Readers of The Wages of Wins Journal have seen all season that, according to Win Score and Wins Produced, Kevin Durant was not having a great rookie season. Now that he has been named Rookie of the Year, I thought a brief review of Durant’s entire rookie season might be worthwhile.
Reviewing the WoW stories
Let me start this review by re-hashing the stories offered on Durant since he was drafted last summer.
As the following post from July 7, 2007 indicates, the college numbers fully supported the notion that Kevin Durant would be an excellent NBA player.
Looking Back at the NBA Draft, Part Two
Less than two weeks after this post, though, we began to see evidence that Durant might not be a stellar rookie. As the following posts indicate, he did not play well in summer league action. He was also not good in the exhibition season. And as the regular season progressed, he was again not very good.
July 17, 2007: Disappointing Durant
July 21, 2007: Durant Disappoints Again
October 31, 2007: Will Kevin Durant Be the Best Rookie?
November 16, 2007: Choosing the Best Rookie in November
November 27, 2007: Evaluating Future Stars in Baseball and Basketball
November 28, 2007: The Top Rookies, Again
November 29, 2007: Re-Hashing Durant, Melo, and Stack
December 31, 2007: Should the Rookie of the Year Help His Team Win More Games?
February 13, 2008: The Assistant Coaches Choose the Best?
March 25, 2008: Horford Also Tops Durant in March
The Rookie Durant
Now that the season is over, we now know that Durant was not a great rookie. Okay, we who look at Wins Produced (which might just be me) know this. The sports media has selected Durant as the Rookie of the Year. In other words, the media thinks Durant was the best rookie.
Durant did lead all rookies in scoring. But when we look at all the stats, it’s clear that Durant has problems his rookie season.
A player can accumulate points by being an efficient scorer and/or taking a large number of shots. Durant’s scoring was really about taking shots. His adjusted field goal percentage (45.1%) was well below average. He was also below average with respect to steals, turnovers, net possessions (rebounds + steals – turnovers) and assists.
For a shooting guard he did show that he could rebound and block shots. He is also able to get to the free throw line. But these positives were swamped by his negatives, and hence when we look at Win Score – which summarizes all the box score statistics – we see a below average player. An average shooting guard will post a Win Score of 6.1 per 48 minutes played. Durant only offered a Win Score of 5.2 (per 48 minutes played).
A similar story is told by Wins Produced and WP48 [Wins Produced per 48 minutes]. For the season he produced 0.7 wins and posted a 0.012 WP48. Again, these are well below average marks.
What of the other rookies? When the All-Rookie team – selected by the coaches — is announced I will comment on the entire 2007-08 rookie class. For now I will simply say that of the rookies who received votes from the media, Al Horford, Jamario Moon, Luis Scola, and Carl Landry were above average performers. Each of these players would have been better choices than Durant (in fact, of those receiving votes, only Juan Carlos Navarro and Al Thornton offered less than Durant).
Answering the Arguments for Durant
Supporters of Durant will offer excuses for why he did not produce this season. They will also try and argue that he seemed to improve as the season progresses. And they will argue that someday Durant will be the best player chosen in 2007.
To these arguments I say…
1. The fact is he did not produce. The award for Rookie of the Year should go to the player who did play well, not the player who might have played well in different circumstances. If we are going to follow that logic, give the award to Greg Oden. He also might have been the best rookie if he simply didn’t get hurt.
2. Durant might have played better in March and April. The award, though, is Rookie of the Year. It’s not, Rookie of March-April. The first four months of the season count and we should not ignore these games in giving out awards that represent a player’s accomplishments for a season.
3. And to the argument that Durant might someday be better…I think that is entirely possible. But again, this is an award for what a player did as a rookie. And as a rookie, Durant was bad. No matter what he does going forward, that fact is not ever going to change.
– DJ
Our research on the NBA was summarized HERE.
The Technical Notes at wagesofwins.com provides substantially more information on the published research behind Wins Produced and Win Score
Wins Produced, Win Score, and PAWSmin are also discussed in the following posts:
Simple Models of Player Performance
What Wins Produced Says and What It Does Not Say
Introducing PAWSmin — and a Defense of Box Score Statistics
Finally, A Guide to Evaluating Models contains useful hints on how to interpret and evaluate statistical models.
Wages of Wins Journal 
antonio
May 1, 2008
i am one who has made some defenses of durant, but i just want to be clear that i never thought durant was rookie of the year. I just think he has shown enough this season to make people believe he will become great (which is why I think in fact he won the award). Anybody who seriously looks at the statistics knows Durant was not the best. I just think the voters made the mistake of voting on flashes of greatness and potential
Tomo
May 2, 2008
Well I think that a quick look at how Horford has been such a big part of the Hawks 2 wins against Boston told us that he was the best Rookie this year and will become a great player in this league, probably an all-star…
Durant showed talent, no doubt about that, but did he really try to find a way to make his team and teamates better….I don’t think so…
Horford did it for 87 games (82 + 5 playoffs)…
LeoneL
May 2, 2008
Great work on the stats. I must say that I was rooting for Horford to bag this award. But then again, I know that those who voted for Durant had their own valid reasons.
By the way, Durant said: “I talked to Al (Horford) and he congratulated me yesterday. I was very happy for him and his team. I’d rather trade this in to be where he’s at right now. To be in the playoffs.”
See, even Durant may agree with you. hehehe
Tommy_Grand
May 2, 2008
Your argument is very persuasive. Except this sentence:
“Durant might have played better in March and April. ”
Why say “might?” Why not simply admit/deny he played better in March and April? Do you lack an opinion on this assertion?
Is this trend relevant to ROTY? I have no idea. I suppose it would be reasonable to use as a tiebreaker. It certainly shouldn’t defeat the sound case in favor of Horford, Moon, Scola, or Landry. But is his marginal late season improvement a fact or merely a supposition?
To my eyes, it looked as though KD improved as the season wore on and peaked in the last week.
Owen
May 2, 2008
Teh word “might” can be used in that sense without conveying that the assertion is false. As in, “George Bush might have been the best candidate for the Republican Party but that doesn’t mean he was the best prepared to be President.”
At least, that is how it seems to me…
Jason
May 2, 2008
I suspect that many pro-sports awards are like political elections in that inertia becomes a more powerful force than reason. My sense is that Durant won the award sometime around the time that Oden’s microfracture was diagnosed. Prior to that point, there was a race, but at that point, Durant became the most “electable” rookie. The only thing he had to do to avoid *losing* it in the mind of many voters was to score points. That’s the most conspicuous thing a player can do, and though in isolation it doesn’t win games, it wins awards.
Jason
May 2, 2008
It looks like the Hawks’ coach similarly thinks the award was not deserved: “http://sports.espn.go.com/nba/news/story?id=3377015”
from the article: “Durant lived up to expectations after being the national college player of the year at Texas and the No. 2 overall pick in the June 2007 draft.”
I guess expectations were that he’d shoot all the time and score many points at the expense of many, many missed shots and wasted possessions and that he wouldn’t do much if anything to actually boost the win total.
Faizan Hassan
May 2, 2008
It’s the first time I find myself agreeing with Mike Woodson. It’s a travesty
Cal
May 2, 2008
Durant would have helped a good team make the playoffs as well.
k
May 2, 2008
Individual players have effects on their teammates which cannot be ignored. Unfortunately, your analysis ignores those effects.
Harold Almonte
May 2, 2008
And nobody have said the award means Most Valuable Rookie. Durant deserves the award, even not being his game as balanced at both side of the floor as Horford’s, but It should have been a little more closed. The votes don’t reflect the competition.
Faizan Hassan
May 2, 2008
Unfortunately, this award is also not about what would have and could have happened on different teams. It is about what actually happened.
Nick
May 2, 2008
Question that I think is relevant for this debate: how does (or simply does) win score or wins produced account for the different levels of risk demanded by different roles, not positions, on a team?
It would seem that Durant, being his team’s primary scorer and offensive option, would have more possession, face tougher defenders, and often have more difficult shots. Certainly, some of those shots are of his making, but inevitably teams have to take difficult shots b/c of good defense/shot clock constraints and usually they look to their primary options to take those shots.
Put another way: not all FGA are created equally, and not all difficult shots are the product of poor decision making. Primary offensive options face a more difficult task, and it seems misleading to compare their FGA to someone’s like Horford. Moreover, it doesn’t seem exactly right to compare Durant to Ronnie Brewer a 4th/5th option SG. Shouldn’t he, instead, be compared to other primary perimeter options like Kobe or Lebron? He still compares unfavorably–and I’m not saying he played well–but it seems to put him in a more relevant context.
I think maybe this would fall into the “why” as opposed to the “how,” but if that’s the case doesn’t it in some part dilute the “how?”
Tommy_Grand
May 2, 2008
“…Teh word “might” can be used in that sense without conveying that the assertion is false. As in, “George Bush might have been the best candidate for the Republican Party but that doesn’t mean he was the best prepared to be President.”
No joke, bro. But it’s stronger to say “Regardless of Durant’s clear improvement in March and April…”
or
“Even though no other Rep. candidate was better prepared than Bush, that doesn’t imply….”
The point: if you’ve no idea how Durant played in March and April (32% of the season?), you shouldn’t posit a conclusive opinion on his season.
If you do know how KD played in March and April (as this author surely does), you shouldn’t elide that evidence. To say Durant “might” have improved, if+when you know he did, is churlish and weakens the (otherwise sound) arg.
Jason
May 2, 2008
Not all FG attempts are equal, but whether or not Durant is the focus of the opposition’s defensive focus (and I’m leaving that as an “if”) does not make his misses any less costly. If, against strong defensive pressure, he’s unable to make a shot, he shouldn’t be taking the shot. He’s the “first option” but based on his shooting percentage, he probably shouldn’t be. I know many cite player X Y or Z’s poor shooting and will often defend it on the grounds that his team didn’t have anyone else, but more often, the disappearance of a high volume-low percentage shooter does *not* result in lower FG percentage by the remaining players.
What a player does is a product of many things, but on empirical grounds, his own ability seems to be the greatest single factor, far more than the focus of the defense or the quality of his teammates.
Nick, the WP model compares individuals at a position against others at that position by adjusting for the average production at that position. In this way, Durant is not expected to shoot as well as Horford, nor rebound as well. Their disparity is in addition to the relaxed requirements of production at their respective positions.
Point simple: Durant hurt his team by shooting as often at such a low percentage. He does not somehow become more valuable because he did this to spare teammates from missing the same shots.
antonio
May 2, 2008
What I will say about Durant’s shooting percentage- higher than Lebron’s in his first year or Carmelo. There was another player too (i cant remember who, it was an espn graphic), but either way, it helped me put Durant’s season in perspective. In other words, I don’t what other people think he will become or thought of his first year, but shooting efficiency can some time to develop, especially for high scorers
Nick
May 2, 2008
Jason,
“the disappearance of a high volume-low percentage shooter does *not* result in lower FG percentage by the remaining players.”
But isn’t it the case that Iguodala and Korver, the two remaining players who got most of the shots after AI, shot much worse (eFG%)? Further, didn’t Miller’s effective shooting percentage drop when he came to Philly (year 1)?
Also, what about the case of Joe Johnson/Jason Terry.
Jason Terry leaves ATL where he was one of two primary options, takes fewer shots in DAL, and improves dramatically. Joe Johnson enters the new offense and sees a remarkable drop (not to mention a huge p/min TO increase).
Or how about Pierce having by far the best eFG% of his career once he’s playing with more talent than ever before?
I’m not saying that this will explain all players–some players are just bad shooters no matter their team/role–but in some instances couldn’t it be the case that they’re stuck being the primary/secondary option in a bad offense?
You say that a player shouldn’t take a shot he’s unable to make, but that’s a simple-minded view. Sometimes, when the offense is bad, there’s only bad shots. And someone has to take them. To blame the volume shooter seems too easy. I mean, Philly has shot the same FG% for the past three years and AI only played in 87 of those games.
Couldn’t it be the case, they’ve just got a bad offensive concept there?
Jason
May 2, 2008
Well, thank you for deriding my view as “simple minded”, Nick.
It’s not so much blaming the volume shooter as it is not giving someone credit for shooting in high volume. No, Philly’s offense didn’t change significantly when Iverson departed and others started taking the shots (though the team *did* improve marginally). Nonetheless, the popular opinion was that Iverson was a great player, largely because he scored so much and the popular opinion (expressed by the projections that the team would tank without him) was that this was valuable and would be hard to replace. It wasn’t. Blame him? Not necessarily, but neither should he get for being the guy who pulls the trigger in a losing effort.
I get the impression that many people believe that players like Durant and Iverson take their high volume shots because everyone else on their team is so bad that if they didn’t miss 60% of their shots, their teammates would miss even more. I don’t think the data bears this out more often than not.
antonio
May 2, 2008
The Sixers I dont think is an example that works becuase they improved defensively once Iverson left. Which is not what WoW states would happen. WoW would predict an improvement on the offensive side of the ball when iverson was traded for miller that did not happen. While WoW predicted the improvement, it predicted it for the wrong reasons in that case
antonio
May 2, 2008
“I get the impression that many people believe that players like Durant and Iverson take their high volume shots because everyone else on their team is so bad that if they didn’t miss 60% of their shots, their teammates would miss even more. I don’t think the data bears this out more often than not.”
I agree with this statement. I bet Durant does too. Look at his game logs from march and april and he really started to take less shots for the most part. A lot of times he would take shots somewhere in the 13-17 range. And most of the time when he took more, he was “hot” for the game and shot very well. I think when you take great scorers like Durant out of college, it takes a time of adjustment to learning not to take every shot. In college they got it off with such ease. Some players like Iverson just don’t learn it. But being that Durant is younger than Iverson when he came out and just a different person, there is no reason to give him the “volume shooter” label or as somebody who is going to be a low percentage shooter just yet.
antonio
May 2, 2008
Sorry for the consecutive posts, but to add on, especially considering Durant’s extremely high free throw percentage. This indicates a good “pure stroke”, and as he gets more comfortable in the league and gets stronger, I think he will be a pretty high percentage shooter. Almost all players who shoot 87% at the free throw line are great shooters
Nick
May 2, 2008
First, I didn’t mean to mock by calling that particular view simple-minded. Sorry if it was taken that way. But I do stick to my argument that bad shots are not coextensive with bad decisions. Any statement to that effect, such as “If, against strong defensive pressure, he’s unable to make a shot, he shouldn’t be taking the shot,” is simply wrong unless we’re assuming some kind of ideal offenses, but those don’t exist.
But more to the point: My problem with Win Score is not that it doesn’t give credit to players for taking a lot of shots, but that it rewards players who do not. Put another way: there weren’t any more or less missed shots in Philly before or after AI, they were just spread out among more players after he left. So when AI was there, the other players Win Scores were rewarded because he assumed the brunt of the risk. Take him away and the team becomes no more efficient, at least in shooting; they just spread the inefficiency across more players.
So the question is: are we getting a good picture of a player’s value if we are attributing to him, individually, shooting inefficiencies that appear to be more inherent to the offense than to his individual game?
And isn’t it of some importance that Denver’s shooting has remarkably increased since his arrival and Iverson, individually, is shooting better, by far, than he ever has?
Now insofar as people thought Philly would fall apart without AI, that was wrong, and there’s no question that in the mainstream AI is overrated. But I think that, ultimately, Win Score might be flawed for individual players insofar as it is unable to distinguish the inefficiencies of a player from the inefficiencies of the teams. Essentially, it can’t tell what is and what is not a tough shot or, more importantly, why one player is taking those shots. More often than not, it will be a very good indicator, but in some instances I think it misunderstands high-volume shooters.
So in the case of Allen Iverson, it has underrated him, at least as a shooter, because he simply swallowed the inefficiencies of Philly’s offense instead of his teammates. Now someone Antoine Walker…well, that’s spot on.
Jason
May 2, 2008
With a position adjustment, there is *not* a reward for failing to take shots since taking shots is still a pre-requisite for making shots and made baskets are worth more than no baskets. Since the position adjustment assumes an average production at the position, sitting on the court and doing nothing results in a penalty. Granted, a player who takes shots without making shots is penalized *more* than the player who never takes them, but this is not the same as rewarding someone.
Nick
May 2, 2008
Position adjustment will only not reward players if you assume that difficult shots are proportionately distributed among players no matter how many shots they take.
What I am arguing is that “volume shooters,” either by their own decision or because of the offense itself (or both), take a disproportionate number of the bad shots because as the “go-to guy,” they’re expected to take them. Thus, position adjustment tells us little (in regard to shooting at least), since it compares players who take few difficult shots, Ronnie Brewer, to players who take many, Kevin Durant or Allen Iverson.
Now if the bad shots are the result of poor decision-making (Antoine Walker), then it makes sense to penalize him for taking them (and missing most of them).
But in the case of Allen Iverson, just as he did not create shots, he did not create misses. Those misses were apparently part of the offense.
Thus, he’s overrated insofar as you think that he was making more of the difficult shots than his teammates could; they apparently were just as capable. But he’s underrated insofar as you think his game produced the misses. The offense in Philly can apparently only create about 46 made shots out of one hundred no matter who’s shooting them.
Now if you have a case where an offense shoots better when a volume shooter moves on, then you likely have either poor shooter who couldn’t make good shots or a poor decision maker who took bad shots.
Conversely, if an offense shoots worse when a shooter moves on, then he was likely creating more good shots for the offense or was better at making difficult ones.
dustin
May 2, 2008
according to 82games.com
Brewer (08): Utah has a 2.1% increased eFG% when he is on the court. He shot at a .567% eFG%
Horford (08): Atlanta has a .7% increased eFG% when he is on the court. He shot at a .499 eFG%
Durant (08): Seattle has a 1.5% decreased eFG% when he is on the court. He shot at a .452% eFG%
Iverson (08): Denver has a .7% decreased eFG% when he is on the court. He shot at a .488 eFG%
Iverson (07): Denver has a 1.9% decreased eFG% when he is on the court. He shot at a .481 eFG%
I know this is hardly proof, but it appears the anecdote you provide is incorrect.
Efficient, low-volume shooters such as Brewer and Horford do not seem to be passing “bad shots” onto their teammates, as the team as a whole shoots better when they are on the court.
Inefficient, high-volume shooters are not “hogging” the bad shots all for themselves, as the team as a whole shoots worse when they are on the court.
antonio
May 3, 2008
i dont think that works either. This seems like selective choosing. The point is, neither can be proven with examples. There will be contradicting players if you try to provide player examples
Nick
May 3, 2008
I’m not sure that these statistics prove your point.
First, to be clear, the point I was making was that volume shooters will take a disproportionate amount of the difficult shots. In some cases, this is because they’re bad shooters or bad decision makers, thus the team will improve when they’re gone. Sometimes, however, it’s because the offense as a whole is inefficient and they’re simply the best player to take the difficult shot, in which case the offense should get worse or, at least, stay the same when they’re gone.
Now with that in mind, your examples:
Durant – I never said Durant was the good kind of volume shooter. Rather, I’d say he’s a bad one who takes unnecessary bad shots. Thus, it isn’t surprising the team shoots better when he’s off the floor.
Brewer- It’s true that the team shoots better when he’s on the floor. Now first, I’m not sure what this is really supposed to indicate, since the argument is that guys like Brewer are more likely to be taking good shots. So that the eFG% goes down without him could just as well indicate that the offense itself is less efficient without him–less good shots are available–so more bad shots have to be taken. But whether he’s there or not, it’s the same guys taking the bad shots, the “go-to” guys, not Brewer (or not as often Brewer).
Even if it is relevant though, it is also the case that all the Jazz starters have high +/- on eFG%. In fact, Brewer’s is by far the worst. The closest to him is Kirilenko at -2.8. So I think what you’ve really found evidence of here is that the Jazz backups can’t shoot.
Horford- the argument is the same. 1) I’m not sure this is even relevant 2) if it is, his +/- is the worst among Hawks starters. Once again, they have a poor second unit.
Iverson 06-07- I don’t know how much stock I can put into this statistic, since it includes the like 1000 minutes before he was even on the team. But that said, it does look like he was a drag on the offense in a bad way, in spite of his improvement as an individual. In fact, his individual improvement may have been the product of the better offense. Further, it doesn’t seem that odd that a player as obtrusive as Iverson would have a negative impact when immediately thrust into a new team. But yeah, he was probably the bad kind of volume shooter here.
Iverson 07-08- This one I think is misleading. Obviously, the transition period is over here, so if Iverson is still a drag, there’s not much excuse.
It doesn’t seem like he is though. Granted, it’s true that the eFG% drops some (.7%) when he’s on the court, but that’s only part of the story. The points per 100 poss also drops 8 points when he’s off the cour and the turnovers increase by 5. So what I would guess here is that you have an offense that, without Iverson, is turning at least some of its poor possessions into TOs instead of bad shots. Thus, you get better shooting at the cost of worse offense. In fact, the point and TO differentials of Iverson are the greatest on the team.
So it’s not clear that he’s taking the bad shots with him to the bench, since they may just be lost in the TOs.
But now once again, I wasn’t talking about Iverson with the Nuggets; I was talking about Iverson with the Sixers.
So let’s look at a couple of those years (note: I don’t use 06-07 b/c he only played 15 games: that’s a weak sample. But even then there’s only a -.4 difference, which considering he was shooting .429 on a .458 team probably speaks more to my point):
Iverson 05-06: His eFG% .467; The Sixers .487
But when he’s on the court: .494
When he’s off: .461
Iverson 04-05: His eFG% .453; The Sixers .474
But when he’s on the court: .482
But when he’s off: .446
Now if these +/- stats are relevant (and I’m still not sure they are), those seem to speak pretty loudly to the fact that AI was making the hard shots or creating good ones to a certain extent.
Once again though, I’m not sure +/- is really the right stat here.
Harold Almonte
May 3, 2008
“If, against strong defensive pressure, he’s unable to make a shot, he shouldn’t be taking the shot. He’s the “first option” but based on his shooting percentage, he probably shouldn’t be.”
Statistics and Economy pretend to convert this approach in “the refutation of a basketball mith”. But, like somebody said, the offensive system, the defense, the limitations of the floor, and the limitations of the individual skills to keep ever handling the ball more than two seconds; don’t produce neither the same, nor equivalent shooting opportunities for everybody. Simple formulas pretend that the strenght of the opponent defense is the same in all shooting situations, that good shooting opportunities can be automatically created at every possible attempt, that the true and only quantifyable scoring help is a final pass, and that players are the same type of shooter at every quantity and quality level of usage.
Basketball is more complex than that. In order to get a pair of cheapy “two points” from a low usage-high efficient shooter, it’s needed lots of expensive (somebody once said “riskies”) “two points” fom others, because all shots aren’t free throws. Like somebody said, sometimes the efficiency of a player is just the efficiency of the system, and some players are more “owners” of his points than others.
Basketball metric actually passed away prehistory, but still far from taking the automatic command of strategies, and awards.
Harold Almonte
May 3, 2008
When against”relatively” easy defense, a player is unable to score… this is how “options” are decided, not the opposite way.
Stewart
May 3, 2008
Your argument against Durant is flawed in one glaringly obvious way. You place too much of an emphasis on Win Shares and a player’s impact on wins. Obviously winning is the goal of any team, but Durant played on a team that had basically tabbed this as a “developmental year.” Furthermore, Durant was the only one of the rookies you listed that was expected to shoulder the scoring load. Horford had Johnson, Smith, Childress, and recently Bibby. Landry and Scola had Mcgrady, Ming, and Alston. Moon had Bosh, Ford, and Kopono. It’s a lot easier to put up efficient stats when you’re the third or fourth option. Also, those other options obviously make your team better and more likely to win. I think Durant showed remarkable poise for a 19 year old facing defenses planned exclusively for him. I think the sky’s the limit for this kid…
Jason
May 3, 2008
“Stewart”, the argument was originally that Durant didn’t produce much this year, that his contributions didn’t positively effect his team’s win total. That’s not a flawed argument. That’s empirical reality. That there are reasons for this (e.g. the team was rebuilding, he was expected to shoulder the scoring load even if it was going to cost wins) is a different issue. It’s an issue that may explain why he could be expected to get better–and no one has said that this isn’t possible–but it’s different from saying that this year he wasn’t particularly valuable. Regardless of why he wasn’t, he wasn’t. That’s just reality. If ROY is for the most valuable rookie, it’s hard to say a player who didn’t produce a positive effect on his team’s win total is particularly valuable. Or are you arguing that it’s an award for “rookie with most potential to be really good in the future”?
The premise that it’s easier to put up “efficiency stats” when you’re something other than the primary option seems to be a supposition presented often but supported with data far less often. It’s an interesting question and may or may not be true, but it certainly seems that the arguments surrounding it are based on assertion that it’s so more than anything else.
Nick
May 3, 2008
Well, the question is what is sufficient evidence. To be specific: what I’ve been arguing is that Iverson is being unfairly attributed inefficiencies (missed, difficult shots) that appear more inherent to the Philly offense and combination of players.
Now the evidence that I have is:
1) Even though Iverson took around a quarter of his teams shots during 04-05 and 05-06 and shot significantly worse than the team as whole, the team shot significantly worse when he was off the court.
2) The 06-07 team, despite losing a shooter who shot far below their team eFG% and took 24 shots a game, did not shoot any better after he left.
3) The four players who absorbed most of Iverson’s shots after he left–Iguodala, Korver, Smith, and Miller–all shot worse in the post-AI.
4) In 07-08, the Sixers continue to shoot the same as they did in 06-07 and 05-06.
5) Iverson, since leaving the Sixers, has improved his shooting to career high numbers.
Knowing this, I’ve asked two questions:
1) If the loss (whether by trade or just substitution) of Iverson’s apparently poor (i.e. far below team average), high volume shooting did not and has not improved Philadelphia’s shooting in the past four years, and if his exit from that offense has significantly improved his shooting, is it possible that whatever inefficiency his shooting reflected during those years is actually caused, at least in part, by the Philly offense rather than by Iverson’s shooting or shot selection?
2) If that is the case, then does Win Score give us an accurate estimation of a player’s value if it attributes to players inefficiencies of an offense that would apparently be there whether or not that particular player was there?
Applied more to this case: Offenses have bad possessions either because of good defense or bad offense. Those possessions will typically lead to a bad, missed shot or a turnover. If one player is the cause of these shots or these turnovers, then a) should be discredited for these wasted possessions and b) the team should improve when he’s gone.
But if “b” isn’t true then couldn’t it be true that the player isn’t actually the cause, but instead the offense (the sets and player combos) is the cause. So then should we discredit one player for taking the risk that had to be taken (someone had to shoot) but was created by the inefficiency of the whole team?
I don’t think this will be true of all or even most volume shooters. And it may be true of role players who for some reason end up assuming most of the risk for their teams. But I think it is true for Iverson, who appears to be the scapegoat for what has been a pretty poor offensive concept in Philadelphia for a long time.
In part, this is likely due to the fact that he has been overrated. The Sixers apparently at some point determined that he would be better at assuming the risks and even if that were true in 04-05 (and it’s tough to tell), it doesn’t seem it was by 06-07.
But it underrates Iverson to imply (as Win Score does) that he was the cause of these shots that apparently no one on Philly was going to make anyhow.
Also: it is true that Philly’s offense as a whole (out side of just shooting) has improved, but it’s improved by accumulating more possessions. Fewer turnover, more steals, and more offensive rebounds. So I think you would likely have a good argument that at least Iverson was counterproductive in regard to TOs (I’m not sure how much oREB would have to do with him).
But the TOs shouldn’t be surprising since Iverson is not a good PG though the Sixers were playing him there (he’s a shooting guard), and he was replaced by a very good one in Miller.
magicman
May 5, 2008
Durant didnt have any allstars around him this yr like al horford or the rest of the rookies did. Most of the people on his team were nobodies. If you would have put an allstar or kept one on this team wit him he might have had as good of a season as horford or maybe even lived up 2 the hype
Vince
May 5, 2008
I agree with Nick’s points to a certain extent. High volume shooters probably take a higher % of late / difficult shots. And coaches are responsible for offensive design / play calls and so players are not the sole agent on their stats. But they also get the first best looks a lot by design and be grabbing them. Ultimately first option guys should be judged against others and down the usage line.
dustin
May 5, 2008
H0w do you konw who a “first” option is just by looking at the box score?
Magicman, it is empirically proven that there is DIMINISHING returns with respect to putting good players together (with winscore). Also, while Joe Johnson was an all-star according to the coaches, win score put him at about average.
Jason
May 5, 2008
Magicman, if the problem were merely Durant’s supporting cast, it wouldn’t explain why the Sonics appeared, by-and-large, to play *better* without him on the court. There are problems with plus-minus approaches, but by most of them, the Sonics performed better when he was on the bench. Durant’s Sonics were bad, but their scoring margin was *worse* when he played. It’s hard to blame his teammates on his performance when they didn’t affect his replacements in the same way. His replacements out-performed him.
Vince, a few months ago when someone was commenting on a ‘star’s’ shooting percentage reflecting that he would always be the guy to take the shot as a possession was winding down. The thought was that even a bad attempt with a low chance of going in was better than letting the clock run out. That thought is correct. However, the evidence that this need to get of some sort of shot impacts the overall stats of certain stars is high, highly lacking. The Nuggets took 7% of their shots with less than 3 seconds left on the shot clock. That percentage is completely identical to Iverson’s own shot clock distribution as it was ‘Melo’s. They were not taking a disproportional percentage of the last second shots relative to their offensive usage overall. Similarly, Kobe Bryant took a percentage of his own shots late in the clock (12%) similar to the Laker’s overall “late clock” percentage (11%). Durant was actually *less* likely to take that late shot than the team’s shot clock usage overall would predict (8% of Durant’s shots in last 3 seconds vs 11% for the Sonics as a whole.)
Vince
May 5, 2008
Jason you are right on Kobe and Iverson. I said probably because I hadn’t researched it. But checking a few others I see DeRon Williams and Tim Duncan with notably higher crunch time % of shots than team. It doesn’t apply to all stars but it applies to some. Lakers and Nuggets are surprisingly normally distributed in crunchtime though that is not necessarily good or bad.
Vince
May 5, 2008
Dirk and LeBron have an elevated share of crunchtime. Yao and Al Jefferson are modestly elevated. It is looking not that common. Good to know.
But Tayshaun Prince takes 27% of his shots in last 3 seconds compared to 15% for team.
Vince
May 5, 2008
Gerald Wallace and Jason Richardson are both pretty elevated. As is Zach Randolph. It may not be as common as I thought might be the case but there is a group who do shoulder a heavier crunchtime load.
Nick
May 5, 2008
This is interesting stuff. I’m a little hesitant about its application though since it hinges on the assumption that shots taken late in the shot clock are bad/difficult shots. I’m not sure that’s true. And the argument is that guys like AI, Durant, and Kobe are taking more difficult/bad shots.
But here’s some interesting trends:
If we look at 05-06 Philadelphia and compare Andre Iguodala and Allen Iverson, we see that Iguodala actually took a higher proportion of his shots after 21+ seconds. And he shot a much better percentage than Iverson. But if you look closer you’ll see that 70% of Iguodala’s baskets were assisted. Only 23% of Iverson’s were. Now we’re still dealing proxies here, but I’m going to guess that an assisted basket is much more likely to be a good/easier shot than an unassisted basket. So one could argue that Iverson is still taking far more of the difficult shots.
But this isn’t as persuasive as I would hope. Because this assisted percentage ratio pretty much matches up for both players no matter when they shoot (i.e. Iverson is rarely assisted at any time 1-24, and Iguodala is assisted at more or less the same rate 1-24). This trend holds true for Kobe as well. He gets far fewer assisted buckets than anyone else on the Lakers (unsurprising). It’s also true for Iverson on the Nuggets (but not true for Anthony however who gets as many assisted buckets as anyone else).
So the real difference it seems between guys like Kobe and AI and other players it seems is that they take unassisted shots, which might at some level imply that they’re taking too many bad shots at least early in the shot clock. This seems intuitively correct. But here’s a surprising stat: on jump shots and close shots, despite not being assisted nearly as often, both players shoot better than their team averages. Compare this to Iguodala (05-06) who, despite being assisted far more often than average, shot worse on close shots and jump shots.
Thus, when you look at AI or Kobe’s low field goal percentages, they’re misleading, since the disparity between their shooting and their teammates lies primarly in the disparity between the number of assisted dunks and tip ins their teammates get. This is especially true for Iverson, who gets no dunks or tips.
So now let’s have an exercise: Interpret these statistical trends:
1) The Sixers have not shot better since Iverson left the team.
2) While he was there, they shot worse when he was off the floor.
3) He’s shot better since he left.
4) He takes far fewer assisted shots than his teammates.
5) Dunks and tip ins aside, Iverson shoots better than his teammates despite getting far, far fewer assisted baskets. However, overall, his eFG% is far worse than his teammates.
6) Since joining the Nuggets, these statistics all still hold true for Iverson. He’s assisted far less often but is still a better jump shooter and close shooter than his teammates.
7) Note this is not true for Carmelo who, despite being assisted as often, is a much worse jump shooter than his teammates.
8) It is also not true for Camby, who despite being assisted on almost all his baskets, shoots dramatically worse than his teammates in both categories.
9) Since Iverson has left, the Sixers have shot much worse on jump shots and gotten fewer assisted jump shots, but have significantly increased their close range shooting and assisted close range baskets. Accumulated though, the Sixers have shot the same since AI left with the exact same percentage of assisted buckets.
So what does this all add up to? Three things for sure:
1) Iverson is a better shooter than his general eFG% would indicate, because its distorted by his inability to get dunks or tip ins. This is not to say that isn’t a problem in itself or that he is a good shooter. But it is misleading for that problem to imply, through overall eFG%, that Iverson is a worse shooter than he is. Note that Deron Williams also has this problem. Ronnie Brewer has that exact opposite since his excess of dunks masks his exceptionally poor shooting (that also highly assisted). Steve Nash is the ideal scenario, in which he gets no dunks or tips, but shoots so superbly that his overall eFG% is far better than his team’s.
2) Iverson creates an unusual amount of his own shots. There are very few players who get as few unassisted buckets as he does. It’s only guys like Nash (who’s the extreme scenario), Paul, Williams, Lebron, Kobe, and Parker. And for the most part like all of them, he shoots better than his teammates (the exceptions here being Parker and Lebron on jump shots).
3) The Sixers probably relied heavily on Iverson’s created offense. He assisted other players and took many unassisted shots. So they were out of whack when he wasn’t there. Thus the lower percentage. After he left, however, they were able to adjust by getting many more assisted close shots. They didn’t improve. But they maintained. Their perimeter shooting was less assisted and much worse, however. The Nuggets on the other hand have had significantly improved perimeter shooting since Iverson’s arrival (though it’s not any more or less assisted)
So, hesitant conclusion:
Iverson has a very positive impact on his team’s perimeter shooting, by both creating his own shot and, to some extent, creating for others. It is not as positive as Steve Nash’s or Chris Paul’s, but it’s more so than role players like Kyle Korver, who’s absence from Philly did not effect their jump shooting. It was heavily assisted though, so the thought is, he was leaving easy shots.
Iverson ultimately is a limited offensive weapon though because he can’t get any very high percentage shots (dunks and tip ins).
There’s a lot more that could be said about this, but I’m too exhausted here. It does seem that things are much more complicated than I thought.
Harold Almonte
May 6, 2008
I think the learning from all of this is that you can use position adjusts in order to compare player’s rebounding, since rebounds grabbed and rebounds opportunities are closed correlated with positions, with occasional variations and almost no diversity, then if you want to compare rebounding, you can do it by positions. But this variation and diversity at scoring, even inside one position, is so wide that you could end sometimes comparing between oranges and apples into the same position.
Although “firsts” scoring options tends to be in the perimeter, this is not a written rule, and the only position correlation is the assisted factor between PGs and the rest of positions. Then, comparing scorers is a more complex procedure, that involves: position adjust for assisted factors; a definition of scoring “option”, that be by (or a combination of) number of attempts, (usage level, assisted shots, double teams faced, fouls drawed and other kind of risks taken like distance, clutch, etc); and different or adjusted efficiency averages, or break even rates, for each “option” defined, if that is the case.
Harold Almonte
May 6, 2008
Yes, the value of a scored shot (and a failed shot on average) in term of accumulated points for the team is the same no matter who scored, and probably the clock time wasted for that relatively doesn’t matter too much either. But, the marginal cost for the team, in terms of ballhandling skills, passes, and posibility of turnovers and defense reset, in order to get a high efficiency- scoring attempt from any given defense, is not the same for all teams, nor scorers.
What it needs to be demonstrated with numbers is first: that the marginal cost for low usage-limited variation in types of attempts (skills)-high efficiency scorers, to produce extra scoring at any kind of attempt the team could get for them, is still higher in terms of efficiency than the high usage scorer at the same types of attempts; and second: what should be the final cost for a team to only look for and strongly rely in a limited kind of efficient attempts.