Last August, Henry Abbott – of TrueHoop – made the following prediction:
Henry had a host of reasons behind this prediction. A sample of reasons included Shaq’s reputation as a poor team defender, his age, and the number of minutes Darko and Shaq would see.
Of these, the latter seemed most important. It was fairly clear that Milicic would get more minutes than Shaq this season. And that led Henry to note the following:
So, here’s the deal: I’m not saying Milicic will score more, play better D, have a better PER or anything else. I’m saying that over the course of this season, smart analysis will show he’ll produce more at both ends of the floor, in total. In other words, heck yes we’ll factor in defense, and heck yes, I’ll take the advantage I’ll get from the reality that he’s likely to play more minutes.
And that may not be fair in judging the better player, but it’s fair when we’re talking about the value of signing this or that player — players who can get on the floor are more valuable than players who can’t.
If the two have similar production, at both ends of the floor, we’ll appoint some kind of commission of stat geeks to poke into all the best available metrics — PER, SCHOENE, plus/minus, WinScore, or whatever they want to use — to break the tie.
Looking back on this, I am not really sure how this prediction was supposed to be evaluated. But now that the season is over, we can compare what the two players offered on the court.
Just as Henry imagined, Darko did see more playing time this year. Milicic played 1,686 minutes in 2010-11 while Shaq was only on the floor for 752 minutes. But although Darko got more playing time, it is hard to look at the box score numbers – noted in the table below — and conclude that Milicic helped his team.
The per 48 minute numbers reveal that Milicic was below average with respect to shooting efficiency, free throw attempts, scoring, rebounds, turnovers, and personal fouls. He was above average with respect to taking shots from the field, steals, assists, and blocked shots. But when we put it all together, Milicic had a very poor season.
To see how poor, consider Milicic’s career entering this season. Here are his marks with respect to Wins Produced and WP48 [Wins Produced per 48 minutes].
2003-04: -0.6, -0.171
2004-05: -1.1, -0.211
2005-06: 0.8, 0.049
2006-07: 2.4, 0.061
2007-08:-1.6, -0.045
2008-09: 1.1, 0.052
2009-10: -0.2, -0.016
Career (prior to 2010-11): 0.8, 0.006
As one can see, Milicic’s worst season of his career – in terms of Wins Produced – was 2010-11. So even though he played more minutes than Shaq, his production of -3.4 wins was easily eclipsed by Shaq’s production of 1.6 wins.
Does that mean Henry lost the bet? Again, I am not sure how this prediction was going to be evaluated. And if you read over what Henry said, he never mentions Wins Produced. He does mention Win Score, though, and if we look at that measure we see that Milicic – with a mark of 241.0 – did more than Shaq. So by that measure, Henry wins the bet.
Of course, that is only because Milicic played more minutes than Shaq. If we focus on production relative to position played (as we do when we look at Wins Produced), Milicic was awful. So I don’t think Henry’s prediction came true.
I would note that someday, though, I think it has to be true. In Kareem Abdul-Jabbar’s last season he posted a WP48 of -0.056. Shaq is old and declining, and if he keeps playing, he would eventually offer as little as Kareem did in his final season. So I think someday, Milicic can offer more than Shaq. Someday, though, was not 2010-11.
- DJ
P.S. One last note… having Kevin Love as a teammate didn’t help Milicic. Love led the league in Wins Produced this season. And the existence of diminishing returns indicates that having Love as a teammate diminishes a player’s production. But the diminishing returns effect –as I have often noted – is not that large; and given how far Milicic is in the negative range, I do not think diminishing returns explains why Abbott’s prediction didn’t appear to come true.
It looks like Shaq is also ahead on Win Shares, PER (and however that gets translated into Hollinger’s value added/wins added), and both 1 and 2 year adjusted plus/minus. So I think you’re on pretty solid ground for claiming a Shaq victory.
“Of course, that is only because Milicic played more minutes than Shaq….So I don’t think Henry’s prediction came true.”
Which is exactly what Abbott said “heck yes, I’ll take the advantage I’ll get from the reality that he’s likely to play more minutes”. He goes out of his way at several points to say that he’s not talking about per-minute productivity or who is a better player when they’re on the floor.
Given what Shaq has been able to offer this year, I think he’s pretty clearly right. Would the Cs be better off now with Darko on the roster than Shaq- if Shaq can’t give them anything because he’s hurt? Almost certainly.
^Not if Darko brings negative productivity
And milicic would be better off hurt Karon. That is how bad he is. He is worse then the average Center in the league and still worse then the average replacement or backup center. Thats where the – wins comes from.
So I would rather take someone that didnt play because I can replace him with the standard backup then taking darko. And shaq in his limited minutes actually contributed. Where as when DArko comes on the floor the team gets way way worse. Because he is a walking missed shot.
@Kamron:
I disagree completely. The fact is that there are readily players available in the D-League who can produce at least as much as Darko. So I think 752 minutes of Shaq, and 934 minutes of Semih Erden, or even Nikola Pekovic would probably be more effective than 1,686 minutes of Darko.
I would take Shaq for 752 minutes every season over Darko for 1,686. To me realistically, the worst case scenario for Shaq, is that he’s backed up by Darko, and that those remaining 934 go to a Darko level player. But that still leaves you with 752 less minutes of Darko than before.
@Kamron: I disagree. They’d be better off playing Nenad at 5, at least he’s producing in the positive range (WP48=0.039), heck they’d be better off playing Troy Murphy (WP48=0.016). Darko not only wouldn’t have helped them, he’d have taken wins away from them. He’d be taking shots he can’t make, playing minutes where he’ll be rebounding below average and turning the ball over–essentially getting in the way.
WP48 numbers are from nerdnumbers.com.
Kamron,
I don’t think that’s quite true. Henry fudged a little to make it difficult to evaluate his claim. But he clearly said this:
There’s no way around the fact that Shaq in limited minutes made a net positive contribution while Darko provided a net negative contribution.
Kamron,
Please explain how having a player with a NEGATIVE win production makes you better off by virtue of playing more minutes? Clearly, the celtics would rather have zero minutes from Shaq than get 1000+ minutes of negative production from Darko.
Saying more minutes for Darko somehow makes him more useful is like when a businessmen sells items at a marginal per -unit loss and tries to “make it up in volume.”
I think the consensus is that everyone disagrees with Kamron.
Just a comment, the new moderation before post rule, seems to lead to multiple people posting the same comment, when previously it was the case that one or maybe two posters would respond, and then the vast majority would stay silent since the major points were put across already. Now it seems you’re going to see multiples of the same message, due to the fact that everyone submits their posts, without seeing what’s already available on the queue.
Nick,
I think this is the first time this has happened.
Pingback: Revisiting Darko Milicic vs. Shaquille O’Neal » Stathead » Blog Archive
wow, I am sooo popular.
I disagree completely. The fact is that there are readily players available in the D-League who can produce at least as much as Darko.
I doubt that; Darko shouldn’t be starting for any self-respecting NBA team, but I don’t think he’s obviously worse than a good D-leaguer. Of course, everyone’s entitled to their opinion.
They’d be better off playing Nenad at 5, at least he’s producing in the positive range (WP48=0.039), heck they’d be better off playing Troy Murphy (WP48=0.016).
I didn’t say they’d be better of *starting* Darko. Worst-case scenario is that Bad Darko is exactly equal to an injured Shaq (assuming the coach agrees with your assessment and never plays him). And even in that worst-case (where Darko is worse than many D-leaguers), Bad Darko is able to step in after Nenad is injured and Murphy fouls out- something that an injured Shaq cannot do for you.
So I would rather take someone that didnt play because I can replace him with the standard backup then taking darko….. Because he is a walking missed shot.
In fantasy basketball, yes you can do that. In the NBA, you can’t grab “one standard backup” off of the shelf so easily. You’re going to get a player that no one else wanted for one reason or another. See the D-league comment.
I mean, look at the Lakers- would they take Darko for free? Not “would they start him over Bynum”, but would they take him in place of the last big on their bench, Theo Ratliff (who I didnt realize was even active anymore)? I think they would. He has big flaws, but is he better than Ratliff?
Would he break into their rotation? Probably not. But if Bynum’s hurt and Gasol gets in foul trouble, maybe he gets some minutes off of the bench. Minutes that would otherwise go to some even-less-capable player, not to “one NBA standard backup guy” or a 0.100 wins-per-game producer.
I think there’s a misconception even about negative wins eg And milicic would be better off hurt Karon. That is how bad he is or <i.Please explain how having a player with a NEGATIVE win production makes you better off by virtue of playing more minutes? A player who produces negative wins is not worse than fielding four players, or worse than those four guys plus somebody’s grandma. Even taking Wins as gospel, if you’re going to talk about rosters you need to be able to articulate (as Nick does) the alternatives you’re proposing. If you’re starting Milicic because *Im* the backup center, then he’s adding a lot of value by keeping me off of the court ;). That doesn’t mean that negative wins aren’t useful for evaluation, but just claiming that he’s producing negative wins so his team would be better of if he was hurt- this is just a failure to understand the game. The difference between positive and negative wins is arbitrarily established by the quality of play in the NBA, it’s not some absolute measure that means Milicic detracts from his team’s play regardless of the other options.
Nick does get this: So I think 752 minutes of Shaq, and 934 minutes of Semih Erden, or even Nikola Pekovic would probably be more effective than 1,686 minutes of Darko.
Abbott didn’t put it in terms of subs like this, so it’s not exactly the question being asked. But it’s an interesting question. It’s a bit deeper than that though:
Let’s say, one team gets 20mpg of Darko (backed up by Erden)
The other plays half of their games with 20mpg of Shaq (backed up by Erden) and the other half with 20mpg of Erden (backed up by the next lower guy in the rotation). That makes the Shaq option a little less attractive I think- when he’s not around Erden is your starter and god-knows-who is his backup. That is, taking Shaq impacts your depth, and depth matters. If you could allocate the minutes as you choose, take 10mpg from Shaq and 10mpg from Erden (and maybe 20mpg from Shaq if Erden was hurt) that’d be a different proposition from the real-world one.
But arguing after the fact that Shaq has better per-minute numbers etc is missing the point- if I said I thought Melo would score more than Chris Paul this year, Id have been right. Doesnt matter that Paul does a lot more for his team etc. Abbott said he’d produce more, in total, and specifically disavowed the idea that he was claiming better per-minute production or advanced stat production. Maybe that’s a stupid question, since a guy who misses a lot of shots etc isn’t that useful. But there is certainly something to be said for the other side of the equation: a guy who isn’t available to play is worthless. A guy who can give you 82 games might not have the per-minute productivity of an oft-injured star, but not having to start your subs and use your 12th man as a key reserve is pretty important too.
[Take it to it's logical conclusion and you'll think that 1 game out of 82 from Lebron is as valuable to a franchise than 82 games from Kirk Heinrich or Jeff Green.]
Does the opposite of diminishing returns exist? Like when a good player is surrounded by possibly the worst supporting cast of all time then his numbers are inflated in some way?
This is an extreme example but I don’t think Kevin Love ever touches a 30 rebound game with someone like Dwight Howard or his team, or even Kris Humphries. His stats would diminish but he wouldn’t necessarily be a worse player.
Kamron brings up the salient point that Abbot’s wager discounted minutes played. Darko has a Win Score of 241 for the season, compared to 186.5 for Shaq. So by this measure, Darko comes out on top. I suspect he would for every other box score measurement as well; they all place less emphasis on scoring efficiency.
Shaq: 341 + 178 + 16 + (.5 x 26) + (.5 x 42) – 201 – 56 – (.5 x 131) – (.5 x 120) = 186.5
Darko: 606 + 360 + 54 + (.5 x 104) + (.5 x 140) – 578 – 151 – (.5 x 115) – (.5 x 229) = 241
Of course, it’s rather silly to discount minutes played when evaluating a player. But it’s also silly to discount turnovers and scing efficiency, and this happens regularly among the mainstream media, fans, and even GMs.
Robbie O’Malley,
It’s just a matter of perspective. You can say that Howard’s rebounding is deflated or Love’s is inflated; both are correct. I agree that he’d be even more unlikely to put up such a remarkable single-game rebounding total.
But it’s not like Kobe putting up 81 points or Scott Skiles dishing out 30 assists, where both players absolutely use a possession and there just aren’t that many to go around. They both used about 3/4 of their team’s possessions to put up those stats. There’s about 80+ rebounds available in the average NBA game; Love didn’t even grab half of the boards available.
As for whether a player is “worse” or “better” – there’s arguably an element of subjectivity to it. Is Lebron “worse” now that he’s playing for the Heat? I don’t know. But he’s certainly been less productive.
All I know is that I’d rather have Bargnani than either of them. :-)
I still believe it would be possible to build a decent team (albeit unconventional) by using Bargnani as a SF on offense combined with a hybrid PF and SF that do most of their scoring in the paint by slashing and posting up. They would also have to be above average rebounders and defenders. I realize it makes no sense to build a team around a mediocre hybrid player like Bargnani, but it could be done.
I can’t imagine creating a decent team with a guy that’s always hurt and another that can’t play a lick.
On another note….
I made a few bets on the NBA championship several months ago. One of the teams I bet on was the Bulls at 30-1. You would think I’d be enjoying how well they are playing these days, but I’m not. I like Derrick Rose. He seems like a great kid, has a great attitude about the game, and is improving rapidly as a player. But there’s only so much media hype a person can take. I keep waiting for these TV media guys to just drop their pants and start chasing around the court. Totally ridiculous. I have to turn the volume off on my television.
@Kamron: I agree with your basic idea that it’s better to have someone than not have someone. If it the Celts only have Shaq and another center, perhaps, but that’s not the case. And never was the case. And while Glen Davis maybe a poor sub for Shaq, he’s still better than Darko. I daresay that Garnett playing the 5 is better than Darko (though I have no proof of this). Meaning, unless a fluke incident suddenly hits the Celts and I’d be better off having someone playing than not because of roster minimum requirements, I don’t see why having a healthy Darko would be better than having an injured Shaq.
Essentially, I’d rather have a few games from Shaq (net positive contributions) + the contribution of whoever has to replace him (Nenad, Troy, Jermaine, Davis, KG or even Jeff Green), than having a net negative from a healthy Darko. My point is, he’s so unproductive that he’ll hurt you just by being on the court. You’re better off not having both him and Shaq than having him because while you may not find a league average substitute, it’s easier to win with a substitute even if he turns out to be less than average than to win with Darko.
Watching Derrick Rose play last night was amazing.
Honestly, he does things I don’t see anyone else does on the court. But he takes some ridiculous shots. The whole season’s discussion about Rose being the probably MVP, and not really deserving it has been discussed over and over. And watching him last night, you can SEE why he appears to be MVP caliber. But at the same time, he took a couple of terrible shots, including throwing up 2 air balls on step back long range 2′s. And yet still, he had a terrific game.
He had 5 TO’s in the 3rd quarter alone, and still had a great game.
He’s a very good basketball player, who makes a lot of mistakes. But is still a .200 WP/48 caliber player now. If he learns to hit the 3 in the high 30′s (by no means easy, or a given), he could end up being a .300 player.
The same discussion goes for Carmelo, he’s a very talented player, and even by WP, an above average player who plays a lot of minutes. It seems Carmelo will never get it fully, and I know a lot of people have compared Rose, Carmelo, and AI…but AI was below average WP most seasons, ended up being a basically average NBA starter, Carmelo is slightly above average, and Rose is already basically a WP star….
Having said that…..Watching Rose is like watching Kobe sometimes, and it frustrates me to no end to watch them take step back fall away 2 point jumpers, 2 feet inside the line, without really moving the ball around on offence.
Basketball is like any good movie, the most interesting characters are flawed, and that’s what we love about them. As much as I hate Kobe’s game, and I enjoy talking about his flaws. I enjoy comparing him to the guys I play against at the Y. It’s why I post comments on a basketball blog about player efficiency.
I love the playoffs.
Would any of this analysis differ if we used a median replacement value, rather than the mean? Or, if we went a step further and looked at actual alternatives available to a team?
We talk so much about replacement value, but it seems the center position in the NBA is so sapped that I can’t think of any decent centers for my Sixers to go after, so why should I criticize Minnesota for missing them?
As far as I can tell, most of the above average centers either play 20 mpg or are injured a third of the year.
I daresay that Garnett playing the 5 is better than Darko (though I have no proof of this).
or even Jeff Green
That might be true for KG, but playing smallball changes their entire game. Might work well for them, might be a real problem. I note that they havent done that so far this season that Ive noticed. So Garnett might outproduce Darko, but playing eg Pierce at the 4 might hamper him and be an overall loss compared to a Pierce-Garnett-Darko frontcourt. Or the defensive & rebounding problems created by going small might outweigh any advantage in offensive production.
There’s a reason that smallball is rarely embraced as a 48-minute strategy- it has serious weaknesses and it’s hard for gains to offset those weaknesses.
(And Green at the 5? I think he’d be such a bad fit there that he’d be worse than Darko. No shotblocking, worse rebounding, no big body to work in the post on either end.)
My point is, he’s so unproductive that he’ll hurt you just by being on the court.
See, I think people are being seduced by the negative number into saying “he just hurts you on the court”. That is not the case- remember that the zero-point here is not some threshold between “player who helps on the court” and “player who hurts on the court”- it’s arbitrarily set based on the quality of play in the NBA.
Negative Wins are I think like negative temperatures- colder than positive temperatures, but shouldn’t lead to the confusion “it’s actually negative warmth out there, it’s beyond a total absence of heat”.
Darko, like all pro basketball players, adds a lot on the court compared to some reasonable absolute zero measurement (ie no one, or a small non-athletic non-basketball player). And, as I said before, I think absolute measurements are a mistake anyway- when you pull one player, you’re playing a different one. In the Wolves case, they’d be playing Pekovic. Did minny err by not playing Pekovic 40mpg? Or are both of them worse than a good D-leaguer? Along with a bunch of other bigs sitting on the ends of NBA benches- again, I suspect most teams would take Darko for free even if it just meant an upgrade to the last guy on their bench. (And yes, I am arguing that Darko is an upgrade on the average last guy- if you think otherwise, find some last-big-on-the-bench guys you’d rather have).
Maybe people just enjoy hyperbole, or dislike Darko for the mismatch between his potential and his achievements (and ego). He’s not a good NBA player. He was probably Minny’s best option at the 5 only because their front office is a bunch of idiots. He should maybe be a rotation player on a bad team or a little-used reserve on a contender. He is not worse than nothing, or obviously unqualified to be on an NBA roster.
@ Italian Stallion.
Bargs as a SF wouldn’t work. They tried that before with JO, Bosh and Him. He was (and still would be) absolutely abused on the defensive end.
Unless you meant he was an SF offensively and guarded the teams weakest big-man on the defensive end…oh wait that’ kinda how he plays now.
What exactly does negative production of wins mean?
Yes, Bargnani is a defensive 5 who is closer to a 4 (3.5) on offense. And Tolliver and Love are better players than Darko at the 5 for Minnesota.
I think that misses the point.
Cumulative stats will always help players that play more. Total rebounds will favour a player that plays 82 games and average 1.0 per game over a player averaging 10 playing 8.
I doubt that, even as a cumulative score, Darko is producing more. Would anyone want a center that played 15 minutes a game to average that 1.o rebounds? Would 82 games of such a player be a net win over ten games of 30 minutes and 10 boards?
IMHO, WP is a lovely measure, as it accounts for exactly this fact, and accounts for hideously unproductive players very well.
I think that misses the point.
Cumulative stats will always help players that play more. Total rebounds will favour a player that plays 82 games and average 1.0 per game over a player averaging 10 playing 8.
On the contrary, that is exactly the point. We can debate about whether Abbott’s metric is well-chosen, but we cannot substitute our metric for his and then say he was wrong. He made a prediction based on cumulative production. He appears to me to have been correct. He specifically said that part of his judgment was based on the fact that Darko would probably play more.
Debates about the utility factor of injury etc are interesting, but they don’t alter this basic fact: Abbott specifically disavowed per-minute productivity or that he was trying to determine who was a better player. He said what he thought was going to happen. It happened.
I doubt that, even as a cumulative score, Darko is producing more.
There’s no reason to speculate, since cumulative stats aren’t derived.
Darko produced
606 points, 360 reb, 104 ast, 54 stl, 140 blk, 151 TOs, 229 PFs
Shaq produced
341 points, 178 reb, 26 ast, 16 stl, 42 blk, 56 TO, 120 PFs
That looks pretty definitive to me.
Would anyone want a center that played 15 minutes a game to average that 1.o rebounds? Would 82 games of such a player be a net win over ten games of 30 minutes and 10 boards?
Two things:
1)I don’t know what anyone wants, but I do know what Abbott predicted. If one guys averages 2rpg for 80 games and the other guy averages 10rpg for 10 games, then the first guy grabbed more rebounds. If I predicted that he would grab more total rebounds, then I am right. I am not ‘actually wrong’ because the second guy is a better rebounder.
2)If you’re read the thread, you’ll already know what I think- whether you want 20 games from a star or 80 games from a marginal starter depends very much on the rest of your rotation- specifically, who gets the minutes in the 62 games that your star misses. If that player is really bad, then 80 games from the marginal stater is probably a good deal. If you have a lot of depth then the star might be a better deal.
IMHO, WP is a lovely measure, as it accounts for exactly this fact…
I know of no system, or even competent observer of the sport, who would think that a starting C who averaged 15mpg for some reason and only got 1rpg was anything but a disaster. If this is the wisdom of Wins, then it is not telling us very much…
But I told myself I wasnt going to debate the underlying statistics when I started this conversation. And it’s not necessary. Because:
1)Abbott is still right, regardless of whether you think his judging criteria are foolish
2)You cannot use Win scores- or any other metric, even a perfect metric- to judge the wisdom of playing a particular player without knowing what the rest of the that team’s roster is. You cannot use Win scores to decide the value between a a better player with an injury history and a worse player without one without knowing what the rest of that team’s roster is.
3)A negative Win score is arbitrary, and does not mean that the team would be better off if that player was injured or that playing him is always a bad decision, or that playing him is worse than playing 4-on-5.
Kamron,
Your argument around a negative Win Produced player being arbitrary is a bit off. First off it is not legal to play 4 on 5 so let’s not even leave that off the table. Second, there is a collection of D-League players and other players that can be signed to a 10 day contract, vet minimum or signed under other conditions. Your assumption is that Darko as the 2nd worst big man in the league is a better option than a random big man from the D-League. I doubt that. The truth is Shaq in his time on Boston was helpful. Darko was the opposite. In fact Darko is the very sort of example you keep falling back on. Darko is what happens when a team decides to do something rather than play an level NBA player.
Your argument around a negative Win Produced player being arbitrary is a bit off. First off it is not legal to play 4 on 5 so let’s not even leave that off the table. Second, there is a collection of D-League players and other players that can be signed to a 10 day contract, vet minimum or signed under other conditions.
Neither one of those are about my position on negative Wins; my point about negative Wins is that a player who contributes zero wins in the NBA is actually an unusually gifted basketball player when considered against the overall population. It isn’t that he isn’t contributing to his team, it’s that his contributions are less than the average NBA player. Likewise, a negative Wins player isn’t- as someone said- so bad that his team would be better off if he was hurt.
So a zero-win player isnt like a player who isnt there or a player who contributes nothing- he’s a player who’s contribution is lower than average, period. There’s nothing magical about the zero-win threshold (unlike, say, the Kelvin scale of temperature, where zero has a special meaning related to temperature).
Your assumption is that Darko as the 2nd worst big man in the league is a better option than a random big man from the D-League. I doubt that.
My assumption is that Darko is not the 2nd worst big man in the league, and that anyone who actually thinks that doesn’t understand basketball very well. Look at the end of the benches around the league if you want to find the 2nd worst big man (hint: he won’t see a lot of floor time). At least you could try to argue that Darko is the 2nd worst big who gets minutes, but even there Id disagree.
You think Darko is worse than ‘a random big man from the D-League’. I think that says that we don’t have much common ground to talk.
“Unless you meant he was an SF offensively and guarded the teams weakest big-man on the defensive end…oh wait that’ kinda how he plays now.”
That’s what I meant.
It sure can’t work when the rest of the team is horrible and anyone that’s even mildly productive is always hurt. I think it would work better than anything with Milicic on the court and Shaq in a suit though. :-)
I’m confused. Abbott made a bet:
http://espn.go.com/blog/truehoop/post/_/id/18744/shaquille-oneal-vs-darko-milicic-a-bet
So, what does that mean? How do we assess “help more”? The argument has evolved to a debate of what helps more – small minutes for an injured player or longer minutes for an inept player? That was the crux of Abbott’s question, and I am not sure we are closer to answering it.
How do we answer what “helps more”?
Kamron is of the view that “You cannot use Win scores- or any other metric, even a perfect metric- to judge the wisdom of playing a particular player without knowing what the rest of the that team’s roster is.”
That seems to me to be an argument that we can’t measure it at all, so I’ll disregard that.
The idea of WP is also rejected, which seems strange to me, so what are we left with, realistically, only cumulative stats.
In how many minutes? That matters, because Darko played 1,686 minutes and Shaq played 751. What if Darko had played 3000? Or what if those stats were over a 5000 minute career? Who helped more in those cases?
Surely, there comes a point at which “smart analysis”, to quote Abbott’s phrase, says that player A is so bad they hurt, and less minutes from them and Player B + player C is better. Darko for 1600 or Shaq for 750 and Player Y for 850 is the question.
IMHO, cumulative stats do not work because, as the Reductio ad absurdum I offered demonstrates, there is a point at which a player so under produces that no matter the cumulative stats, they are worse than less of another player.
So we are left with the same questions, and have no way to decide between 82 shit games played (Darko) and 36 OK games played (Shaq), and no framework to compare that at all.
I’m glad I sorted that out in my own head :)
Kamron is of the view that “You cannot use Win scores- or any other metric, even a perfect metric- to judge the wisdom of playing a particular player without knowing what the rest of the that team’s roster is.”
That seems to me to be an argument that we can’t measure it at all, so I’ll disregard that.
I said that outside of the context of Abbott’s bet, considering the larger question of “who was more valuable to their team, Darko or Shaq?”, which is IMO more interesting than Abbott’s bet. But that was my point #2.
Point #1 was “Abbott is still right, regardless of whether you think his judging criteria are foolish”. I should probably add “IMO” to that.
Besides, I just said we couldn’t just look at the player, we had to look at the roster- not that it was unknowable. In the Wolves case, behind Darko we’ve got a rookie Euro (Pekovic) who fouls even more per minute than Darko, and then they have to start playing Love, Beasley, Anthony Randolph, or Anthony Tolliver at the C. That’s a collection of really bad C options I think. That’s the kind of roster where a marginal starter is probably better than an oft-injured good starter; with Shaq, they’ve got 50 games of starting Pekovic and playing a lot of smallball when he takes a break (or fouls out after 2 quarters)- which as Ive said, could hurt the entire team at both ends of the floor.
For the Cs- well, when he made the bet they were starting Perkins and playing Shaq off of the bench, with additional backup of Davis and J ONeil (and Ergadan on the end of of the bench). With Shaq out they’ve still got plenty of C depth. Subtracting Shaq and adding Darko seems like a bad choice there- they’ve got a lot of options so trading quality for depth doesn’t make sense.
So if I had to say it, Id say the general sense of Abbott’s comment was wrong: Shaq is more valuable to the Celts than Darko, and Darko is more valuable to the Wolves than Shaq.
IMHO, cumulative stats do not work because, as the Reductio ad absurdum I offered demonstrates, there is a point at which a player so under produces that no matter the cumulative stats, they are worse than less of another player.
Sure, cumulative stats are bad for understanding a player’s contribution, I think we agree. But I still think they’re the most reasonable interpretation of Abbott’s proposition. He pretty clearly does not mean something like Wins or PER, even if those are much better at determining value.
All of this does get at an interesting thing though, for statistical observations of basketball- the value of a substitute is important in understanding injury risk. So a player who gives you 12 wins over 82 games might be more valuable than a player who gives you 15 wins over 41 games, if your rotation is so think that you’re going to be playing negative-Wins players when the 41-game player is out (or if it messes with your rotation in other ways and reduces the effectiveness of other players).
Yes, Darko produced more points, rebounds, and assists than Shaq, but he also “produced” more turnovers, missed shots, and, presumably, by virtue of having played many more minutes, bad defense. So to compare these players we’d need a metric that nets out the good and bad and gives us a total net contribution by a player. WinScore is one such metric, but it doesn’t fully reflect defensive contributions. But at least I could imagine that Henry had in mind something like an adjusted WinScore.
Henry conceded last night:
http://espn.go.com/blog/truehoop/post/_/id/28124/shaq-vs-darko-losing-ugly