Is PDSS a good way to analyze the Raptors defense?

[Ripp]

What is the formual for individual drat and team drat? Maybe the two cannot be correlated by the formula you use.

The formulas for individual DRat were posted in the previous thread. This post by Boris has most of them:
viewtopic.php?f=32&t=1047850&start=420#p24798256
viewtopic.php?f=32&t=1047850&start=345#p24789518

Most of the rest of the stuff is in Dean Oliver's book, which is also in google docs. Honestly, it is unfortunate there is not just a Wikipedia article with everything in it, it is annoying as hell hunting around for one place where you can find his entire methodology w/o buying his book.
Team DRat is calculated just be adding up the total number of points scored on the team, then dividing by the number of possessions they played in. The values on basketballvalue.com calculates this value exactly for the Raps (113.2), and www.basketball-reference.com estimates it indirectly from season box score information (113.15).
I use the BBValue numbers in my spreadsheet, but you could also use the estimate if you like and not change things much.

[dagger]

So what we have here is anger over results that you don't like, therefore you question the method (without understanding it particularly well). So I guess we're not supposed to pay attention to your numerous mistakes of fact, we'll sweep all those under the rug - we've got some hardcore venting that something has come along to challenge statboy assumptions.

I'm going to get offensive here but it's not necessarily directed at you personally for the most part. I get really tired of attitude pretty quickly: I have no truck for it. I hate the attitude that comes from a certain segment that says, "who needs stats? I get everything I need with my eyes" and the other segment that says, "I've got my spreadsheets and some methods I really like and this is gospel and anything that doesn't agree goes out the window". And to be perfectly honest, you are acting very much the latter in this case. I'm not passing judgment on you or your basketball acumen, but in this discussion your conduct is very much the latter.

Post of the year.

Boris, there are some obsessive compulsive posters here who cannot ever admit that there might be empirical conclusions that vary from their own, so they stick with their methodology like a Republican who won't stop calling Obama a Muslim, basically because it feels good, not right.

Ripp, like Harry Palmer, can't ever back down on anything. Challenged, they try even harder to woo the suckers with their pseudo erudition. (And of course, this debate goes all the way back to the 2006 draft and trying to justify support for, or opposition to, the player selected in June 2006.)

There are lots of examples of the leaky dam in nature. A small leak leads to a breach of the entire dam. The concept is pretty damn simple. A leaky perimeter leads to the collapse of the entire defence. Forces of nature.

However, I don't much care whose methodology is right because I'm not going to sit and chart these things or analyze statistical websites all day. My hat's off to those who try, but basketball has a much higher inter-dependence among teammates than baseball so I don't accept the importance of stats to the same degree that I do for a pitcher.

I do laugh, however, for those who are so unprofessional that they can't conceive of the notion that their sweeping, one-sided conclusions should be more nuanced - more grey than black and white.

[supersub15]

"I've got my spreadsheets and some methods I really like and this is gospel and anything that doesn't agree goes out the window".

Is this directed at me?

[elmer_yuck]

Attitude aside, the question is extremely valid. If individual PDSS shows Jose as the only rotation player that's below average then why does the overall defense still stink more often than not when he's not in the lineup? Especially considering PDSS shows the other players as above average individual defenders.

The answer is, "team" (uncontested action) isn't really being accounted for. 8% of the team's possessions went unaccountable to any specific player this year, and that is going to account for how things look for individuals.

So adjust your numbers so they pass the smell test. Apportion those team possessions or something like that. The individual values should be related to the team value. Otherwise the numbers don't make sense.
A little less ego and willingness to accept criticism would be appreciated too.

[dagger]

Attitude aside, the question is extremely valid. If individual PDSS shows Jose as the only rotation player that's below average then why does the overall defense still stink more often than not when he's not in the lineup? Especially considering PDSS shows the other players as above average individual defenders.

The answer is, "team" (uncontested action) isn't really being accounted for. 8% of the team's possessions went unaccountable to any specific player this year, and that is going to account for how things look for individuals.

So adjust your numbers so they pass the smell test. Apportion those team possessions or something like that. The individual values should be related to the team value. Otherwise the numbers don't make sense.
A little less ego and willingness to accept criticism would be appreciated too.

So they past the smell test? You mean adopt your views, which is ridiculous.

You haters are so insecure. You hate the message, so you want to shoot the messenger.

[Reignman]

So what we have here is anger over results that you don't like, therefore you question the method (without understanding it particularly well). So I guess we're not supposed to pay attention to your numerous mistakes of fact, we'll sweep all those under the rug - we've got some hardcore venting that something has come along to challenge statboy assumptions.

I'm going to get offensive here but it's not necessarily directed at you personally for the most part. I get really tired of attitude pretty quickly: I have no truck for it. I hate the attitude that comes from a certain segment that says, "who needs stats? I get everything I need with my eyes" and the other segment that says, "I've got my spreadsheets and some methods I really like and this is gospel and anything that doesn't agree goes out the window". And to be perfectly honest, you are acting very much the latter in this case. I'm not passing judgment on you or your basketball acumen, but in this discussion your conduct is very much the latter.

Post of the year.

Boris, there are some obsessive compulsive posters here who cannot ever admit that there might be empirical conclusions that vary from their own, so they stick with their methodology like a Republican who won't stop calling Obama a Muslim, basically because it feels good, not right.

Ripp, like Harry Palmer, can't ever back down on anything. Challenged, they try even harder to woo the suckers with their pseudo erudition. (And of course, this debate goes all the way back to the 2006 draft and trying to justify support for, or opposition to, the player selected in June 2006.)

There are lots of examples of the leaky dam in nature. A small leak leads to a breach of the entire dam. The concept is pretty damn simple. A leaky perimeter leads to the collapse of the entire defence. Forces of nature.

However, I don't much care whose methodology is right because I'm not going to sit and chart these things or analyze statistical websites all day. My hat's off to those who try, but basketball has a much higher inter-dependence among teammates than baseball so I don't accept the importance of stats to the same degree that I do for a pitcher.

I do laugh, however, for those who are so unprofessional that they can't conceive of the notion that their sweeping, one-sided conclusions should be more nuanced - more grey than black and white.

Actually, in this instance both sides make valid points and counter points. Boris made his and now there are some very big questions that remain outstanding. Asking for clarification doesn't necessarily mean they aren't willing to budge, they just want a reason to budge.

[Reignman]

Attitude aside, the question is extremely valid. If individual PDSS shows Jose as the only rotation player that's below average then why does the overall defense still stink more often than not when he's not in the lineup? Especially considering PDSS shows the other players as above average individual defenders.

The answer is, "team" (uncontested action) isn't really being accounted for. 8% of the team's possessions went unaccountable to any specific player this year, and that is going to account for how things look for individuals.

Is that really the "answer" or is it the only variable that's unaccounted for and is 8% really what's causing these numbers to be this skewed?

IMO, PDSS generates more questions than answers and for a guy like me that bases his basketball analysis mostly on empirical evidence I just don't think this portrays an accurate description of what I see on the floor. I could be wrong but after going through all the information provided over the last two days it just isn't clicking for me.

[Ripp]

Boris, a few questions for you:

1) Can I go from PDSS Drtgs to understanding the performance of the lineups? If so, how do I do this?
2) If my individual DRat is 110 and the team drat is 113, am I a good or bad defender? What if it is 110 versus 106?
3) If I have a team with team Drtg of say 113, and say only one player with an individual DRat higher than this, is it safe to say that he is the worst defender on the team? Should it not then be true that lineups that don't have this guy have Drtgs better than the team average of 113?
4) You said the following:

I understand your criticism, and I've tried to interact with it fairly (I wish this method would get the same treatment from you!). But it's funny that when you sum up all the individual defensive possessions, and you estimate the value of FGA and sum that all up from PDSS data, you get something that looks an awful hell of a lot like the team's actual DRat of 113.2. It gives you 113.9, and that accuracy will improve next year when I specifically track 3FGA so I'm not left guessing about the value of FGA.

Can you put the calculation in a spreadsheet, and then on Google docs for us to see?
5) Regarding the following:

So the "smell test" that you kept referring to, doesn't seem quite so outrageous. It seems to more or less work, far more accurately than you're willing to give it credit for.

My objection is this. You have a technique, an algorithm for rating player defense, yes? Now, I make available to your algorithm the Team DRtg (in this case, 113.2). Your algorithm then spits out individual DRats, and a way to combine them to get the Team Drtg again. Now, if my only metric for judging how good this algorithm is is that it gives me the correct Drtg back, then this is not particularly impressive. Even the fake RippRating algorithm I gave you earlier can do a very good job of giving you something close. An even crummier algorithm can just say that each player's Drtg is just exactly equal to the Team Drtg.
So do you see my objection to justifying your approach only because it gives the correct Drtg (something that it does not appear to do, as of now?) It has to do more than just give me back the answer I told it.

[elmer_yuck]

Dagger, I'm not complaining about the message.
I't's the numbers. And the attitude.

[Ripp]

So they past the smell test? You mean adopt your views, which is ridiculous.

You haters are so insecure. You hate the message, so you want to shoot the messenger.

Dagger. Listen. Imagine his algorithm as a magical genie that will tell you who is doing a good job defensively, and who is not. Suppose this genie says to you, "Before I answer your question, I'll ask for you to give me the following pieces of information, including the Team's Drtg."

Now, this genie then takes all that information, and then writes on a sheet of paper the ratings for each of the players. How are you to know whether this genie actually gave you something close to the truth or not? Suppose he is a lazy genie, and instead just takes the Team's Drtg and then randomly adds +1 or -1 to it, and tells you that is the rating for each player. How are you supposed to independently verify that the genie's answers are correct?

We need some sort of way to "test" the genie's answers. The most basic test I can think of is to see if the numbers can be combined in a way that can yield the team's Drtg. In particular, if the genie writes down a list of players, and says that Jose is by far the worst defender and that everyone else is above-average (relative to the team), then I expect that lineups not involving Jose should be better than average. If this is not the case, why should I trust what the genie is telling me? He might be lazy, lying, or not even knowing what he is doing.

[BorisDK1]

Is this directed at me?

Not directed at anybody in particular.

[BorisDK1]

So adjust your numbers so they pass the smell test. Apportion those team possessions or something like that. The individual values should be related to the team value. Otherwise the numbers don't make sense.
A little less ego and willingness to accept criticism would be appreciated too.

"Adjust my numbers" - what? "Team" scores allowed and stops produced happened. People forget that PDSS data isn't derived from some dude reading NBA.com PBP and guessing who was responsible for what: I watched every single possession this past year, until my ears were bleeding and my retinas detached from watching some of the defensive gong show I had to sit through.

I think you misunderstand: "team" is a category for defensive possession outcomes that couldn't be attributed to any specific player. 3 second violations, for example, or guys stepping out of bounds when not guarded by anybody in particular or whatever. I can't "adjust" those: they happened. They actually happened. I've asked for help from APBRmetrics in distributing those numbers, I received no ideas for help. So they have to be dealt with for what they are.

[dagger]

It's basketball, not baseball, and there is more interdependence in basketball, so I don't think there is a methodology - yours or his - that explains all, and that's why teams employ scouts and not robots working on algorithms. Given the imperfect nature of advanced basketball stats, there is always going to be a gray zone called interpretation in which the objective becomes the subjective. There is no reconciling differences at the subjective level. You agree to disagree. From my point of view, Boris' entry point - ball pressure, which begins at the perimeter - is just as valid as anyone's theory that it all hinges on the center.

In basketball, indeed in sports generally, there are lots of factors that just don't fit in a spread sheet: When you star leaves the bar, whether he's had a flu shot, whether he's playing with a full deck. It's impossible to isolate one player's performance. We spend all kinds of time arguing that Bosh was a sensational player and everyone else held him back, then turn around and deny the relationship called inter-dependence in a team game. It makes zero sense.

There is very little in the first 35 pages of the Bargnani defence thread about the team. It's basically 35 pages of self-justification of one's views that we should have drafted Rudy Gay, the Overpay.

It's juts another manifestation of some Raptors' fans need to constantly reformulate their negative view of Bargnani, which from year to year keeps evolving with his game. He's not the perfect player or the dream #1 overall pick. But even when one admits that, they are not allowed to like the contribution he makes because in the eyes of the hater, there is no middle ground, there is no compromise, there must a scapegoat, always a scapegoat.

[Ripp]

Boris: I want to make something clear, I absolutely love the PDSS sheet you did. It is a wonderful idea to tabulate additional defensive statistics like this. My issue is not with you or that piece of work, or even Oliver's work (which I think has some cool ideas in it.) My issue is, I don't think that the translation from these additional counting statistics you've calculated, to Stop%, then to individual Drtg and then to statements like, "Bargnani is the 3rd best defender on the team, and Calderon by far the worst" is necessarily meaningul, unless we have some independent way of verifying that the individual Drtgs you came up with are correct. Like, if you have some sort of ranking system for players, then it is always nice when you have independent ways of making sure that they are consistent with what actually happened, etc.

[ranger001]

Attitude aside, the question is extremely valid. If individual PDSS shows Jose as the only rotation player that's below average then why does the overall defense still stink more often than not when he's not in the lineup? Especially considering PDSS shows the other players as above average individual defenders.

The answer is, "team" (uncontested action) isn't really being accounted for. 8% of the team's possessions went unaccountable to any specific player this year, and that is going to account for how things look for individuals.

I haven't looked at the formulas but as an illustration of how a small sample can skew things lets say players Alex and Paul are great one on one defenders. The offense however decides to go with pick and roll and Alex/Paul end up giving up layups 90% of the time since they don't know how to defend the pick and roll. In this case the team will look bad but the individual drtg of Alex and Paul will look good.

[darth_federer]

Boris: I want to make something clear, I absolutely love the PDSS sheet you did. It is a wonderful idea to tabulate additional defensive statistics like this. My issue is not with you or that piece of work, or even Oliver's work (which I think has some cool ideas in it.) My issue is, I don't think that the translation from these additional counting statistics you've calculated, to Stop%, then to individual Drtg and then to statements like, "Bargnani is the 3rd best defender on the team, and Calderon by far the worst" is necessarily meaningul, unless we have some independent way of verifying that the individual Drtgs you came up with are correct. Like, if you have some sort of ranking system for players, then it is always nice when you have independent ways of making sure that they are consistent with what actually happened, etc.

So what youre saying is that you cant believe that Bargnani is the 3rd best defender. There MUST be some problem with his work right?

[Ripp]

So what youre saying is that you cant believe that Bargnani is the 3rd best defender. There MUST be some problem with his work right?

Did you not understand my point about averages? If you remove the biggest number from a list, then the average must decrease. Remove the smallest, average has to go up. Similarly, if Jose is the worst defender and by far the worst defender, then I'd better see lineups w/o him doing pretty well...

[BorisDK1]

1) Can I go from PDSS Drtgs to understanding the performance of the lineups? If so, how do I do this?

The specific lineups? No. You can, however, use it to project performance, like any other efficiency-based individual metric.

2) If my individual DRat is 110 and the team drat is 113, am I a good or bad defender? What if it is 110 versus 106?

If you're an individual with a DRat of 110, you are not a particularly effective defender - regardless of what you're team's DRat is.

3) If I have a team with team Drtg of say 113, and say only one player with an individual DRat higher than this, is it safe to say that he is the worst defender on the team? Should it not then be true that lineups that don't have this guy have Drtgs better than the team average of 113?

Not necessarily, because the entirety of team play isn't encapsulated in one individual's defensive performance.

You seem to be forgetting that what was documented, actually happened. I watched these games, man! Every single possession and documented it accordingly. I'm not guessing with lineup data, I'm not guessing what happened with imaginary counterparts - what happened, happened.

Can you put the calculation in a spreadsheet, and then on Google docs for us to see?

Already have, it was on my original spreadsheet. Just sum the fields in the final sum field for "Stops" and "ScPoss" and voilà - individual total possessions.

My objection is this. You have a technique, an algorithm for rating player defense, yes? Now, I make available to your algorithm the Team DRtg (in this case, 113.2). Your algorithm then spits out individual DRats, and a way to combine them to get the Team Drtg again. Now, if my only metric for judging how good this algorithm is is that it gives me the correct Drtg back, then this is not particularly impressive. Even the fake RippRating algorithm I gave you earlier can do a very good job of giving you something close. An even crummier algorithm can just say that each player's Drtg is just exactly equal to the Team Drtg.

Okay, but if you make each player's DRat = team DRat, that completely negates any concept of individual contribution to a team's defense. Why shouldn't the play of, say, Jarrett Jack (a decent defender, in the main) indicate different performance than Jose (who couldn't stop BP's stock value from skyrocketing, let alone an NBA opponent)?

Again, might I suggest you didn't do a particularly good job summing the individual possessions? And my lack of including 3FGA has hurt the accuracy of "points allowed" a little bit. It works a lot more than you'd want to think, no? Just because you tried summing DRat's instead of summing points allowed / possessions faced doesn't reflect poorly upon the method. It reflects poorly upon your time contraints/reading comprehension because I warned you of that pitfall several times.

So do you see my objection to justifying your approach only because it gives the correct Drtg (something that it does not appear to do, as of now?) It has to do more than just give me back the answer I told it.

And I think the methodological error is on your part more so than Dean Oliver's, who published this in a field for peer review with a lot of testing (commissioned by the WNBA). Again, my mathematical skills are limited by lack of education and I do have to trust people appropriately qualified both in basketball and in theoretical mathematics - both of which Dean Oliver has chops for, no?

[Truthrising]

It's basketball, not baseball, and there is more interdependence in basketball, so I don't think there is a methodology - yours or his - that explains all, and that's why teams employ scouts and not robots working on algorithms. Given the imperfect nature of advanced basketball stats, there is always going to be a gray zone called interpretation in which the objective becomes the subjective. There is no reconciling differences at the subjective level. You agree to disagree. From my point of view, Boris' entry point - ball pressure, which begins at the perimeter - is just as valid as anyone's theory that it all hinges on the center.

In basketball, indeed in sports generally, there are lots of factors that just don't fit in a spread sheet: When you star leaves the bar, whether he's had a flu shot, whether he's playing with a full deck. It's impossible to isolate one player's performance. We spend all kinds of time arguing that Bosh was a sensational player and everyone else held him back, then turn around and deny the relationship called inter-dependence in a team game. It makes zero sense.

There is very little in the first 35 pages of the Bargnani defence thread about the team. It's basically 35 pages of self-justification of one's views that we should have drafted Rudy Gay, the Overpay.

It's juts another manifestation of some Raptors' fans need to constantly reformulate their negative view of Bargnani, which from year to year keeps evolving with his game. He's not the perfect player or the dream #1 overall pick. But even when one admits that, they are not allowed to like the contribution he makes because in the eyes of the hater, there is no middle ground, there is no compromise, there must a scapegoat, always a scapegoat.

+100

People need to be objective sometimes and not always take one sided views. As mentioned numerous time he's not a franchise player and is not paid as one, but for 8.5 million/yr compared to Rudy Gay for example but sorry i'd rather take barg's contract.

[Reignman]

Attitude aside, the question is extremely valid. If individual PDSS shows Jose as the only rotation player that's below average then why does the overall defense still stink more often than not when he's not in the lineup? Especially considering PDSS shows the other players as above average individual defenders.

The answer is, "team" (uncontested action) isn't really being accounted for. 8% of the team's possessions went unaccountable to any specific player this year, and that is going to account for how things look for individuals.

I haven't looked at the formulas but as an illustration of how a small sample can skew things lets say players Alex and Paul are great one on one defenders. The offense however decides to go with pick and roll and Alex/Paul end up giving up layups 90% of the time since they don't know how to defend the pick and roll. In this case the team will look bad but the individual drtg of Alex and Paul will look good.

So using your example, if you and I are playing 2 on 2 and the opposing team scores layups 90% of the time using the PnR, can we really be called "good" individual defenders? Not in my books.

So it seems to me that PDSS would be a great formula to use when grading defenders in a game of 1 on 1 but since we're using this to measure the defensive impact in an NBA 5 vs 5 game the formula holds much less weight?