall that a certain Mr. Michael Jordan left the Bulls before the 1994 season and Chicago won a grand total of... two fewer games.

    (I realize this isn't a totally apt comparison -- for reasons I've written about in the past -- but I want to illustrate the folly in focusing purely on a team's perceived lack of change when a player leaves, at the exclusion of other, more meaningful pieces of evidence.)

  39. P Middy Says:

    We have a rule at work. 3 emails or more, and you pick up the phone. Similarly, you get 3 rebuttals or more, and it's not a bad idea to go to The Eye Test.

    Son was elite. Now he's not.

  40. David Friedman Says:


    I could just as well say that you seem "fixated" on the idea that the Wizards collapsed in 2009 primarily because of Arenas' absence--and there is a lot more evidence supporting my case than there is supporting yours. We have three years of evidence that even when Arenas was healthy (and had at least one All-Star caliber teammate) his teams were mediocre. I cited one example of an elite player (T-Mac) whose team did much better over a similar period of time when he played than when he didn't play. I asked you to provide counterexamples supporting your apparent thesis that someone can be an elite player without having much impact on winning but you have declined to do so.

    Furthermore, we have nearly a full season's worth of evidence that without Arenas--coming off of perhaps his best season no less--the team did not get any worse. These pieces of evidence suggest that Arenas had little positive impact on winning when he played (his team was right around .500) and that his absence had little negative impact on winning (his team was right around .500 when he didn't play). There is no evidence that Arenas' absence in 2009 had anything to do with the team's record.

    If the "data point" concerning a team's record with/without a player is that unimportant then it is odd that you chose to mention it so prominently; if it is important to consider all other relevant factors (as in the case of the 1994 Bulls, etc.) then perhaps you should have alluded to such factors in your original article. One could make the case that the Bulls did not drop off much in terms of regular season record sans Jordan precisely because Pippen was indeed an elite level player.

    I just don't understand the value of "efficiency differential" in this case if the actual, real world results differ so widely from what this stat would have predicted. You are reluctant to explicitly say it but we both know that if in November 2007 I had said that the Wizards would not miss Arenas that much in terms of win/loss record you would have answered that Arenas' "efficiency differential" proves his value and that for anyone to say otherwise betrays a lack of understanding of the superiority of advanced basketball statistics compared to regular box score numbers and/or the "eye test."

    Aren't you the least bit curious why "efficiency differential" does not seem to work in this case? The predictions spawned by the Theory of Relativity have produced great practical results for decades now, yet physicists still constantly run experiments to test if the theory is correct. The theory of "efficiency differential" is much newer than the Theory of Relativity and has not produced nearly the same kind of practical results yet you seem disinterested in trying to ascertain if it is really accurate, what its margin of error may be or if there may be other, better measurements.

    The semantics of defining elite (top five, top 10, top 20) is one issue but determining the predictive value and/or measurement accuracy of "efficiency differential" would seem to be an important issue for anyone who relies on this stat to make player and/or team evaluations.

    P. Middy:

    I don't have Neil's phone number and I thought that the comments section is the place to respond to articles posted here. I have raised some very important questions not just about Arenas but about player analysis in general.

    Also, I thought that at stat-oriented sites we are forbidden to rely on the "eye test." :)

    However, if we are going to use the "eye test" then my eyes tell me that during his prime Arenas was an undisciplined player who took bad shots, was indifferent at best defensively and frequently defied his coach (the latter is easy to verify by recalling how many times Eddie Jordan expressed great dissatisfaction with Arenas, including the times when Arenas seemed more focused on trying to score 50 points against the coaches who were on the staff that cut him from Team USA than he was on leading Washington to wins). Arenas was clearly an All-Star level talent due to his athletic ability, shooting skills and passing skills (when so inclined) but he was not consistent enough or disciplined enough to be ranked at the same level as truly elite players.

  41. P Middy Says:

    Yeah, but he shouted "HIBACHI" at dudes when he busted on them.

    But seriously, check out his 05-07 stats. 29 points, 6 assists, 4 rebounds, 2 steals. Sure the overall FG% isn't so hot, but his 3FG% is good. And his FT% is excellent. If 29,6,4, and 2 isn't elite for you, then I dunno what to tell you. You're maybe too discerning.

  42. P Middy Says:

    Which is not necessarily a bad thing!

  43. David Friedman Says:

    P Middy:

    There is no question that Arenas put up some good boxscore numbers during that time period and it is not like I am saying that he was a terrible player--but boxscore numbers are not everything. Stephon Marbury regularly put up 20-8 during his prime but there was a consistent pattern throughout his career that teams he joined got worse and teams he left got better; his individual productivity just did not correlate much with winning. I question just how much impact Arenas really had on winning, which is--as Herm Edwards admonished the world--why we play the game.

    I don't define elite solely by some kind of statistical threshold but rather by the overall impact that a player has and by whether or not that player could realistically be the best player on a championship team. Guys like Kobe, LeBron, Dwight Howard, Dirk Nowitzki, Shaq back in the day, Tim Duncan until recently are/were clearly elite players. As good as Arenas was in his prime, I just don't see him in that light. Put it this way: picture any of those guys I just mentioned playing for the Wizards instead of Arenas from 2005-08; do you think that the Wizards would have never won 50 games in a season and would have never gotten past the second round of the playoffs? The funny thing is that Nowitzki basically has Wizards South in Dallas (Butler, Haywood, DeShawn Stevenson comprised a significant portion of the Wizards' rotation during Arenas' prime) and the Mavs are one of the best teams in the league (yes, I know that Nowitzki has other good players, too, but Arenas had a younger Butler plus Jamison in his prime).

  44. Mike G Says:

    The 2008 Wizards had several of the league's most improved players, relative to 2007 --

    Caron Butler, Haywood, Roger Mason, Blatche, and Jamison all stepped up significantly to fill the void left by Arenas' injury.
    One might speculate that if Arenas had been healthy, and had delivered another 11 Win Shares (as he had in '07), then the Wizards win several more games. A conservative estimate would be about 50.

    As Neil said, one season's win total is one data point. Fixating on one point is like contemplating your navel for days on end.

  45. Mike G Says:

    Here's a Win Shares take on how the 2008 Wizards managed to be just as good without Gilbert Arenas' 10.8 WS.
    (formatting help?)

    returning   2008    2007    2008    gain   added
    Wizards      Min    WS/48   WS/48   WS/48    WS
    Jamison     3060    .123    .144    .021    1.34
    Stevenson   2564    .061    .077    .016    0.85
    Butler      2314    .103    .150    .047    2.27
    Haywood     2228    .095    .146    .051    2.37
    Daniels     2161    .116    .099   -.017   -0.77
    Mason       1708   -.005    .096    .101    3.59
    Blatche     1675    .033    .085    .052    1.81
    Songaila    1554    .078    .053   -.025   -0.81

    In their 2008 minutes, these 8 returning players are estimated to have contributed an additional 10.66 Win Shares over what their 2007 WS/48 would predict.

  46. P Middy Says:


    I understand and appreciate your perspective on the importance of winning when judging a player. However, I think it can be deceiving as well. Winning is a team effort, and you have to look no further than Paul Pierce to see that. Boston doesn't pull the trigger on those trades, and Pierce becomes a footnote, if that, in Celtics history. Now he's considered in their pantheon of greats. One could argue that Garnett has more to do with that than Pierce does.

    In any case, Gilbo, because he was ALWAYS a headcase, would never be an elite winner, even in a situation like Pierce has. He would screw that pooch. So I agree with you there.

  47. David Friedman Says:


    I am not focusing merely on one data point. Actually, I am doing the opposite: I am saying that less focus should be directed on the 2009 "data point" because it is unclear what that "data point" means: we neither know how good Arenas would have been at that time had he played nor do we know what impact his presence would have had on the Wizards' record in light of the team's other problems/weaknesses. Meanwhile, three other "data points"--Arenas' healthy seasons during his prime with the Wizards--show that Arenas' team had a mediocre record even when he played.

    We know that the Wizards were never very good when Arenas played and we know that the Wizards did not decline when Arenas missed virtually the entire 2008 season, so we have four "data points" that strongly suggest that Arenas' presence did not have much impact on winning. The 2009 season does not really have much to do with the Arenas story; certainly the data from 2010 and this season do not indicate that Arenas' return helped the Wizards, so why should we assume that he would have turned the Wizards around in 2009? Arenas had no discernible impact on winning when he played from 2005-07 or when he missed 2008 or when he played in 2010 and so far this season but we are supposed to assume that most of the Wizards' losses in 2009 are because he did not play? That is why I object to lumping in the 2009 season with Arenas' prime (as Neil did in this article) and then using that "data point" to suggest that the Wizards' won-loss record was dramatically affected by Arenas.

    Saying that other players' "win shares"/productivity/efficiency improved during the 2008 season is tautological; obviously the other Wizards played better or the team's record would have declined sans an All-NBA Second Team player. The important questions are:

    1) If Arenas was truly an elite player why did he have such a minimal impact on winning games during his prime?

    2) If Arenas was truly an elite player why did his teammates apparently play better/somewhat better/much better without him? (This "data point" contradicts the idea that Arenas was elite, because one would expect an elite player to make his teammates better/create more open shots for others/draw extra defensive attention).

    P Middy:

    There is no question that winning a championship and then making another Finals appearance has changed the way that Pierce will be perceived historically (playing alongside Kobe as a Laker has even more dramatically transformed Gasol from a one-time All-Star into a likely future HoFer)--but even before KG and Allen arrived, Pierce led the Celtics to the ECF (in 2002) playing alongside one All-Star (Antoine Walker) and an otherwise fairly pedestrian supporting cast that was not as good as what Arenas had around him during his prime. Also, the one year that KG had two other All-Stars around him in Minn. he led the team to the WCF. Arenas had two other All-Stars around him and could neither lead the Wizards to 50 wins nor get past the second round.

  48. Anon Says:

    "If Arenas was truly an elite player why did his teammates apparently play better/somewhat better/much better without him? (This "data point" contradicts the idea that Arenas was elite, because one would expect an elite player to make his teammates better/create more open shots for others/draw extra defensive attention)."

    This "data point" is a central issue in the skill curves model proposed by Dean Oliver, which has been shown to be valid in several stat studies. Arenas taking on the burden of his team's offense while remaining efficient (notably in 2006) was helping the Wizards, not hurting them.

    Even if you want to just go by the adjusted +/- regressions, that data shows him ADDING point differential to his teams when he played his best ball. So you can't look at these two pieces of data and draw the conclusion that the Wizards played better BECAUSE Arenas was not in the lineup "stat padding", as the old-schoolers would put it. Perhaps some Wizards players made improvements in their own games to contribute to their success without Gil - and as Neil noted, those improvements came on the DEFENSIVE end of the floor. They clearly missed Arenas offensively.

    I think you're looking at the glass half-empty here.

  49. James Says:

    Arenas was always a dividing figure for fans for a ton of different reasons. His attitude, attention grabbing antics, shot jacking, etc but the central reason was always that when Gilbert played well he was utterly dominant because he would routinely knock down low percentage shots that most teams/coaches will gladly give him or anyone else.

    The problem is that just like Antoine Walker that horrible shot selection was the most consistent part of his game other then his ability to get to the line and convert. So if you watched Gilbert play on any given night you might've seen an obvious Hall of Famer or you might've seen an Antoine Walker that could get to the line and convert.

    This is why Arenas was only able to lead teams to 40-49 wins in any given year despite having a fairly talented roster around him. He could lead them to wins against anyone but he would routinely kill his own teams against anyone too.

    In 05/06 he shot 50/40/83% in the Wizards 40 wins against 40/34/81% in the Wizards 40 loses.

    In 06/07 he shot 49/43/86% in the Wizards 39 wins against 34/27/83% in the Wizards 35 loses with him.

  50. David Friedman Says:


    Would Oliver's formula have predicted that the Wizards would perform the way that they did in 2008 sans Arenas? Plus/minus regressions, efficiency differential and other advanced stats can be useful tools but something is amiss here if the "advanced" numbers say that Arenas was adding wins but in his three prime seasons his team was only mediocre and when he missed most of his fourth prime season the team did not get any worse. Mike G's comment about multiple Wizards playing much better sans Arenas in 2008 suggests that a plausible theory worth investigating is that the gaudy individual numbers that Arenas put up did not really contribute much to winning and may have come at the expense of his teammates' individual numbers; it certainly seems like that in 2008 the Wizards absorbed the loss of Arenas through some combination of improving defensively and playing better offense by committee instead of relying primarily on Arenas--and this leads right into the very apt numbers that James provided: while Arenas may have been nearly as productive as Kobe or LeBron for a brief period of time according to "efficiency differential" the reality is that Kobe and LeBron are consistent players who have the ability to be explosive while Arenas was very much a high variance, all or nothing player. The way that I once described Arenas' impact is that if he shot .390 from three point range what the Wizards were really getting was one 6-9 performance in which they might lose anyway because of his porous defense and one 1-9 performance in which they almost certainly would lose because those long rebounds would essentially be like turnovers starting the other team's fast break. James' numbers fit that description: Arenas was "hot" when the Wizards won but erratic when the Wizards lost. James is also correct that, make or miss, Arenas took a lot of shots that the opposing team was happy to see him take (and that Eddie Jordan was not so happy with)--and the reason that the opposing team was happy was that, in the long run, those shots are losing plays, even if they lead to isolated victories here or there.

    Arenas famously dropped 60 on the Lakers once and Kobe took some heat when he--candidly and correctly--said that Arenas had taken some "terrible" shots. What most people don't remember is that the next time the teams played, Kobe had 39-6-6 on 14-26 field goal shooting as the Lakers cruised to a 118-102 victory over the Wizards while Arenas had 37-5-4 but shot just 9-29 from the field, including 3-15 (!) from three point range. So in that little "exchange" the Wizards squeaked out an overtime victory the first time around and then got their doors blown off in round two--a .500 record and a microcosm of why Arenas was not an elite level player.

  51. Anon Says:

    "Would Oliver's formula have predicted that the Wizards would perform the way that they did in 2008 sans Arenas?"

    I'm sure one could run a Monte Carlo simulation to answer that question.

    But David, even if you had no knowledge of the stats presented here, keep in mind that the Wizards went from 4th in offense to 12th in 2008, dropping two points in efficiency in a league environment that was more conducive to offense. Alot of that is because their star player wasn't in the lineup (the Wizards essentially had the same roster both seasons). Isn't that a reflection of what Gilbert brought to the table when he was healthy and playing at his best?

  52. Mike G Says:

    The Wiz did have an essentially unchanged lineup in '07 and '08, and it's a pretty good laboratory for the effect of removing a star player. But no formula can predict that 6 of 8 players are going to make dramatic improvements from one year to the next.

    Did Arenas 'depress' the statistics of his teammates? He probably took more than his share of shots, but he shot a lot better than most of the others.

    He missed the last 9 games of 2007, and the team went 2-7 in that span.
    Non-Arenas Wizards had shot .492 eFG% before Arenas went down; those last 9 games, they dropped to .476.
    Wiz opponents had shot .516 through 73 games, then .523 in the last 9.

    The team's TO% was 12.6 with Arenas, 13.3 without him.

    Team FT/FGA was .274 with Arenas playing, .251 without him.
    Opponent FT/FGA rose from .249 to .283 .

    Team ORtg was 110.1 for the year, after 9 games at 104.1 .
    DRtg wound up at 110.6, with 9 games at 109.4 .

    Losing 6.0 ppg on offense, gaining 1.2 on defense, translates to a net loss of 4.8 ppg.
    A team with a +4.8 point differential wins 51-52 games.

  53. David Friedman Says:


    I had knowledge of the stats being presented here regarding Arenas. Back when he was actually putting up all of those stats I wrote that I did not think that a team with him as the best player would ever get past the second round of the playoffs and all of those stats were recited to me, along with the charge that I must "hate" Arenas; I think it is safe to say that no one who strongly believes in the value of those statistics would have ever predicted--or even considered it possible--that the Wizards could subtract Arenas (in 2008) without suffering at all in terms of winning percentage. I don't "hate" Arenas, nor do I hate those particular stats--but I am not convinced that Arenas was ever an elite player, nor I am convinced that those stats have much predictive value concerning an individual's impact on a team's performance.

    Even if the stats that you are citing accurately describe Arena's impact offensively--and I am skeptical of that because Arenas was such a high variance performer--half of the game is defense and Arenas made very little contribution at that end of the court, something that not only affected his individual matchup but also trickled down to the rest of the team: when the team's best/most vocal player does not place a high value on defense it is very difficult to build a defensive-minded team.

    So, let's assume that you are correct and Arenas' presence made the Wizards much more efficient offensively; since their record did not get worse without him it is therefore logical to say that his presence at the defensive end had just as much negative influence: if you are going to say that he won games with his offense then you also have to say that he lost games with his defense. Otherwise, you are just looking at half of the picture and cherrypicking numbers to try to support the premise that Arenas was an elite player.

  54. David Friedman Says:

    Mike G:

    If a formula has no predictive powers (or very limited predictive powers) then what is its value? Obviously, no formula can perfectly describe reality or predict every possible outcome but if a formula says that a given player is an elite player and we end up with an almost perfect laboratory experiment that strongly suggests otherwise then it makes sense that at some point we should examine the formula to see if it can be improved in some way. This situation (the 2008 Wizards) is so fascinating precisely because it is about as controlled of an NBA experiment as one could hope to see in the real world: the Wizards lost Arenas in his prime for virtually a whole season but otherwise had essentially the same player rotation (and the same coach with the same philosophies).

    However, your example regarding the end of the 2007 season has serious issues. One, the sample size is obviously very small. Two, you neglected to mention that Butler also missed those games; I mentioned that early on in this thread, so perhaps you missed my remarks about this, but I pointed out to Neil that not only should the 2009 games that Arenas missed not be assigned to Arenas' prime but the Wizards' winning percentage during Arenas' prime was more sensitive to Butler being out than to Arenas being out: "(In 2007-08) the Wizards went 33-25 (.569) with Butler and 10-14 (.417) without him--and five of the losses with Butler also came with Arenas in the starting lineup. Washington's best starting lineup [in 2008] (by winning percentage, with a minimum of 10 games) was Butler, Daniels, Antawn Jamison, Brendan Haywood and DeShawn Stevenson. That group went 23-16 (.590) for nearly half a season without Arenas, which projects to a 48-34 record, a mark that would exceed the Wizards' best season since acquiring Arenas."
    The reason I noted that five of the losses with Butler included Arenas' presence is that we have not proven that Arenas had a positive impact on winning, so Arenas' presence may very well have "contaminated" the sample with Butler--for all we know, the Wizards may have won even more of the games that Butler played in if Arenas had been replaced by Daniels in the losses.

    In 2006-07, the Wizards were 37-26 when Butler played and 39-35 when Arenas played. If Butler had played more games in 2007-08 it is likely that the Wizards' winning percentage sans Arenas would not just have matched their winning percentage with him but even surpassed it. Butler missed 19 games in 2007 and 24 games in 2008 so our "laboratory experiment" is sound (in terms of rotation similarity other than Arenas being in one year and out the next) but since the Wizards fared much worse during this period without Butler than they did without Arenas it is certainly fair to speculate that if Butler had played, say, 80 games in 2008 then the Wizards may very well have broken the 50 win barrier sans Arenas (something that they never did with him).

    We also have another example, which I cited above, regarding T-Mac; the "laboratory experiment" is not quite as controlled but during the first several years that he played in Houston the Rockets essentially had the winning percentage of a 50-plus win team when he played and a lottery-bound 24 win team when he did not play (T-Mac's absence made much more of a difference than Yao's absence during that time frame, a time frame that pretty much coincides with Arenas' prime Wizards' years).

    So, as we look at our various "data points" and try to draw the most accurate possible conclusions, we see that when Arenas played alongside other All-Stars in Washington during his prime the team was barely above .500 and when he missed virtually an entire season the team hardly missed a beat--but when another team during that same era was without the services of their All-NBA guard their record dropped dramatically. I just don't see how one can look at all of these "data points" and state with great confidence that Arenas was an elite player.

  55. Anon Says:

    David, I would argue that an individual's impact on team defense in basketball isn't as strong as his impact on offense. Perimeter players aren't as important on that end as the men in the middle are, and while Gilbert certainly wasn't a lock-down defender (1.8 defensive win shares per 3000 minutes) I wouldn't tie so much of the jump in defensive efficiency to his absence from the lineup. Unless you think that Antonio Daniels was so much better than Gilbert at defense than Gilbert was than Daniels at offense.

    FWIW - and I admit that the sample size is small here - in the games that Gilbert played in 2008, the Wizards were actually pretty good on defense.

  56. David Friedman Says:


    The point is that there is not a valid way to prove exactly what an individual's impact on defense is; this is one of the shortcomings of basketball statistical analysis at this stage.

    Chess masters note that one incorrectly placed piece makes the whole position bad and I would argue that an analogy could be made to team defense in basketball: one incorrectly placed player breaks down the whole scheme. Also, as a point guard Arenas served as the first line of defense: when he broke down by allowing dribble penetration and/or leaving his man open this resulted in a chain reaction.

    Also, I don't understand why we should believe samples of questionable value (the 2009 Wizards' record, posted two years after Arenas last played at a high level and after he had injuries that likely permanently reduced his abilities), small samples or advanced metrics with questionable predictive value when we have larger, more significant samples that all suggest that even at his best Arenas was not an elite player:

    1) We have three seasons when Arenas was healthy and put up his best individual numbers yet could not make the Wizards into anything more than a slightly above .500 team.

    2) We have almost a full season that the Wizards played without Arenas right after arguably his best season and the Wizards did not decline despite making no other significant changes to their rotation.

    3) We have the stats that James provided, which support my observation about Arenas being a high variance player.

    4) We have the example of T-Mac, an All-Star/All-NBA guard who had a much greater impact on his team's record than Arenas did during the mid-2000s.

    All of these "data points" indicate that Arenas did not add many wins when he played, that the Wizards did not lose any wins when he was out and that a legit elite player (i.e., T-Mac) does affect his team's record. On the other side, we have individual metrics that indicate that Arenas performed efficiently offensively but these numbers tell us nothing about how this correlates with wins or how much Arenas cost his team defensively--and we have the fact that the Wizards were terrible in 2009, but it could be strongly argued that the absence of Haywood and other factors were more important than Arenas being out that season (particularly since Arenas being out in 2008 had no discernible impact in the standings).

  57. Anon Says:

    "Chess masters note that one incorrectly placed piece makes the whole position bad and I would argue that an analogy could be made to team defense in basketball: one incorrectly placed player breaks down the whole scheme."

    It widely known that some players are more important to defense in basketball than others. Namely the center and interior defenders - mainly because especially at the pro level, even the best perimeter defenders can be beat off the dribble, and a great defender in the paint can make up for it. These guys ANCHOR their squads defensively. The other important thing to keep in mind is that a staple of a great defense is great help defense. One-on-one confrontations certainly take place, but only within the context of a team defense that is primed and in position to defend wherever the ball is on the floor. Defense in basketball is more team-based than offense is, and the best defenses boast these two qualities.

    With all that said, it's hard to believe that Antonio Daniels, a player that started for Gilbert when he was injured, was THAT much better than Gilbert defensively (if he even was at all; he was ALSO a below average defender) he offset the advantages that Gilbert brought to the team offensively, thus enabling the Wizards to maintain a solid record. You're gonna have to rethink that one.

  58. westcoastslant Says:


    Perhaps Gilbert wasn't that much worse than Daniels on the defensive end, but he strongly influenced the entire squad as the "go to guy." As has been mentioned, Arenas is known to take shots not conducive to winning and was an "all or nothing" type player. Daniels, on the other hand, was a drive and get fouled type player who didn't force shots. He would more often than not make the right play. Case in point, Arenas has a career 1.7 assist/turnover ratio, while Daniels is at 3.09. While the overall offensive team rating dropped, the individual performances (as mentioned elsewhere here) of the other players suddenly became "career years" offensively when Arenas missed time.

  59. David Friedman Says:


    If you talk to any good basketball coach, he will tell you that a good defensive team has to be "on a string"--if one player is not on that string, then the whole defense collapses. You are right that it is very important to have good interior defenders but it is also important and valuable to have a guard who either puts pressure on the ballhandler or at least stays in front of him and prevents him from easily getting to the hoop (this depends on whether the coach's philosophy emphasizes pressure or just playing sound, fundamental defense--for instance, Larry Brown's best teams were known for their "jump/switch" defense, while Popovich's Spurs do not go for a lot of traps or steals).

    You are right that good NBA defense is not primarily based on one on one confrontations but my point is that Arenas is a bad team defender; he takes gambles and is often out of position and those kinds of things break down a team's defense.

    I am not sure what you are asking me to rethink. The author of this post theorized that Arenas was at one time an elite player and the first piece of evidence that the author cited was the Wizards' record with/without Arenas. I pointed out that 2009 should not really "count" as part of Arenas' prime and that during Arenas' prime we have three years in which he played and the team was little better than .500 plus one season in which he missed virtually the entire campaign without the Wizards declining. The body of evidence suggests that those who theorize that Arenas was an elite player have to do the rethinking.

    Also, I am not completely convinced that the issue here is as simple as figuring out how much Arenas added offensively versus how much he subtracted defensively. Arenas was a high variance player offensively during his prime, as James' stats pointed out. When Arenas was hot the Wizards had a better chance to win but victory was hardly assured--but when he was cold the Wizards were almost certain to lose. Removing Arenas from the lineup in 2008 probably led to much less variance offensively and that is another reason that the Wizards did not suffer as much in the standings as some might have expected.

  60. Anon Says:


    Offensive production for the Wizards dropped across the board sans Gilbert in the lineup. The only players who "improved" on that end were Haywood and Butler, and Butler went right back to his usual offensive production in 2009 (a season which Gilbert was also out of the lineup). This goes right back to the efficiency/usage model that has been discussed here multiple times. "Bad shot" Gilbert was still taking and converting tough shots that other teammates couldn't do as effectively without him on the floor.


    I'm not saying that perimeter defense isn't important, just not AS important as interior defense. You can go back throughout NBA history and see a great big man for pretty much every great defensive team (even the 90s Bulls, with terrific perimeter defenders like MJ and Pippen, had guys like Grant, Cartwright, Rodman, etc. in the paint). The bigs are more valuable to your team defensively, and you'll "miss" them more from your lineup than a perimeter defender.

    You're right about Neil using the Wizards record with and without Gilbert as evidence, but he isn't one to emphasize win-loss record with/without players in the lineup. I think he was mainly focusing on the points he made in the following paragraphs as proof of Gilbert's impact. By the way, WOUDLN'T it make sense that in games the Wizards didn't win, Gilbert didn't play well? He carried alot of his team's offense -- the difference between him at his his best and guys like LeBron who are obviously better players is that the teams they played on could at times compensate for subpar games from their stars with their defense. Something that the Wizards were never known for.

  61. David Friedman Says:


    This is not about the relative importance of interior defense versus perimeter defense; as I said, team defense must be played "on a string" and if one player is not on that string then it is impossible for a team to be very effective defensively. Moreover, the central question here is the thesis posed in this article that Gilbert Arenas used to be an elite player. I disagree with that thesis for the reasons that I already explained and I disagree with citing the Wizards' 2009 record as "evidence" that Arenas had a great impact on his team's record because (1) there are many other factors that explain why the 2009 Wizards were not good, (2) there is no evidence that Arenas had a great positive effect on the Wizards' record from 2005-07 and (3) the Wizards survived virtually an entire season without Arenas and posted the same slightly above .500 record that they posted with him during his prime.

    It makes sense that "in games the Wizards didn't win, Arenas didn't play well"--but it does not make sense to call a high variance offensive player who was also a poor defensive player an elite player unless there is some strong statistical and/or anecdotal evidence showing that he had an elite level impact. We have established that Arenas did not have a significant impact on his team's record and we have established that the Wizards were quite capable of absorbing the loss of Arenas' individual offensive efficiency. So what reason is there to believe that Arenas was an elite player. Neil made it quite clear why he enjoyed watching Arenas but there is a big difference between being a popular/fun to watch player and being an elite player on the level of Kobe or LeBron.

  62. Mike G Says:

    "We have established that Arenas did not have a significant impact on his team's record..."

    No, WE have not. You have this foregone conclusion, and evidence be damned.
    Several Wizards were entering their prime just as Arenas' chronic injuries began. The loss of Arenas negated major improvements among his teammates. With a healthy Arenas, the Wiz almost certainly win 50+ games in '08.

    This is of course just an educated guess. You guess that the team would not have been greatly better with Arenas. That's your prerogative, and maybe Arenas is one of those rare all-NBA players who do not add many wins. Would there be any others?

  63. Mike G Says:

  64. David Friedman Says:


    I don't have a foregone conclusion; I am simply reporting what four full seasons of evidence say: three seasons with Arenas healthy during his prime plus one season with Arenas not playing due to injury right after arguably his best season. I am also suggesting that the 2009 season is not relevant to this discussion, an important distinction because when we remove the "noise" in Neil's sample from those irrelevant games we are able to see that during Arenas' prime his presence/absence had no real effect on the team's record.

    You asked about other examples. I have already cited T-Mac as an example of an All-NBA player who had a very significant impact on his team's record. Since Neil is the one who wrote this article and proposed the thesis that Arenas was an elite player the burden of proof is really on him to find examples of elite players who had such little impact on their team's record for an extended period during their prime. The Arenas sample covers four years.

    During Arenas' three best healthy seasons with the Wizards the team never won more than 45 games or made it past the second round of the playoffs even though he had at least one All-Star caliber teammate plus a good supporting cast each season. A useful comparison is that Kobe Bryant led the Lakers to 45 wins in 2006 in the tougher Western Conference with no All-Star teammates, with a starting pg (Smush Parker) who never had success anywhere else and soon ended up out of the league and with one of the worst starting centers in the league (Kwame Brown). So, the individual efficiency stats that are being thrown around here say one thing but when we look at the bigger picture we see something else entirely: Arenas had a good supporting cast in a weaker conference but could barely get his team above .500 during his prime--the same team that could play slightly above .500 without him--while an elite player like Bryant was able to carry a very weak team to the playoffs in a stronger conference.

    A team's win-loss record is a fact. Individual offensive efficiency, as defined by the stats you are citing, is a theory; someone else could weigh the boxscore numbers differently and come up with a different theory. Also, even if Arenas was as efficient offensively as you believe there is still not an adequate measure of how inefficient he was defensively.

    Regardless of what the theory about Arenas' offensive efficiency says, the facts show that Arenas missed virtually the entire 2008 season, the Wizards made no significant changes to their main rotation, Caron Butler missed 24 games and the Wizards still posted roughly the same winning percentage that they did during the previous three seasons. Another aspect of this situation that is worth remembering is that the Wizards' record during this time frame was far more sensitive to Butler missing games than to Arenas missing games. You say that the Wizards might have won 50 games if Arenas had been healthy in 2008 but there is no evidence supporting that contention: he was healthy for the three previous seasons and they did not win 50 games. However, if you look at the Wizards' record with Butler but without Arenas during 2008 it looks like the Wizards might have won 50 games if Butler had been healthier.

    Which specific key Wizards' players do you think just entered their prime in the 2008 season? Jamison was 31, Butler was already an All-Star and Arenas' replacement Antonio Daniels was 32. There is a fallacy in believing that the Wizards who showed improvement sans Arenas would have been even better in 2008 if Arenas had played; perhaps these players performed better in 2008 precisely because they were not sharing the court with a shot-happy guard who did not play much defense.

    If your thesis is that Arenas had a significant positive impact in the win column during his prime--which would also mean that his absence with no other significant roster changes would have a negative impact in the win column--then you are should look for better evidence to support this thesis. My evaluation of Arenas is that during his prime he was an All-Star caliber player (i.e, one of the top 20 or so players in the league) but he was a high variance offensive threat who was below average defensively and displayed poor leadership qualities on and off of the court. When he did not play the Wizards missed his offensive explosiveness to some degree but ultimately they could live without this by performing better in other areas (playing better defense, being less explosive but more consistent offensively, having better game to game focus and work habits with guys like Butler and Jamison having a bigger voice). My evaluation is based on actually watching what took place when Arenas played and when he didn't play, as opposed to assuming that certain stats are infallible or that a player who is considered fun to watch by some (Neil mentioned that he is a big fan of Arenas) must be having a big impact on winning.