Batter's Box Interactive Magazine Batter's Box Interactive Magazine Batter's Box Interactive Magazine
I have been pouring too much time into basketball research recently, but there have been some side benefits. I have developed two metrics which I will explain briefly, that measure a player's defensive contributions. These are not perfectly original, however I believe that no such metrics have been developed before.

Incidentally, look for me in an upcoming Aaron's Baseball Blog discussing more basketball-related stuff.


Individual Defensive Value

The first is called Individual Defensive Value (IDV). IDV is simply the total point value of the defensive events that a player is responsible for... his blocked shots, steals, defensive rebounds, and his personal fouls.

How do you determine the point value of a defensive event? For these formulae, I am indebted to John Hollinger and his terrific book Pro Basketball Prospectus. A less modest man than John would have called his book The Hidden Game of Basketball, because his work has had at least that much impact in terms of revolutionizing the study of basketball statistics.

To illustrate the Hollinger approach, let's take the point value of a blocked shot.

Blocked shots have value where they are rebounded by the defensive team, and where they are, they generally have the value of one possession (i.e. the point value of one possession). This is because the opponent has spent a possession, and received no points. So a player's blocks are valued as

Blocked Shots * VOP * league DRB%

Where "VOP" is the average value of one possession (almost exactly one point to date in 2003/04... 1.0012 points to be exact) and "league DRB%" is the percentage of rebounds that are grabbed by the defensive team - 71.3% in 2003/04 to date.

Point values for other events are similar. The value of a steal, is one possession (i.e., simply VOP). The value of a defensive rebound, is the VOP times (1 - league DRB%)... i.e. the chance that the offense would have grabbed that board.

The value of a personal foul, is more complex, but is essentially the value of the free throws which are generated by the foul. You take the average number of free throws generated by each personal foul, and multiply that by the value of those free throws (minus the "possession value" of the free throws), to get the negative value of a foul.

Add it all up, and you get the contributions of a defensive player, measured in points. You then need to adjust each player for the pace his team plays at. Once you've done that, you have IDV.

Adjusted Defensive Value

The second stat is called Adjusted Defensive Value (ADV) and it is, as the title implies, IDV along with an adjustment. The adjustment in this case has to do with team field-goal defense; it gives players who play on teams with good FG% defense a boost. Essentially, a player gets credit for his percentage of the field goal misses by opponents, keyed to his time on the floor.

The value of a field goal miss generally (to the defensive team) is the same as that of a blocked shot. We figure the number of misses a team forced, subtract the number of misses an average team would have forced, assign the blocked shot value for each miss (a forced miss works the same as a block), then divide up the team's points among all its players based on minutes played. That amount, when added to a player's IDV, is his ADV. I prefer ADV to IDV because though the adjustment amount is usually very small, it does give players on good defensive teams a boost. The largest ADV adjustments for this year are Tracy McGrady (-12.95) and the wholly undeserving Cuttino Mobley (+18.81). Rockets and Spurs players get the biggest boost (thanks largely to Yao and Tim Duncan) and Magic and Blazers players take the biggest hits.

So who are the NBA's best players in Adjusted Defensive Value in 2003/04?

Top 15 in ADV (All data calculated as of February 7)


Ben Wallace DET 307.9
Kevin Garnett MIN 262.3
Tim Duncan SAN 245.9
Andrei Kirilenko UTA 225.2
Shawn Marion PHO 207.1
Jermaine O'Neal IND 193.1
Marcus Camby DEN 176.2
Donyell Marshall TOR 167.2
Theo Ratliff ATL 159.8
Ron Artest IND 153.7
Kenyon Martin NJN 147.9
Paul Pierce BOS 145.0
Lamar Odom MIA 142.6
Jason Kidd NJN 135.9
Carlos Boozer CLE 135.7
Yao Ming HOU 134.0
Rados Nesterovic SAN 131.5
Erick Dampier GSW 128.3
Dirk Nowitzki DAL 127.6
Brad Miller SAC 126.8


ADV/48 Minutes (Min 400 minutes)


Wallace 7.49
Duncan 6.46
Garnett 6.45
Chris Andersen DEN 6.44
Camby 6.41
Kirilenko 6.35
Dan Gadzuric MIL 5.40
O'Neal 5.20
Shawn Bradley DAL 4.87
Marion 4.86
Martin 4.83
Theo Ratliff PHI 4.82
Marshall 4.81


Well, the 14 guys on these lists are all superb defensive players, really. If you were looking for an All-Defensive team, most of these guys would be on your shortlist, though the list is dominated by shotblocker/rebounders.

The top 5 guards in ADV/48 minutes (min 400 minutes)


Brevin Knight WAS 4.03
Manu Ginobili SAN 3.90
Jason Kidd NJN 3.71
Paul Pierce BOS 3.55
Darrell Armstrong NOR 3.54


Anyway, Shawn Marion plays some guard as well, enough for All-Defensive consideration. (An All-Defensive team might have a token guard, but really the best team defenders in the game are all frontline players).

My All-Defensive Team for 2003/04 so far, then, is Ben Wallace, Kevin Garnett, Andrei Kirilenko, Shawn Marion, and Jason Kidd. Sure, it's hard on Tim Duncan, but ultimately I'd rather have Garnett or Wallace doing my dirty work as the last man back. Call it a personal preference.

My Second Team would be Pierce, Ginobili (with some misgivings), O'Neal, Camby, and Duncan.

The worst twelve players in ADV/48 minutes (min 400 minutes)


Michael Curry TOR -0.08
Danny Fortson DAL 0.10
Steve Smith NOR 0.14
Jarron Collins UTA 0.24
Tyronn Lue ORL 0.40
Othel. Harrington NYK 0.43
Kyle Korver PHI 0.58
Chris Wilcox LAC 0.67
Raja Bell UTA 0.73
Corl. Williamson DET 0.74
Tony Massenburg SAC 0.75
Chris Wilcox LAC 0.77
Raul Lopez UTA 0.81


I don't know if this last list is really terrible defensively, though it is full of forwards who are hopeless rebounders (Fortson excepted) and who foul a lot (Fortson in spades). Some of these guys (Fortson, Wilcox, Bell) are good enough on the offensive end to be good players overall, and most of the rest (Williamson, Lopez, Lue, Smith, Collins, Korver) are at least decent.

Curry, of course, has the unique distinction of being the worst offensive *and* the worst defensive player to receive significant minutes in the NBA this year. (If you'd seen him play twice a week like I have, you'd know why I say "of course".) I think it's safe to say that's never been done before, and we'll see if he makes it. Curry's actually not a bad man-on-man defender; he's pretty good. But he just doesn't contribute anything at all on the defensive end in terms of making plays, and is the league's worst rebounding forward by miles. If Kevin Garnett is the ultimate impact player, then Curry is his exact opposite.
Craig Names His NBA All-Defensive Team For Some Reason | 40 comments | Create New Account
The following comments are owned by whomever posted them. This site is not responsible for what they say.
_Donkit R.K. - Monday, February 09 2004 @ 11:49 AM EST (#79199) #
Best...title....ever
_Donkit R.K. - Monday, February 09 2004 @ 11:57 AM EST (#79200) #
I have been ragging on the Raptors insistance of playing Curry since the start of the season, especially when he started and Lamond Murray was mostly an afterthought. Frankly, he's useless and I know that both anecdotally and now statistically. Fine work, Craig. Except for Dirk Diggler, I thought the first list was excellent. Dirk's height gets him enough rebounds and blocks to get in there, I guess.
Craig B - Monday, February 09 2004 @ 01:12 PM EST (#79201) #
Yeah, Nowitzki was a surprise. He does get a lot of blocks, though, and gets steals as well. Playing 38.5 minutes a game helps him, as the list is additive. And for all his faults as a player, he's killer on the defensive boards. Nowitzki ranks only 22nd in rebounds per game because he usually plays 20 feet from the basket on offense. In *defensive* rebounds, he is 10th in the NBA per game.

He also doesn't commit a lot of fouls despite playing major minutes and playing a lot against opposing centers.
Mike D - Monday, February 09 2004 @ 01:33 PM EST (#79202) #
I think this list gives further credence to Donyell Marshall's high basketball IQ. He just never does anything dumb on the floor, which explains, in part, his "quiet" 23/14 games.

I'd have him replace Camby on the second team, just because of Camby's lack of durability and in-game endurance.
Craig B - Monday, February 09 2004 @ 01:39 PM EST (#79203) #
Possibly, Mike. I agree 100% about Donyell Marshall, who is the best player on the Raptors and gets no credit for it. He's five times the player Jalen Rose is and gets far less hype. Marshall would be even better on a team that had some people who can pass - or even one person who can pass. In Utah, he would get two or three easy buckets a game just shaking his defender in the low blocks, and then getting his hands up and catching a pass just as it arrived from Stockton.

Camby, though, while not durable, is a fearsome defensive player who changes shots on a constant basis... something Marshall just doesn't do. The Nuggets are also a terrific defensive team, something the Raptors aren't quite. So I'd give the nod to Camby.
Mike D - Monday, February 09 2004 @ 01:49 PM EST (#79204) #
Camby, though, while not durable, is a fearsome defensive player who changes shots on a constant basis... something Marshall just doesn't do.

Absolutely agreed on both points. The question is, to draw a parallel to a batting race or an ERA title, does Camby "qualify" based on games played and minutes played?
_S.K. - Monday, February 09 2004 @ 02:17 PM EST (#79205) #
I think Rose is a pretty good passer. It's his only real skill.

On the Raps website I go to people keep advocating including Marshall in a package for "a real center" (usually someone like Adonal Foyle). I keep trying to tell them, without Marshall this team would be SCARY.

Btw Craig, in ADV/48 you have Ratliff listed as being on Philly. Just thought you might want to know.

Oh... by the way.. great stuff! It's good to finally get some context for defensive stats.
_Young - Monday, February 09 2004 @ 02:24 PM EST (#79206) #
Speaking on Camby... I think it is unfair to say that Camby is injured a lot in the current season. He has played 48 of 52 games for his team. Only 3 others on his team has played all 52 games, Boykins, Miller and Anthony. His mpg isn't all that shabby when taken into the team's context either, he is at 28, while the other 4 of the denver starting five is getting 35, 34, 32 and 32 mpg.

So he doesn't play as much as the others, but not drastically a lot less. He hasn't missed a lot of games unlike some other superstar players this NBA season, Shaq and Webber comes to mind. Can he be better? Maybe. Do people in Toronto who saw him play way back when have negative views of him? I know I do, but still...

I think its the same view that people have of Vince Carter. How he is still gets the most votes for the all-star team is all based on past year performances. Have you guys watched games of the Raptors lately? Has Carter dunked more often than he has layed the ball in? I know he is injured, but his star status is based on his dunking ability, and this supposed dunking ability is hampered by his inability to jump... sigh... days bygone
_Young - Monday, February 09 2004 @ 02:27 PM EST (#79207) #
Hollinger also mentioned that he hoped one day someone would start tracking charges taken. Seen anyone on the web who has done this?
Mike D - Monday, February 09 2004 @ 02:43 PM EST (#79208) #
hoped one day someone would start tracking charges taken

If incorporated, it would vault Morris Peterson to the top of the list!
_Scott Carefoot - Monday, February 09 2004 @ 02:56 PM EST (#79209) #
http://www.raptorblog.com
Craig: One of my RaptorBloggers told me about this article. I share your admiration of John Hollinger's work, but I disagree with one aspect of your measurement. I don't believe defensive rebounding should count as a defensive stat. If Robert "All Scottish, No Finnish" Archibald makes somebody miss a shot and Donyell Marshall grabs the rebound, why should Marshall get credit for Archie's defensive stop?

Danny Fortson is a perfect example of a guy who would be overrated by your stat. He's a great rebounder but a terrible defender. And Michael Curry is obviously underrated as a defender. If he was truly as bad as conventional statistics indicate, he wouldn't be in the NBA. Rick Carlisle and Kevin O'Neill have both given him significant roles on their teams. These are not stupid men.
Pistol - Monday, February 09 2004 @ 03:31 PM EST (#79210) #
Does anyone track +/- in basketball?
_Paul D - Monday, February 09 2004 @ 03:45 PM EST (#79211) #
http://www.thewolfshack.com
Speaking of John Hollinger, anyone know what happened between him and SI? Why isn't he writing for htem anymore?
Also, Craig this is great work. I do question the value of defensive rebounds though. The way I've had it explained to me is that a defensive rebound has alot less value then an offensive rebound. Party for what Scott said, and also because if you don't get the defensive rebound, it's likely that someone else on your team will. Unlike an offensive rebound. So it may be overated as a defensive stat. (Those ideas aren't mind though, they're igor's, from asbntr)
Craig B - Monday, February 09 2004 @ 03:57 PM EST (#79212) #
in ADV/48 you have Ratliff listed as being on Philly

Old habits die hard.

I don't believe defensive rebounding should count as a defensive stat. If Robert "All Scottish, No Finnish" Archibald makes somebody miss a shot and Donyell Marshall grabs the rebound, why should Marshall get credit for Archie's defensive stop?

Interesting question. The answer lies not in the stop, but in the rebound. Let me explain.

The stop, unfortunately, we have no way (yet) of tracking, that's even halfway decent. What I do it give a small amount of credit to everyone a team, when that team makes a lot of stops - or vice versa. That's the "adjusted" in ADV. I wish I could do more on this.

But the rebound, that's another story. When the shot is missed, there is still an open potential for scoring to occur off the rebound. In other words, the "stop" doesn't take place until the rebound is secured. So defensive rebounding, in fact, is a huge factor in defense... a guy who can't rebound the ball, like Curry, is hugely less valuable than a guy like Jerome Williams, who plays the same position but grabs lots of misses. A team like last year's Warriors, who can't be collectively bothered to rebound the opponent's misses, are going to be allowing lay-in after lay-in. (Which is exactly what happened... GS had the worst defensive record in the NBA).

When a stop is made (unless it's a Team Rebound) someone is responsible for completing the stop. Now they don't get the whole credit for the stop. Most of that credit (in fact, I can even tell you how much... it's 0.71 of the 1.01 points for the stop) goes uncredited - unless the shot was blocked.

The defensive rebounder gets about 0.30 points (VOP times 1-DRB%) for the rebound. That's because when the rebound is in the air, there is still a live potential for the offense to retrieve the ball, regain its possession, and try again. In order to complete the stop, you gotta clean the glass.
Craig B - Monday, February 09 2004 @ 04:01 PM EST (#79213) #
Defensive rebounds do, absolutely, have less value than offensive rebounds - at least for the player getting them. For the *team*, they have the same value. But that doesn't apply here.

Not for the individual. If Player X didn't get the defensive board, his team was still likely to get it. If Player Y gets the offensive board, it's a huge play because it's not likely to occur.

My IDV/ADV method gives only individual credit for the rebound. So if player T blocks a shot and player U grabs that rebound, T gets credited with 0.71 points and U gets 0.30 points. This is even though without either act, the team would be down 1.01 points.

So Paul, I agree! The method takes your concerns into account.
Craig B - Monday, February 09 2004 @ 04:23 PM EST (#79214) #
Michael Curry is obviously underrated as a defender. If he was truly as bad as conventional statistics indicate, he wouldn't be in the NBA. Rick Carlisle and Kevin O'Neill have both given him significant roles on their teams. These are not stupid men.

Oh really? :)

I kid, I kid. Rick Carlisle is, of course, not a stupid man.

Yeah, the method underrates Curry, I agree 100%. It underrates him because it doesn't manage to capture the things he does do well defensively (contesting shots, possession and position) and does look at all the things he doesn't do well (everything else - and I mean EVERYTHING).

Now the things Curry does well are important, but they're certainly not as important as the other stuff - even on the defensive end. On the defensive end, the stuff that ADV measures is about half of defense - the other half is the stuff Curry does well. He's easily the worst player in the league at all the stuff that's measurable, so there's no way - even if he were some sort of defensive superstar at the other stuff, which he's not - that he's not a bad defensive player. But there's no way he's really the league's worst defender.

Time for my rant...

Despite all this, Curry has huge lapses where he plays with his head down on both ends of the floor (I can't believe that O'Neill doesn't notice *this*, because it drives me batshit crazy watching on TV) and as a result he doesn't make steals that other players would make (because he doesn't see the passing lane) and doesn't ever give anything significant in the way of help defense. He also gets screened pretty easily. Sure, he gives a hell of an effort out there, and he denies his man the ball better than any Raptor since Alvin Robertson. But he never, ever makes plays because half the time he's busy running around after his man instead of playing basketball.

NBA coaches and other "basketball people", just like baseball people, often come to believe that players have special skills or talents that only they can see. I know this; I'm a former basketball coach myself. A guy like Curry gets props for doing things fundamentally right (except the head-down thing) and trying really, really hard. It doesn't cut it.

And that's not getting into his offense, which is putrid. It's worse than his defense. It may be worse than ANYONE's defense. At least he (like Rose) knows how to pass. Of course, he never has the ball so it doesn't matter much (there's no point in getting him the ball either, because you can't pass to a guy who has his head down). And he never even sets screens. Sigh. Rant off...
_Scott - Monday, February 09 2004 @ 04:24 PM EST (#79215) #
http://www.raptorblog.com
Craig, you said: "When the shot is missed, there is still an open potential for scoring to occur off the rebound. In other words, the 'stop' doesn't take place until the rebound is secured."

I disagree, and so does the NBA Rules committee. That's why a foul on a rebound is a loose ball foul instead of an offensive or defensive foul. Once the shot is missed, the stop has been completed and the ball is "loose" or open for anyone.

If a player misses a shot and the ball goes out of bounds without a rebound, it's still a stop. If the offense gets the ball back, the previous possession ended in a stop and now there's a new possession with a new shot clock.

Rebounding is a completely different discipline from defence, but they both rely heavily on hustling and size so certain types of players (Ben Wallace being the prototype) tend to be good at both.

To muddy the waters further, steals and blocks are also not necessarily measurements of great defenders. Some prolific shot-blockers and steals take excessive gambles to boost their stats in these areas and get regularly burned as a result.

When it comes right down to it, the most effective measure of a player's defensive ability is probably how a given player limits his opponents' offensive stats compared to that player's offensive stats against other teams. So a player who averages 50% from the field, 25 points and 6 assists per game might only average 40%, 18 points and 4 assists when guarded by a certain player. That defensive effectiveness might not show up in rebounding, steals or blocks, but it shows up in overall points allowed and ultimately in the win column.
Craig B - Monday, February 09 2004 @ 04:39 PM EST (#79216) #
I disagree, and so does the NBA Rules committee. That's why a foul on a rebound is a loose ball foul instead of an offensive or defensive foul. Once the shot is missed, the stop has been completed and the ball is "loose" or open for anyone

Yeah, I know, but what matters is the possession. At least, that's what matters to me as an analyst. It's a hell of a lot easier to analyze the game if you treat the ball as never out of the hansdof the defense.

Anyway, as a former coach and player I can tell you that *I* always had it drilled into me that you haven't made a stop until you have the rebound.

Rebounding is a completely different discipline from defence, but they both rely heavily on hustling and size so certain types of players (Ben Wallace being the prototype) tend to be good at both.

No. Look, it's not that I don't understand your point (it's a common understanding that shooting, passing, rebounding, and "defense" are all separate skills) but from a functional point of view, that's not correct. There is only offense and defense... putting points on the scoreboard and keeping the opponents off.

steals and blocks are also not necessarily measurements of great defenders

Yes. But players who make steals and blocks are making plays on defense. No, all of defense isn't being captured by these. Lots of matador players are gamblers. It's the nature of the system... it's not perfect.

I will, however, take the stealer over the non-stealer and the shotblocker over the non-shotblocker sight unseen. All the system can do is evaluate players for making the plays we have a record of.

the most effective measure of a player's defensive ability is probably how a given player limits his opponents' offensive stats compared to that player's offensive stats against other teams.

Yes, though that doesn't take into account the single most important element of defense in the NBA environment - help defense. If you've ever played in or against, or coached or coached against, a man-to-man defense or a matchup zone, you'll know that help - where, when, and how much - is the key to stopping opponents. It's why the Rockets are the best field-goal percentage defense team. Because they play in a fabulous defensive system that provides lots of support for weak defenders, and because they have a center who provides a ton of help just by wiggling his eyebrow.

It's why the most popular set offensive play in the NBA is the clearout - because it reduces the availability of help defense.

I don't know if we'll ever get to a satisfactory analysis of defense. It is fun, though.
Craig B - Monday, February 09 2004 @ 04:39 PM EST (#79217) #
italics begone. Sorry.
_Rob - Monday, February 09 2004 @ 04:55 PM EST (#79218) #
Here is a webpage that lists tons of stats.

http://82games.com/03TOR8D.HTM

For example, Michael Curry limits opponents (the whole team, I think) to 43.1% shooting (*) while they shoot 45.2% when he isn't there - for a net gain of 2.1%.

(* this is effective FG % which counts made 3 pointers as 1.5 makes)

Mind you, he hurts the offense even more at -3.3%!
_NDG - Monday, February 09 2004 @ 06:20 PM EST (#79219) #
Wow Rob, that site you linked to is awesome.

Really shows how underrated Donyell Marshall is and how overrated JYD and AD were.
Pistol - Monday, February 09 2004 @ 07:25 PM EST (#79220) #
What a great site 82games.com is. I've found a new site to surf.
Pistol - Tuesday, February 10 2004 @ 09:27 AM EST (#79221) #
http://www.aarongleeman.com/
Craig – great column over at Aaron’s site today.

You may notice something else about this list. The top scorers are not necessarily on it. Allen Iverson, Kobe Bryant, Vince Carter, LeBron James... these guys are high on the scorer's list, but not as high in PER and similar metrics. Iverson leads the NBA in scoring, yet is 26th in the league in WinsAR at 5.5, tied with Carlos Boozer and Rashard Lewis. Why is this?

The analogy I use is that points scored is an awful lot like RBI in baseball. Both are primarily a product of opportunity. The top scorers, like the top RBI men, are all good players. They are all good at converting opportunities into baskets or runs. But that does not make them the best players; the players who are setting up those opportunities have equal claim to offensive prowess. The players who are more efficient (who take fewer shots and waste fewer of a team's opportunities) have a better claim to offensive prowess. Possession in basketball is like outs in baseball... it is your commodity.


Admittedly, I’m new to this, so perhaps I’m spouting non-sense here.

I’m not certain that points and RBIs are a good comparison to each other. In baseball you can’t, for the most part, decide who comes to bat in certain situations. If you have the bases loaded you can’t just send Delgado up to the plate. You’re stuck with the order you submitted prior to the game (save for pinch hitting). In basketball you can have much more of an influence of who’s taking a shot. A PF who average 6 points a game might be more efficient than a SG who averages 20 points a game, but if you have one possession where you have to score would it be wiser to get the ball to the PF? Likely not as he probably can’t even get a shot off on his own and knows that he can only get the garbage points that his teammates set up for him.

Take Iverson for example. He ranks lower than how ‘traditional’ people would evaluate him. I’m pretty sure that a lot of it is a due to a low shooting percentage. But perhaps there’s a few things to explain that? If you’re one of the few players on your team that can force double teams (by penetration forcing the rotation, or being good enough in the post to force a double) you’re going to have the ball in your hands more and likely get more shots – including shots that you might not necessarily take if not for a shot clock. I think this would be a drain on the shooting percentage that is teammate dependent. Also, in the case of Iverson I frequently see him get by his defender and draw another defender towards him. He’ll miss the shot, but because a 2nd defender had to come over to him to prevent an easy layup it leaves someone open for an easy offensive rebound and put back. So in a case like that Iverson gets a missed shot, and a teammate gets an offensive rebound and basket. But I assume in the WinsAR calculation that Iverson is being hurt in this metric (in this particular example) and his teammate is being helped even though I believe that all of the net benefit is due to Iverson – that is, the teammate getting the rebound and basket is only getting it because Iverson beat his defender off the dribble.

Another area where I think a player can get undervalued is when he’s the first pass of a 2 (or more) pass sequence to get a basket. For example, say Duncan gets the ball in the post and a double team comes. Say he passes the ball out to Ginobili, the defense anticipates that and rotates. Ginobili then makes another pass to, say Parker, who has an easy look at a 3 pointer. If Parker makes the shot he gets credit for a 3 pointer, Ginobili gets an assist, and Duncan gets nothing, even though the entire sequence is caused because Duncan can force a double in the post.

Looking at this from a personal perspective, I play basketball once a week at the gym. If I’m on a team where I’m forced to be the primary offensive option my team is in trouble. My shooting percentage will be low and my turnovers will be high, which would yield a pretty poor performance if you threw it into a formula and spit out a number. On the other hand, if I’m on a team where there’s someone that the defense has to focus on I’m going to be more effective. I can hit open jumpers, make good passes, and get lost on the baseline for easy baskets. In that situation, if measured statistically, I’ll look a lot better than I would in the first situation, just because of who I have around me. I’m a pretty good complimentary player, but if I’m supposed to be more than that I won’t be nearly as effective.

In baseball it’s just a 1 on 1 battle - the hitter versus the pitcher (obviously ignoring defense). My teammates wouldn’t have much of an impact on the result of my AB (typically G6).

Maybe it’s because I’m new to basketball sabrmetrics, but I’m not at the point where I can see a list like WinsAR and think of it the same way I would think of something like RARP in baseball. I just think there’s much more teammate dependence in basketball than there is in baseball.
_DS - Tuesday, February 10 2004 @ 09:43 AM EST (#79222) #
Mike D,

Funny you should mention Peterson's ability to take charges. I went to a Michigan State basketball camp a few years back, and the only defensive drill they ever covered was how to take the charge. No drills on defensive positioning, defensive rebounding, mirroring the ball, nothing. Clearly the ability to take a charge was key for any State player.

Needless to say, it was definitely the worst camp I ever went to.
Craig B - Tuesday, February 10 2004 @ 09:54 AM EST (#79223) #
Yeah, there's more to it than that. I am being pretty hard on Iverson by comparing him (if only by analogy) to Joe Carter.

Taking shots in basketball is very important - not just because of the "choice" element you discussed, which is very important, but because of the inexorable shot clock. You *need* a shooter when you have a 24- or 30- or 35-second clock on you.

The system doesn't, incidentally, punish Iverson for the whole of the field goal miss. The chance that his team will rebound that ball and score is taken into account - though the *increased* chance that Iverson's misses will be rebounded versus another player's isn't. I think, with all due respect, that that's a non-zero factor, but not a very important one.

PER, because it's an additive system, actually means that the 10/23 shooter will score better - usually much better - than the guy who goes 3/6 in the same number of minutes. The same goes for WinsAR. That's as it should be, as you point out.

But what it also does, is sort out who is an efficient 30-point scorer from who isn't. There's just no way that a guy who scores 25 points on a 10/27 performance from the field and 5/6 from the line - an entirely typical Iverson performance - is helping his team as much as 25 points of Shaq, when he goes 11/18 from the field and 3/6 from the line. Iverson might drop ten dimes on you... in which case sure, he's had a real good day. He might have turned the ball over six time, in which case he hasn't. The system takes all that information, and looks at what its impact is. That's where the value lies in this stuff... we have no way of knowing whether Antoine Walker getting 8 rebounds a game in 39 minutes is good, bad, or indifferent, until we evaluate the impact of that on the scoreboard, and compare it to what other players are doing.

Another area where I think a player can get undervalued is when he’s the first pass of a 2 (or more) pass sequence to get a basket.

Yes, this is true. However, in most set offensive systems there isn't one guy who racks up lots of assists on set plays, and other guys who don't. Now in transition, for example, or on the pick and roll, one guy gets the credit for most of the assists (these are usually one-pass plays anyway, so all the credit should go to the setup man). Otherwise, assists are relatively well-distributed. No one player is particularly hurt by awarding all the "setup" credit to the assister only.

Generally, all the WinsAR system can measure is who is making the plays. There's a lot more to basketball than just making plays - though making plays (assists, field goals, drawing and giving fouls, rebounds, blocks, turnovers, steals) is an extremely important thing. WinsAR is not going to capture the other stuff, just as traditional baseball stats can't measure defensive range or productive outs.
robertdudek - Tuesday, February 10 2004 @ 10:10 AM EST (#79224) #
Craig, have you calculated replacement levels based on position?

I had an idea way back when to come up with "position scores" to classify players along a basketball position spectrum similar in concept to Bill James's run element ratio.

For example, stats would be classified as either "centre-like","guard-like" or "neutral". Assists and three-point attempts (perhaps steals) would be an example of the latter and rebounds and blocks the former. You could then assign every player a score on a scale of 1 to 5. Some players listed as guards would have a position score more like small forwards because they get rebounds (I am thinking here of Jason Kidd).

You might be able to distinguish 6 different roles: true centre, centre-forward, power forward, small forward, shooting guard, point guard, or perhaps just 3 (frontcourt, swingman, backcourt). In any case, however you classified the players, you could calculate replacement levels for all the role-types.

On Aaron's Blog, you commented that each defensive player has equal responsibility (or somthing to that effect). I don't think this is correct: the players guarding opponent players who get the most öffensive chances" would have more defensive impact.
Craig B - Tuesday, February 10 2004 @ 10:17 AM EST (#79225) #
Robert, except at center (where there are more players due to the physical demands of the position and the relative fragility of almost anyone who stands seven feet tall) replacement levels are very close to equal. I will publish some work on this shortly, but I need to run data for five years first to get a better idea.
_S.K. - Tuesday, February 10 2004 @ 01:59 PM EST (#79226) #
Craig - will we ever see WinsAR for the whole league, or are you still tweaking the formula? I could help you with some of the data entry if that's all it is... I want to see which Raptors I should be picking on...
Craig B - Tuesday, February 10 2004 @ 02:04 PM EST (#79227) #
I'm working on it, I can post it here soon. I want to tweak the formula first, because the one I am using right now is a lousy kludge.
_S.K. - Tuesday, February 10 2004 @ 02:18 PM EST (#79228) #
Excellent. If you need any help with the droid work, just ask.
Mike D - Tuesday, February 10 2004 @ 02:45 PM EST (#79229) #
However, in most set offensive systems there isn't one guy who racks up lots of assists on set plays, and other guys who don't [...] assists are relatively well-distributed. No one player is particularly hurt by awarding all the "setup" credit to the assister only.

I don't think this quite addresses Pistol's point. Using his Spurs analogy, your argument indicates that your formula won't overrate Ginobili at the expense of, say, Parker or Bruce Bowen -- and vice versa. But either way, it substantially undervalues Duncan.

Duncan's counting stats are so insane, and his shooting percentage is so high, that your formula reveals Duncan to be the stud he is. But his mere presence in the post is responsible for way, way more open looks than is a replacement-level C or PF, for the reasons (i.e., drawing an automatic double team) that Pistol mentioned.
Craig B - Tuesday, February 10 2004 @ 02:55 PM EST (#79230) #
Yeah, Mike, you and Pistol are right. This in particular is one of the small factors that go into stars making the players around them better - the unavailability of help defense from Duncan's defender, the weakside cut that is opened when Shaq forces you into a zone.

I feel confident that we'll never measure it.
Craig B - Tuesday, February 10 2004 @ 03:14 PM EST (#79231) #
I want to see which Raptors I should be picking on...

Well, for the Raptors, it's obvious. Curry and Palacio, to pretty much any observer, are badly overmatched everywhere and have dragged the team down something fierce. Curry's easily the worst offender (lowest WinsAR in the NBA), but Palacio is a turnover machine who can't shoot (from inside, from 3, or from the line) and whose 20 minutes a night was just killing the Raptors.

The Raptors also have easily the most players playing under the replacement level. By WinsAR (figures are for the whole year and include play with other teams)

Marshall 6.3
Carter 5.3
Bosh 2.7
Peterson 1.0
Williams 0.7
Rose 0.7
Baxter 0.6
Bateer -0.1
Moiso -0.1
Murray -0.2
Mason -0.4
Archibald -0.4
Palacio -0.9
Curry -2.1

Overall, the Raptors' 23-25 record (on Feb. 7th) overstated their performance by 2.77 games according to Pythagorean, meaning their Pythag. record was 20-28.

One principal factor going into their poor performance on WinsAR is that the Raptors are a very good defensive team against shooters, and team FG% defense isn't yet included in the WinsAR stat like it is in my ADV stat. I'm leaning towards putting it in, which would mean most of the Raptors players would receive a small boost, maybe 0.1 or 0.2 win shares each. Some of the Rockets and Spurs would get a big boost, because those teams have been murder on opposing shooters. Some of the Blazers, Magic, and Kings would get knocked back, as teams shoot the lights out against them.

But I'm hesitant to do this, because I don't think it accurately encapsulates *individual* value, which is what I'm trying to do. In a nutshell? It's the Cuttino Mobley problem. The Rockets are killing opposing shooters; the Rockets are three points a game better than the league average, solely on forcing misses.

Now Cuttino Mobley leads the Rockets in minutes played. How much of the Rockets' success against field goals is due to Cuttino Mobley? Probably close to zero. About 50% is probably due to Yao Ming, and 50% due to a solid coaching staff. Mobley is a part of that team effort, but he's never been a particularly good man-on-nman defender.

So I have to ask myself whether making the adjustments will, in general, result in more or less accuracy. I'm leaning to more, but I may end up doing two metrics anyway (shades of Baseball Prospectus!)
Craig B - Tuesday, February 10 2004 @ 03:41 PM EST (#79232) #
Mason's got a funny stat, too. He has 16 fouls in his 88 minutes, or 8.7 fouls per 48 minutes. For a guard, that's utterly insane... there are no guards anywhere close to him (well, Darrick Martin has 8 per 48, but only has 48 minutes).

Archibald, by the way, among all players with 100 minutes, has 10.2 fouls per 48 minutes, which ties him for second-most with Jahidi White. They are just behind Yogi Stewart, leading the way with 10.5. I guess my appellation of "Angus MacPersonalfoul" for Archibald is pretty justified.

White, whose similarity score with an ox is about 958, is the easy winner for "most pain inflicted per minute" this season.

Actually, fouls per minute is a nice, quick-and-easy way (for guards) to see who is defensively overmatched. Guards who have no problem staying with their man rarely foul. The top foulers among guards all have slow feet and/or aren't mentally able to play NBA-caliber defense yet (yet, or any more). Look at this Top 16 list (80 minutes to qualify to get Mason on)... Roger Mason, Dan Dickau, Bimbo Coles, Mickael Pietrus, Dajuan Wagner, Travis Hansen, Marcus Banks, Rick Brunson, Erick Strickland, Mike Wilks, Mitchell Butler, Dahntay Jones, Bob Sura, Devin Brown, Raul Lopez, Richie Frahm.

Now I'll give Sura and (especially) Brown a pass. Brown is probably getting those fould mixing it up under the boards inplace of the AWOL Bruce Bowen - Brown is the best-rebounding guard in the league.

The rest of those guys do look like an all-matador team to me.
Craig B - Tuesday, February 10 2004 @ 03:47 PM EST (#79233) #
Lowest fouls per minute - Latrell Sprewell, Chris Whitney, Wesley Person, Michael Finley, Reggie Miller, Damon Jones, Nick Van Exel, Jason Williams, Earl Boykins, Allen Iverson, Mike Bibby, Antonio Daniels, Tracy McGrady, LeBron James, Jason Kidd.

Person and Miller have signed non-aggression pacts with the league, but other than them those 13 guys are probably all in the 30 fastest players in the league.

Forwards who don't foul? Almost all wing players. Peja Stojakovic, Tayshaun Prince, Jamal Mashburn, Eddie Robinson, Ben Wallace, Shawn Marion, Kevin Garnett, Bruce Bowen, Ron Artest, Glenn Robinson, Tim Duncan, Andrei Kirilenko.

Now look at those lists and tell me the refs don't give stars the benefit of the doubt!
Gerry - Tuesday, February 10 2004 @ 04:15 PM EST (#79234) #
Rick Carlisle and Kevin O'Neill have both given him significant roles on their teams. These are not stupid men.

Oh really? :)

I kid, I kid. Rick Carlisle is, of course, not a stupid man.


Craig, it sounds like you think O'Neill could do better. Care to elaborate?
Craig B - Tuesday, February 10 2004 @ 11:22 PM EST (#79235) #
Nah, I'm just having a shot at O'Neill, who has been given a pile of chickens*** and asked to make a bowl of chicken salad. He's done remarkably well to get them near .500.

I do think he plays favourites with bad players, and is inflexible. O'Neill's walk-it-up approach is horribly suited to his personnel, notably Williams, Carter, and Peterson (especially Mo Pete) who are very good players when running the floor. But O'Neill's style depends on never playing in transition.

And it won't matter, because even when the team wins, O'Neill's brand of basketball is utterly unwatchable. I remember his Marquette teams well.
_Kevin - Wednesday, February 11 2004 @ 10:49 AM EST (#79236) #
i don't know if it's still true because I find the NBA nearly unwatchable nowadays but I would explain away some of the excessive fouling on the young guys to getting no respect from the officials. In those ambiguous charge/block situations, the rookies always get screwed. Once they get a year or two under their belt and the officials start cutting them some slack, the defense really picks up.
_Nick M - Thursday, February 12 2004 @ 01:33 PM EST (#79237) #
Excellent article. I have two suggestions:

1. Incorporate offensive-fouls taken into IDV. Maybe:

OF = VOP

Bill Laimbeer, Dennis Rodman, and Scottie Pippen were masters of drawing charges. Though I suspect this statistic will favor power forwards & centers more than any other position.

2. How about a little historical perspective - what are the ADV values of Jordan, Pippen, and Payton in their primes? And whatever other excellent non-PF / non-C defenders you can think of.
Craig B - Thursday, February 12 2004 @ 02:42 PM EST (#79238) #
NickM, I believe that offensive fouls taken isn't an officially-kept stat, sadly! At any rate, it's not in my stat package.
Craig Names His NBA All-Defensive Team For Some Reason | 40 comments | Create New Account
The following comments are owned by whomever posted them. This site is not responsible for what they say.