SoTD: Bargnani's defensive numbers over 4 years

[Junkball]

I've never been fond of Bargnani, but conceded, at one point last year, that he is better than I thought.
Everybody needs to learn to like him, because he isn't going anywhere. How do these headlines sound?

"Colangelo Admits Big Bargnani Mistake"
"DeRozan Raptors' Top Dog"

They ain't gonna happen. Bargnani is the first offensive option on a team without many offensive options. He has his weaknesses and it's BC who deserves the ire if Bargnani is not paired with a defensive-minded centre who can rebound.

[BorisDK1]

While I agree that the foundation of any team defense is the ability to pressure the ball, that pressure is primarily dependant on the defensive ability of the big man anchoring the defense. The best defensive teams in the league always have good defensive big men which allow your wing players to pressure the ball heavier than when they have poor defensive big men playing behind them. There are many examples of this around the league (ie: no matter how good a defender Kobe was in 05-07 and no matter how much he could shut down a perimeter player, his team still ended up ranked in the bottom 3rd of the league in defense because of the inept big men. Whereas you look at a team like Orlando, who have surrounded Howard with some pretty stinky perimeter defenders who end up looking passable to good playing beside him AND the team, regardless of these players (Rashad, Turk, Vince, Jason Williams, Redick, Lee, Arroyo… ), still ranked as one of the top defensive teams in the league.

It's easier to limit Jose's defensive deficiencies on the perimeter than it is to limit Barg’s defensive deficiencies within the construct of overall team defensive impact.

OKC has guys just as poor defensively as far as difference-makers defensively as we do, yet they were a really good defensive team last year. Really, is Bargnani any more poor in supporting the perimeter than Nenad Krstic is? I don't think so. So the category "always" needs to be rejected, probably. And, I think we need to put a tremendous asterisk beside any team Dwight Howard has played on the past 2 1/2 years because he is an exceptional and singularly effective defender. You can put almost anybody out there and he'll make them look almost manageable, but he is the rarest of exceptions in that regard. There are no Dwight Howards in the league, except for Dwight Howard.

Nobody is arguing that a poor interior defensive presence is desirable. But the point is, when you go through the exercise of building a defense (which every coach has to do), it starts off with guarding the basketball. Everything breaks down when there's no pressure on the basketball. You can't defend screen sequences, you can't pressure passing lanes, you have to defend the post differently, etc. It's hard to summarize everything that is directly affected by inability to pressure the ball. I think it's a naive statement to say you can just "limit Jose's defensive deficiencies". Jay Triano did everything possible to do that this year, and it didn't make a difference. The only thing you can do that's effective in limiting Jose's defensive deficiencies is to 1) dress him in civilian clothes, 2) bring him off the bench for limited minutes against lesser talent.

[BorisDK1]

Boris: Did you manage to take a look at the three posts I made on page 25? There are some pretty big questions about the methodology you are using (the BoP stuff) here. If you could read those posts and address some of the concerns I raised, I would really appreciate it.

Like, the biggest one to me is that the Dean Oliver Drtg for Player X is very different from what actual basketballvalue.com style Drtg (which as you might know, just calculates the total number of points given up when player X is on the floor, and divides by the number of possessions.)

So in other words, there is a gap between Oliver's Drtg and reality. Anyway, that is just one of various concerns I had, so it would be great if you could address them.

I'll get into the methodology as much as I can. There are always a few things here and there that might need supplementary questions.

Firstly, about Dean Oliver's DRtg. He has two methods: the first is the estimated DRtg based on boxscore data, the second is through the data taken in Project Defensive Score Sheet (which is not published anywhere that I know of, except the stats I got through manually scorekeeping every game this year). The estimated DRtg, in short, doesn't do a bad job estimated the number of stops produced by a given player, but does a poor job estimating the number of scores allowed. This is the DRtg you can see, for example, at basketball-reference.com. Project Defensive Score Sheet (PDSS) gives us all data of all stops produced and scores allowed, not just estimating things based on the boxscore.

As I said before, there are five statistical categories. The overriding principle is "if a guy scores on a defender, he's tagged with a Field Goal Allowed; if he misses, the defender is credited with a Forced Miss, etc."

1. Forced misses (FM). When a player is guarding a player, and the player shoots and misses, this is a forced miss. (Excludes blocked shots). Credit can be shared: if a player drives and another defender comes over and contests the shot and it's missed, both players get .5 FM. Obviously this doesn't just apply to the guy being guarded at the start of the possession.
2. Field Goal Allowed (FGA). When a player is guarding another player, and the player shoots and scores, that defender is attributed a field goal allowed. Credit can be split in two-man game instances (screen sequence breakdown with equal culpability, for example).
3. Forced Turnover (FTO). When a player forces another player to turn the ball over (not including steals). So, if a player drives on another player and dribbles off his leg out of bounds, or travels, or commits a charge, etc.
4. Forced Free Throw (FFT). Successful free throw coming from a foul a player committed.
5. Forced Free Throw Miss (FFTM). Do I have to say it? Same as above, but unsuccessful.

I also keep the category "team" for uncontested action, like some (but not all) transition scores and situations when the defense elects to leave a guy open - help-and-recover scenarios, for instance, or for three-, five-, eight- and twenty-four second violations or stuff that simply wasn't due to any individual. A player dribbling off his foot unguarded in the backcourt, for example.

So the formula for stops is pretty predictable:

Stops = FTO + Stl + (FFT / 10) + (FM + Blk) * Forced Miss Weight * (1 - OppOReb%) + DReb * (1 - Forced Miss Weight)

"Forced miss weight" is a formula to calculate the relative value of a forced missed shot vs. a defensive rebound.

Scoring possessions allowed = FGA + .45 * (FFT + FFTM) * (1 - (1 - (FFT / (FFT + FFTM))^2 ) )

Stop% = Stops / (Stops + Scoring Possessions Allowed)

%DPoss = ( (Stops + Scoring Possessions Allowed) * (Team Minutes / 5) ) / (Game Possessions * Minutes Played)

And the DRat formula I've already posted. It differs from the estimated DRat in that not all players are assumed to have a DPoss% of .2, and the stop% isn't just estimated from boxscore data. So you need to separate what you know from Dr. Oliver's estimated DRat and just use his PDSS formulae. It's clearly the superior method, because you're not just guessing at who did what that wasn't directly related in the boxscore via. blocks, steals, defensive rebounds and guessing at who caused how many free throws based on PF.

I'm sorry I didn't respond right away, I have been busy today (my barbeque sauce was a knockout, and I made elk ribs for the first time with it tonight). Is there anything else you want to ask concerning this method?

[TheKushinator]

Boris, I'm heading out for lunch, so I'll answer you this afternoon. In the meantime, let me know if you think that Jarret Jack is a better defender than Jose. If your PDSS stat says so, I'm going to run team DRTG for Jack/Bargnani and Jack/No Bargnani for you...

Yes, I do. Jarrett was less effective as a starter than as a backup (stop% ~.540 off the bench, ~.480 as a starter), but certainly far better than Jose's .372 stop%.

Enjoy your lunch! I'm going to be busy in the kitchen this afternoon (I work weekends), making a new barbeque sauce.

Alright.

Jack and Bargnani played together 1630.67 minutes. The team DRTG was 114.76
Jack played without Bargnani for 612.75 minutes. The team DRTG was 110.96

Again. The team performed about 4 pts worse with Bargnani than without Bargnani. We can go down the list of players, and the result will always be the same.

Havent read all the posts, but wouldnt that be because Bargs primarily plays against starters who tend to be better than bench players. And Bargs backup (amir) is a pretty good defender so that skews it a bit because Amir can put up better defencive numbers vs bench players. Furthermore Bargs primarily plays with terrible to bad defenders (Bosh, Calderon, DD, Turk).

So you can go down the list with every player, but it will always come out the same because Bargs played the most minutes, versus the toughest competition and had the most inept teammates on the court with him (the starting lineup) for most of his minutes.

[BorisDK1]

Jack and Bargnani played together 1630.67 minutes. The team DRTG was 114.76
Jack played without Bargnani for 612.75 minutes. The team DRTG was 110.96

Again. The team performed about 4 pts worse with Bargnani than without Bargnani. We can go down the list of players, and the result will always be the same.

Now, you're getting that I don't think the on/off court data is all that helpful in evaluating an individual's data, right? Ultimately you're measuring not the individual himself, but peripheries. You're measuring the team with vs. the team without such a player, in which case that player is only 1/5 of what's being measured. The laugh test results don't always look good, either, like when Tim Duncan or Luke Mbah a Moute score very pedestrianly with it.

The superiority of PDSS to me, is that it's usable for a coach. Dean Oliver, of course, was a college assistant coach himself so his methods are designed especially with coaching decisions in mind. As a coach, I can't really do much with On/Off court 82games.com data. It's helpful to know, but I don't know after reading it some key information, like, "who's giving up what, versus who's stopping what?" All we can do is kind of guess with this method who is doing what, and that's not nearly as helpful IMO. And it's not at all helpful in the case of Calderon, whose numbers changed entirely based on coming off the bench vs. starting.

[timdunkit]

Heres my problem with a stat like this ... you can look at it as a 4 pt improvement? But a 4 point improvement to what? 110.96 DRTG still puts you 5th worst in the league and 0.8 pts away from being worst.

So basically, they played terrible defense without Bargnani but Jack but played even more terrible defense with Bargnani/Jack. There is a point of being terrible where statistical analysis should be considered flawed. To me if your DRTG is over 110, you SUCK defensively big time no matter whether your 111, 112, 113.

Thats my problem with presenting a comparative stat like that ... Its so easy to read it as a 4 point improvement, where in reality Its going from really really really sucky to really suck ... though in the end you still suck!

So defensive sucktitude has a finite ultra-suckiness?

Pretty much ... see to me the difference between a DRTG of 110-114 doesn't have to do with actually playing better defense because at that point, I think its safe to conclude that whatever it is with that DRTG is playing awful defense. That difference can come between difference in between OPP strength, bench vs starter and other factors.

Its easy to break statistics down. SS said with Jack/no Bargs we had a DRTG of 110 and with Bargs it was 114. What does that mean in terms of game to game basis (which Boris makes us a good point tells us nothing really)? But if we were to break it down, heres what it would tell us.

The raps played with an avg pace of 93.1, so if we scale the efficiency numbers back to game speed at which the Raps played at, the defense allowed becomes 102.4 vs 106.1 . The problem is that neither Jack or Bargs play 48 minutes a game. Jack played 27.4 minutes a game.

If in a certain game, Jack played all his minutes without Bargs (assuming Jack plays 27.4 min consistently), then the points scored would have been 58.5 points and without Bargs would have been 60.5 points. So inconclusion, if we take the game into account and the number of minutes Jack played, we basically made one more stop with Bargs of the court in the case that Bargs/Jack don't play togther at all vs if they play with each other for all the minutes Jack played (since he played less minutes then Bargs) ... thats right in an in game situation at most the difference between a line up with Jack/Bargs and Jack/No bargs at max is ONE STOP.

But like Boris said, these are just peripheries and like I just did, prove nothing ...

[Ripp]

I'll get into the methodology as much as I can. There are always a few things here and there that might need supplementary questions.

Yep, I understand the methodology...spent some time reading and thinking about it yesterday.

Firstly, about Dean Oliver's DRtg. He has two methods: the first is the estimated DRtg based on boxscore data, the second is through the data taken in Project Defensive Score Sheet (which is not published anywhere that I know of, except the stats I got through manually scorekeeping every game this year). The estimated DRtg, in short, doesn't do a bad job estimated the number of stops produced by a given player, but does a poor job estimating the number of scores allowed. This is the DRtg you can see, for example, at basketball-reference.com. Project Defensive Score Sheet (PDSS) gives us all data of all stops produced and scores allowed, not just estimating things based on the boxscore.

Yes. But neither of these Drtgs are actually what happened on the floor. Like, another way to compute "Drtg" is to divide the total number of points scored in a game by the number of possessions that happened. If you think about it, this is the most meaningful way to understand defensive impact...you take the number of points scored on a team and then normalize by the number of possession they were involved in. Both of Oliver's techniques are surrogates for this play-by-play based Drtg.
Like, if there is a discrepancy between the two, we should trust what actually happened on the floor (i.e., play-by-play Drtg) rather than the Oliver ones.

As I said before, there are five statistical categories. The overriding principle is "if a guy scores on a defender, he's tagged with a Field Goal Allowed; if he misses, the defender is credited with a Forced Miss, etc."

Yep, I understood this. And I took issue with the FFTM variable being taken as a positive, in the posts I made on page 25. Makes no sense to give a player credit for sending someone to the FT line, even if they miss...

Stops = FTO + Stl + (FFT / 10) + (FM + Blk) * Forced Miss Weight * (1 - OppOReb%) + DReb * (1 - Forced Miss Weight)

Again, why am I getting credit for sending someone to the free throw line? Why is that good?

"Forced miss weight" is a formula to calculate the relative value of a forced missed shot vs. a defensive rebound.

Yes, I saw this too, and the formula listed in Oliver's book. But the explanation doesn't seem very convincing. I mean, think about it yourself...what is the natural way to trade off between defensive rebounds and forced misses? There isn't any. And as you'll notice in the link I posted on page 25 frm Oliver's website (http://www.rawbw.com/~deano/articles/JordanvsOlaj.html), when he first did this he arbitrarily decided to weight defensive rebounds by 1/2.

%DPoss = ( (Stops + Scoring Possessions Allowed) * (Team Minutes / 5) ) / (Game Possessions * Minutes Played)

And the DRat formula I've already posted. It differs from the estimated DRat in that not all players are assumed to have a DPoss% of .2, and the stop% isn't just estimated from boxscore data. So you need to separate what you know from Dr. Oliver's estimated DRat and just use his PDSS formulae.

I wasn't really concerned too much about this particular aspect of the formula...the value being multiplied against %DPoss was the more interesting variable to me, not really the stuff before it.

It's clearly the superior method, because you're not just guessing at who did what that wasn't directly related in the boxscore via. blocks, steals, defensive rebounds and guessing at who caused how many free throws based on PF.

Again, neither method is "clearly" superior to anything. Even If your estimate of stop percentage measures reality, then it still isn't clear how good the model is, as Oliver said on his website:

The defensive formulas are much simpler, because methods for defensive evaluation aren't as well developed theoretically. Simply put, I have struggled with evaluating players' defenses for many years now. I came up with a basic method three or four years ago called defensive stops, which are a way of estimating how many times a player stops the opposition from scoring. It's not a bad first approximation, but it misses out on players like Joe Dumars and Glenn Rivers, who prevent their assignments from scoring by not allowing them good looks at the basket. They don't get many defensive rebounds or blocks and don't steal the ball much, but they shut down their men. Doug Steele came up with a good way for accounting for this type of player, the kind of player I call a good man defender. On the other hand, my method does best at evaluating team defense, which would include blocks, steals, and defensive rebounds. Doug has begun including these in his method, too, but he uses a form of linear weights, something I disapprove of rather heartily.

I'm sorry I didn't respond right away, I have been busy today (my barbeque sauce was a knockout, and I made elk ribs for the first time with it tonight). Is there anything else you want to ask concerning this method?

No biggie, hope you enjoyed your ribs. But like I said, hopefully you get the chance to look at the concerns I raised with the methodology on page 25. Like I keep saying, all of this stuff that Oliver does is meant to estimate something that we can calculate directly these days. If you don't have time to read the stuff on page 25, at the very least read the section called "Defensive Ratings" from his site:
http://www.rawbw.com/~deano/articles/JordanvsOlaj.html

Anyway, hopefully you see what I'm saying, and why this approach of his has some pretty significant limitations.

[Ripp]

Jack and Bargnani played together 1630.67 minutes. The team DRTG was 114.76
Jack played without Bargnani for 612.75 minutes. The team DRTG was 110.96

Again. The team performed about 4 pts worse with Bargnani than without Bargnani. We can go down the list of players, and the result will always be the same.

Now, you're getting that I don't think the on/off court data is all that helpful in evaluating an individual's data, right? Ultimately you're measuring not the individual himself, but peripheries. You're measuring the team with vs. the team without such a player, in which case that player is only 1/5 of what's being measured. The laugh test results don't always look good, either, like when Tim Duncan or Luke Mbah a Moute score very pedestrianly with it.

Suppose I have 1000 minutes of lineup A, and 1000 minutes of lineup B. Lineup A is as follows:

C = Bynum, PF = Pau Gasol, SF = Artest, SG = Kobe, PG = Me

and then lineup B is as follows:
C = Bynum, PF = Pau Gasol, SF = Artest, SG = Kobe, PG = Deron Williams

You are telling me that if i have a sufficiently large sample size of BOTH of these lineups, I can tell absolutely nothing about the comparative defensive abilities of me versus Deron Williams?

The superiority of PDSS to me, is that it's usable for a coach. Dean Oliver, of course, was a college assistant coach himself so his methods are designed especially with coaching decisions in mind.

User-friendliness of a tool is meaningless if the tool is giving you false or misleading results. It is better to have correct, but hard to get answers than wrong answers which are easy to get.

As a coach, I can't really do much with On/Off court 82games.com data. It's helpful to know, but I don't know after reading it some key information, like, "who's giving up what, versus who's stopping what?" All we can do is kind of guess with this method who is doing what, and that's not nearly as helpful IMO. And it's not at all helpful in the case of Calderon, whose numbers changed entirely based on coming off the bench vs. starting.

Eh... Stan Van Gundy and Erik Spoelstra both look at +/- data a lot.

[Local_NG_Idiot]

OKC has guys just as poor defensively as far as difference-makers defensively as we do, yet they were a really good defensive team last year. Really, is Bargnani any more poor in supporting the perimeter than Nenad Krstic is? I don't think so. So the category "always" needs to be rejected, probably. And, I think we need to put a tremendous asterisk beside any team Dwight Howard has played on the past 2 1/2 years because he is an exceptional and singularly effective defender. You can put almost anybody out there and he'll make them look almost manageable, but he is the rarest of exceptions in that regard. There are no Dwight Howards in the league, except for Dwight Howard.

Nobody is arguing that a poor interior defensive presence is desirable. But the point is, when you go through the exercise of building a defense (which every coach has to do), it starts off with guarding the basketball. Everything breaks down when there's no pressure on the basketball. You can't defend screen sequences, you can't pressure passing lanes, you have to defend the post differently, etc. It's hard to summarize everything that is directly affected by inability to pressure the ball. I think it's a naive statement to say you can just "limit Jose's defensive deficiencies". Jay Triano did everything possible to do that this year, and it didn't make a difference. The only thing you can do that's effective in limiting Jose's defensive deficiencies is to 1) dress him in civilian clothes, 2) bring him off the bench for limited minutes against lesser talent.

I already mentioned OKC was the exception and the reason for that is their length, youth, athleticism and depth on the wings which again like Howard you have to put an asterisk beside.

I used Howard as an example against Kobe since both have seen poor defenders around them with completely different outcomes. Orlando has seen defensive success regardless of who they put on the perimeter since those poorer perimeter defenders can apply additional pressure knowing the consequences of them getting beat still might not result in a score with Howard behind them. Couple that with a very good demanding defensive coach who explains their rotations, where they would like to direct the P&R’s, who’s stepping up in the paint to play passing lanes, etc… and you still have a very good defensive team. The Lakers have a top 3 on ball defender in Kobe, yet when he had poor defenders behind him (Kwame, Mihm, and Cook) logging major minutes, the team was in the bottom third of the league in defense. Him and his other perimeter teammates could apply all the ball pressure in the world, but when they got beat, or put in P&R situations, the bigs behind them couldn’t produce.

Again, I am not attempting to discredit the fact that good defensive teams start with ball pressure, but how teams are able to apply it, which they do by having good defending big men (always might have been a little strong for emphasis, but let’s say 9 times out of 10).
Examples? Detroit, great defensively when they had the Wallaces and Dyce (top 5 in the league year after year), now with their frontcourt of an over the hill Ben, Maxiell and CV (bottom of the league). Peirce and the Celts were always taking hits for their defense and Shuttlesworth was always considered average at best, but as soon as they were backed by KG and an improved Perkins, they magically transform into ball stoppers? Go look at Houston’s defense with/without Yao. NJ went from a 98DRTG and absolute beast of a defensive team making the finals with Kenyon Martin, to a 103DRTG and second round exits with Vince instead.

1 more for you, go check out a teams’ defensive ranking the last year Mutombo played for them, and compare it with the year after when they played without him.

[timdunkit]

Personally, I find that defensive statistic are measuring the wrong elements. I don't think its fair to judge a player on points allowed because even there, there going to be some mistakes. Take DRTG for example, free throws are accounted by it but nobody defends a free throw ... in fact the Raps had the highest ft % shot against them. This would be accounted into a number like DRTG but in the end shouldn't because it doesn't measure whats its suppose to.

The good news is that statistics are evolving but there hasn't been an defensive statistics that captures the essence of statistic.

Personally, I think for defensive statistics, we'd know more people like Boris who are willing to measure small things that actually mean a players are playing defense. I'd love to know, which big contests the most shots per game, which wings are prone to 1 on 1 penetrations, best players at fighting picks, which wings have the best close outs and which players attempts to take the most charges etc ...

[Indeed]

Firstly, about Dean Oliver's DRtg. He has two methods: the first is the estimated DRtg based on boxscore data, the second is through the data taken in Project Defensive Score Sheet (which is not published anywhere that I know of, except the stats I got through manually scorekeeping every game this year). The estimated DRtg, in short, doesn't do a bad job estimated the number of stops produced by a given player, but does a poor job estimating the number of scores allowed. This is the DRtg you can see, for example, at basketball-reference.com. Project Defensive Score Sheet (PDSS) gives us all data of all stops produced and scores allowed, not just estimating things based on the boxscore.

Yes. But neither of these Drtgs are actually what happened on the floor. Like, another way to compute "Drtg" is to divide the total number of points scored in a game by the number of possessions that happened. If you think about it, this is the most meaningful way to understand defensive impact...you take the number of points scored on a team and then normalize by the number of possession they were involved in. Both of Oliver's techniques are surrogates for this play-by-play based Drtg.
Like, if there is a discrepancy between the two, we should trust what actually happened on the floor (i.e., play-by-play Drtg) rather than the Oliver ones.

If you want to convert defense into numbers, you might have to categorize different situations.
Some separate them by action/reaction (missed FG, shots altering, etc) for individual data
Some separate them by strategy matchup (pick and roll vs man to man high low switch, etc)
Some separate them by opponent strength (shooting, athletic, transition, etc)
Some separate them by assignments (don't leave xxx player, force xxx to pass, etc)

While DRtg sometimes are important to see which lineup works better together (although it is kind of obvious in coaches mind).

[Hendrix]

like when Tim Duncan or Luke Mbah a Moute score very pedestrianly with it.

.

I don't think it's an exact reflection of what is going on. And there is definitly some caevets needed. But with a bigger sample size I think it gives a good general idea. Timmy D over the last 4 years helps the defense 2.4 points/100, and Mbah a moute 3.25/100 better since entering the league. Maybe it's not exact, and maybe Duncan\s not as good as he once was. But I think the fact they are improving a defensive team that is already very good says something. And over a large sample size making a teams defense worse when it is already pretty damn bad also says something imo.

[BorisDK1]

Yes. But neither of these Drtgs are actually what happened on the floor. Like, another way to compute "Drtg" is to divide the total number of points scored in a game by the number of possessions that happened. If you think about it, this is the most meaningful way to understand defensive impact...you take the number of points scored on a team and then normalize by the number of possession they were involved in. Both of Oliver's techniques are surrogates for this play-by-play based Drtg.
Like, if there is a discrepancy between the two, we should trust what actually happened on the floor (i.e., play-by-play Drtg) rather than the Oliver ones.

Well, firstly we need to remember that this is not some sort of Holy Grail one-number metric. There are constituent parts to PDSS ratings: the stop%, DPoss% and finally the DRat. And yes, it does reflect what actually happened on the floor because at some point you have to reconcile individual play with team performance defensively.

I wouldn't say this method is a "surrogate" for PBP metrics, because PBP metrics don't contain defensive data. You don't know who did what, other than if a block, steal, foul or defensive rebound takes place. Secondly, it precedes PBP metrics so it obviously isn't a surrogate anything.

Yep, I understood this. And I took issue with the FFTM variable being taken as a positive, in the posts I made on page 25. Makes no sense to give a player credit for sending someone to the FT line, even if they miss...

Sure, it makes sense: often, you choose to foul to stop an easy score to save your team some points. That's a partial stop being recorded on a possession, so why wouldn't it? Secondly, it doesn't count for much so it's not a big issue. Thirdly, it has to be accounted for at some point.

Again, why am I getting credit for sending someone to the free throw line? Why is that good?

Propose an alternative, then.

Yes, I saw this too, and the formula listed in Oliver's book. But the explanation doesn't seem very convincing. I mean, think about it yourself...what is the natural way to trade off between defensive rebounds and forced misses? There isn't any. And as you'll notice in the link I posted on page 25 frm Oliver's website (http://www.rawbw.com/~deano/articles/JordanvsOlaj.html), when he first did this he arbitrarily decided to weight defensive rebounds by 1/2.

BoP was written about seven years after that article. Oliver clearly worked with this somewhat and came across a more dynamic and accurate method of estimating the value of a forced miss. Improving the method condemns it, how exactly?

I wasn't really concerned too much about this particular aspect of the formula...the value being multiplied against %DPoss was the more interesting variable to me, not really the stuff before it.

That doesn't really apply to PDSS numbers, though.

Again, neither method is "clearly" superior to anything. Even If your estimate of stop percentage measures reality, then it still isn't clear how good the model is, as Oliver said on his website:

The defensive formulas are much simpler, because methods for defensive evaluation aren't as well developed theoretically. Simply put, I have struggled with evaluating players' defenses for many years now. I came up with a basic method three or four years ago called defensive stops, which are a way of estimating how many times a player stops the opposition from scoring. It's not a bad first approximation, but it misses out on players like Joe Dumars and Glenn Rivers, who prevent their assignments from scoring by not allowing them good looks at the basket. They don't get many defensive rebounds or blocks and don't steal the ball much, but they shut down their men. Doug Steele came up with a good way for accounting for this type of player, the kind of player I call a good man defender. On the other hand, my method does best at evaluating team defense, which would include blocks, steals, and defensive rebounds. Doug has begun including these in his method, too, but he uses a form of linear weights, something I disapprove of rather heartily.

And Dean is referring to the estimated DRtg, there, not PDSS...

No biggie, hope you enjoyed your ribs. But like I said, hopefully you get the chance to look at the concerns I raised with the methodology on page 25. Like I keep saying, all of this stuff that Oliver does is meant to estimate something that we can calculate directly these days. If you don't have time to read the stuff on page 25, at the very least read the section called "Defensive Ratings" from his site:
http://www.rawbw.com/~deano/articles/JordanvsOlaj.html

Anyway, hopefully you see what I'm saying, and why this approach of his has some pretty significant limitations.

I've read his website often, but I don't think you're clear in that he doesn't refer to PDSS on his site at all, and it is much older than Basketball on Paper.

[Indeed]

Jack and Bargnani played together 1630.67 minutes. The team DRTG was 114.76
Jack played without Bargnani for 612.75 minutes. The team DRTG was 110.96

Again. The team performed about 4 pts worse with Bargnani than without Bargnani. We can go down the list of players, and the result will always be the same.

Now, you're getting that I don't think the on/off court data is all that helpful in evaluating an individual's data, right? Ultimately you're measuring not the individual himself, but peripheries. You're measuring the team with vs. the team without such a player, in which case that player is only 1/5 of what's being measured. The laugh test results don't always look good, either, like when Tim Duncan or Luke Mbah a Moute score very pedestrianly with it.

Suppose I have 1000 minutes of lineup A, and 1000 minutes of lineup B. Lineup A is as follows:

C = Bynum, PF = Pau Gasol, SF = Artest, SG = Kobe, PG = Me

and then lineup B is as follows:
C = Bynum, PF = Pau Gasol, SF = Indeed, SG = Kobe, PG = Deron Williams

You are telling me that if i have a sufficiently large sample size of BOTH of these lineups, I can tell absolutely nothing about the comparative defensive abilities of me versus Deron Williams?

You sure it is ONLY you being substitute? Unless they both played the same position and against the same offense and system, otherwise, you have two different formulas here:

i) DRtg1 = x1 + x2 + x3 + x4 + x5
ii) DRtg2 = x1 + x2 + x6 + x4 + x7
=> DRtg1 - DRtg2 = (x3 + x5) - (x6 - x7) --- By substitution law (i and ii)
is not equal to Drtg1 > DRtg2 <=> x5 > x7

[supersub15]

Now, you're getting that I don't think the on/off court data is all that helpful in evaluating an individual's data, right? Ultimately you're measuring not the individual himself, but peripheries. You're measuring the team with vs. the team without such a player, in which case that player is only 1/5 of what's being measured. The laugh test results don't always look good, either, like when Tim Duncan or Luke Mbah a Moute score very pedestrianly with it.

The superiority of PDSS to me, is that it's usable for a coach. Dean Oliver, of course, was a college assistant coach himself so his methods are designed especially with coaching decisions in mind. As a coach, I can't really do much with On/Off court 82games.com data. It's helpful to know, but I don't know after reading it some key information, like, "who's giving up what, versus who's stopping what?" All we can do is kind of guess with this method who is doing what, and that's not nearly as helpful IMO. And it's not at all helpful in the case of Calderon, whose numbers changed entirely based on coming off the bench vs. starting.

I know what team DRTG is measuring, but when 4000+ minutes of data with Bargnani and another 4000+ minutes without Bargnani show 2 completely different outcomes, with the same teammates, the same system, and the same opponent, there's only one conclusion to make. I am not using it haphazardly here.

You told me that Jack was better than Calderon defensively according to PDSS, yet we got destroyed worse than with Calderon, when Jack played with Bargnani.

I am probably not going to convince you otherwise, and that's fine. Let's agree to disagree.

[Ripp]

If you want to convert defense into numbers, you might have to categorize different situations.
Some separate them by action/reaction (missed FG, shots altering, etc) for individual data
Some separate them by strategy matchup (pick and roll vs man to man high low switch, etc)
Some separate them by opponent strength (shooting, athletic, transition, etc)
Some separate them by assignments (don't leave xxx player, force xxx to pass, etc)

While DRtg sometimes are important to see which lineup works better together (although it is kind of obvious in coaches mind).

Yeah, that is the ultimate question. Like, sometimes you see kooky results in Drtg or On/Off sort of results. This probably means that your data is biased, and/or you don't have enough data to wash away the noise. Oliver's core idea is very good....use different events that happened on the floor to "correct" for small errors and biases that show up in your +/- data. My main beef is the specific way he is doing this, not necessarily the core idea.

Like...imagine if Forced Misses and all those sorts of variables were tracked in box scores. That would be wonderful, because you could then come up with a new PER formula that takes into account defense more accurately. Maybe Dwight Howard has the highest PER in the league if you took Forced Misses into account and assigned them some weighting.
So basically, I like this idea, but would want to see sort of...expanded box scores for the season. I can then take those box scores and make a new PER formula, new Statistical +/- formula, etc.
In fact, I wish Synergy would publish advanced box scores like this...there is a lot one could do if you had access to that type of data.

[Crazy-Canuck]

I can't be bothered to read all these posts, but is the crux of Boris's argument that the defensive woes of the Raps were no Bargnani's fault, but someone else? Or is he saying Bargnani is a good or passable C defensively?
.

Boris is arguing that Jose was the biggest reason for our defensive woes based on his stats.

Everyone else is saying Bargs was the prime culprit based on drtg (on vs off).

[Local_NG_Idiot]

I can't be bothered to read all these posts, but is the crux of Boris's argument that the defensive woes of the Raps were no Bargnani's fault, but someone else? Or is he saying Bargnani is a good or passable C defensively?
.

Boris is arguing that Jose was the biggest reason for our defensive woes based on his stats.

Everyone else is saying Bargs was the prime culprit based on drtg (on vs off).

And he's attempting to defend this POV with the theory that perimeter defense and ball pressure is more important than interior big men who play defense which is so far out ther I'm surprised no one else has picked up on this.

[cdel00]

Dood I been preaching ball pressure for years, Boris has stats that back what observation and knowledge of sports has said for years. When you take away a person's time and space they make far more mistakes than when you let them set up the play as designed. Ball pressure, bounce passes, boards and bank shots are the killer B's of basketball.

Best set of ballboy posts ever!

Boris has made an excellent statisitically based counter argument with a wealth of insight and analysis.

I respect Sub's findings as more than just anomolous based on the size of the data set. However I suggest that since Bargnani has been used more as a prime time player he has been ON vs other team's best and when he is OFF it's vs the other team's bench therefore when you see the ON/OFF numbers you must consider the nature of the data set as you are comparing 19-25 PER stars vs 10-15 PER roleplayers.

[Indeed]

If you want to convert defense into numbers, you might have to categorize different situations.
Some separate them by action/reaction (missed FG, shots altering, etc) for individual data
Some separate them by strategy matchup (pick and roll vs man to man high low switch, etc)
Some separate them by opponent strength (shooting, athletic, transition, etc)
Some separate them by assignments (don't leave xxx player, force xxx to pass, etc)

While DRtg sometimes are important to see which lineup works better together (although it is kind of obvious in coaches mind).

Yeah, that is the ultimate question. Like, sometimes you see kooky results in Drtg or On/Off sort of results. This probably means that your data is biased, and/or you don't have enough data to wash away the noise. Oliver's core idea is very good....use different events that happened on the floor to "correct" for small errors and biases that show up in your +/- data. My main beef is the specific way he is doing this, not necessarily the core idea.

Like...imagine if Forced Misses and all those sorts of variables were tracked in box scores. That would be wonderful, because you could then come up with a new PER formula that takes into account defense more accurately. Maybe Dwight Howard has the highest PER in the league if you took Forced Misses into account and assigned them some weighting.
So basically, I like this idea, but would want to see sort of...expanded box scores for the season. I can then take those box scores and make a new PER formula, new Statistical +/- formula, etc.
In fact, I wish Synergy would publish advanced box scores like this...there is a lot one could do if you had access to that type of data.

I haven't fully read Oliver's formula, but from your conversation, it is not perfect.
It doesn't cover mis-matches or strategy.
Eg. When a big posting up on a guard, it doesn't mean the guard is a bad defender to allow high percentage of shots. I guess stats doesn't show some of those situation.

Perhaps it is good for isolation or postup, where it is clearly one on one defense, otherwise, a switched could mess everything up.