Totally agree. That is exactly the reason why Bill bashed on rox like that, coz his needy baby boy was F* by our players badly in an earlier game, so papa came out with a sh!ty mouth. It was funny, though, when barry hit him with rox defense stats, he was like embarrassed like an idiot, not knowing what to say no more ... then he had to go like "I saw it with my own blind eyes, blah blah blah".
wnes - pts allowed per possession is nothing more than a rate that INCLUDES possessions per game. You can't get pts allowed per possession without knowing possessions per game. They aren't separate. If you understood that FG% and PPG weren't correlated then why did you say that FG% and PPG of the Rockets are enough indicators to show that the Rockets defense are qualitatively better than other teams? No, using PPG and FG% is misleading, not the other way around. Look at the stats. Memphis gave up 88.5 PPG while the Spurs gave up 88.8 PPG - BUT the Spurs acheived this even with their opponents having the ball 2 more times per game than the Grizzlies opponents and even with the Grizzlies denying their opponents with a lower FG%. Who had the better defense? The Spurs, not the Grizzlies because the Spurs opponents had the ball at least 2 more times than the Grizzlies allowed their opponents YET the Spurs somehow managed to only allow about the same number of points per game as the Grizzlies. And the Spurs did this even with the Grizzlies allowing a lower FG% than the Spurs. Comparing MEM and SAN based on a per possession basis draws this out. SAN defensively last year allowed 100.9 pts per 100 possessions while MEM allowed 103.6 pts per 100 possessions. That's why the points allowed per possession is a better measure - because it factors in pace,fg%,ft,pts allowed, to's, and rebounds into the equation. And if you're not happy with looking at points per possession, look at the pts-per-shot stat. We find that SAN allowed fewer pts per shot (1.13) than MEM (1.18) (which you would think otherwise if you looked only at FG% allowed) which is yet another indicator that the Spurs defense was better than the Grizzlies. Looking at PPG and FG% is misleading because you miss all the above. And none of this is even looking at the offensive side of the equation.
I don't think the definition of a great defensive team should be an ability to win even though they suck offensively. We're talking about defense only...not the overall strength of the team. That's the point. As it turns out, it is probably a difficult argument to make for defense, because teams that tend to be bad, or even mediocre, are generally better offensively than defensively anyway. But I hold that the reverse argument is also true. Offensively, though per possession statistics remain best, they too are flawed by the fact that they normalize pace. Pace is not something that should be normalized...all teams do not play at the same pace, shouldn't pretend like they do. Like I said, I imagine that the Suns are a top offensive team no matter how you look at it - PPG, FG%, Points/Possession. But I think they are all important variables. I think a team can be middle of the pack on a per possession basis but still be a good offensive team because of the pace they force. If a team has a middle of the pack FG% and middle of the pack point/possession, but is still a top 5 team in PPG because they create a lot of possessions, in my opinion, that is still a good offensive team.
But it ignores the key fact that the Grizzlies hold their opponent to less points then the Spurs. Again, I'm not saying that per possession basis isn't the most complete way to look at it...but it does effectively "ignore" pace, a factor that is part of a team's strategy. On as simple a basis as possible, the point of defense is to hold your opponents to as few points as possible, not to hold your opponent to as few points on a per possession basis as possible. In your example above, the Spurs allowing their opponent to have 2 extra possession per game has hurt them in allowing them to give up more points. Fortunately, offense does matter in basketball, though, and the Spurs overall faster pace may have helped them overall as a team. The point is, there is no one key stat. Some are better than others, but you need to see them all to get a good sense of a team's abilities, offensively, defensively and overall.
Pace in itself doesn't determine # of points you allow an opponent. You missed my post above relating to this season's pace. The Portland Trailblazers are #1 in terms of number of possessions they allow their opponents at 86.5 possessions per game. Yet they are allowing 115 pts per 100 possessions which ranks them #27th defensively. The converse also applies. Allowing more possessions doesn't mean your defense sucks. That's why the Bulls this year and the Suns defense would be undervalued if you looked at the PPG. In the above example the Spurs allowing 2 more possessions hasn't hurt them one bit. That's the point with using a rate-based comparison. And pace IS factored in as a part of the points-per-possession. It's not being removed. It's just being converted into a rate so that you can have apples to apples comparisons. Like I mentioned above as well. If you don't like pts-per-possession, take a look at the pts-per-shot allowed stat that I mentioned above. Even on a per-shot basis the Spurs did a better job than Memphis. That means for every shot taken by their opponent the Spurs allowed fewer points than the Grizzlies did. If that's not a measure of defense then I don't know what it is .
I could not believe it when Walton started making these comments. I've been annoyed with stuff he's said before, but this was just ridiculous. To me, the worst part of it all was when JB told him that we were number 1 in all the stats, and asked "How can you say that they aren't the best? They're number one!" Bill Walton's response: "Yes, but how many points do they score?" What does that have to do with defense, Bill???? I seriously don't understand that point at all.
His point was about the pace of the game.....I see where he is coming from but Stats are not based upon pace .....so, the Rockets are the best defensive team in the league...especially with Deke in there. DD
Yes, I know. But you agree that a great defensive should win more easily if they suck offensively compared to a team that's average defensively and also sucks offensively. Right? That's the purpose of a defense. It gives the team more slack on the offensive end. If you play terrible defense, you need to be play great offensive to make up for it. If you play great defense, your offense doesn't need to be as good. Pace is a team characteristic, just like how often they shoot three pointers, or how often they get fast break points, or how many assists they get. Yes, how a team performs can depend on pace (just as it depends on these other things). If you want to look at points per game, is it not also obscuring information like 3-point attempts and assists? No matter what number you come up with, some information about the team is going to be obscured. I wouldn't call that a flaw. As a first pass, you say that Houston was the best defensive team (based on points/poss). Then, you can delve further and look at how they achieved that. That includes the pace they like to play at, how often they give up 3-pointers versus 2-point shots, how often they allow free throws, fast breaks, etc. This doesn't make sense to me. I think you'r defining "good offense" in a somewhat arbitrary manner. A good offense, just like a good defense, should help you win games. It should somehow relieve pressure on what you need to accomplish on the other end of the court. What you describe is a team that's not scoring a lot per possession, but because they play at a high pace they can put a lot of points on the board. Well, that means the other team is also going to have a lot of possessions with which to score, and all they have to do is have average efficiency and they can match this "good offensive team". To me, that's not good offense. It's average offense, played at a fast pace. A team like the Dallas Mavericks, who play excellent offense at a very slow pace (let's pretend they have average defense) would easily outscore such a team, even if they generally don't score as much per game. The reason is because now they get to run that extremely effective offense, but they're getting a ton of extra possessions.
Tango, why would points/shot be an accurate indicator of defense? It ignores free throws and turnovers etc...I think points per possession is accurate, I'm surprised the NBA analysts don't use this stat
One last time on this topic for me. I feel like I'm just repeating myself so let me try a different way to explain the difference between a PPG measure of offense/defense vs a per-possession method. Which car is faster? Car A that reaches 120 MPH after 1000 ft or Car B that reaches 109 MPH after 800 ft? It looks like Car A (120mph) is faster than Car B (109mph). But convert that into a rate and you quickly realize that Car B is actually faster than Car A. Car B: hits 0-60mph in 5.5 seconds (Porsche 911) Car A: hits 0-60mph in 6 seconds (Porsche Boxster) Now let's normalize the distance travelled to both the same 1000 ft. We run the math and we find that if both cars travelled up to 1000 ft the projected final speeds would have been: Car A: 120 MPH @ 1000 ft Car B: 125 MPH @ 1000 ft That's what we're doing with rate based stat like per-possession or per-shot. It's like comparing the actual rate of acceleration between the cars in this example and normalizing the distance travelled vs using the final velocity reached without doing so. The PPG metric is like comparing the final speeds of the cars with different distances traveled which would mislead you regarding how fast the cars really are. Hope that helps! I agree with you whole-heartedly . I was trying to demonstrate to some of the others why rate-based measures instead of using stats based on totals is preferable so I tried a different stat (pts/shot) if the whole possessions issue was tripping folks up.
For the last 5 seasons, not counting the current one, I calculated "normalized" offensive efficiency, defensive efficiency, points scored per game, and points allowed per game for every team. Then I determined how all these metrics correlated with eachother. What I want to look at is how each of these measurements correlates to the team's winning percentage. Intuitively, we can say that "good offense" and "good defense" should correlate well with winning%. Furthermore, offense and defense should be pretty much independent of eachother. How good your offense is shouldn't depend on how good your defense is, and vice versa. Good teams are usually good at both, bad teams are bad both, and middling teams might have some combination. Here were the results: Code: Off/poss Off/g Def/poss Def/g Winning% ----------------------------------------------------- Off/poss | 1 Off/g | 0.834 1 Def/poss | [B]-0.066[/B] -0.159 1 Def/g | -0.168 [B]-0.522[/B] 0.823 1 Winning% | [B]0.676 0.484 0.644 0.454 [/B] 1 Notice (last line) that Off/poss correlates better with winning that Off/g, as does Def/poss to Def/g. Secondly, observe the correlation between Off/poss and Def/poss compared to how Off/g correlates to Def/g. Between Off/poss and Def/poss, it's almost zero (-0.066), but between Off/g and Def/f there's a noticeable negative correlation (-0.522). So, if you only focus on per game numbers, "good offensive teams" will tend to be "bad defensive teams", just as "good defensive teams" will tend to be "bad offensive teams". If you look at per possessions numbers, however, that tendency isn't there -- how good you are offensively doesn't depend on how good you are defensively. Basically, if you depend on per game numbers to judge an offense/defense, you're really not getting a pure evaluation.
I think another thing people should realize is that even points/possession is not 100% accurate in showing defensive effeciency because strength of schedules have been very different for teams. I'm sure most Easter conf teams have had to defend against offensively weak teams for example, so we know their points allowed/possession would be deflated in comparison to teams like the Rockets.
Certainly. It's not adjusting for strength of the opponent. But which of the conventional defensive metrics do that? I'm just saying if you want to look at a number that indicates how good a team is defensively, it makes little sense to look at simply FG% or points allowed per game. Calculating points allowed per possessions isn't terribly complicated (possessions roughly equals FGA + TO + 0.4*FTA - ORB) and it tells you far more about how effective a team is on the defensive end. But over the course of an entire season, even in the currently skewed league, I don't think that adjustment would make much of a difference.
I'd like to quote Dean Smith from his coaching book, "Multiple Offense, Multiple Defense." This is from the introduction to that book: I think Dean Smith knows what he's talking about when it comes to evaluating what wins in basketball. He's been using the per possessions concept since the 1950s! For whatever reason, it's never caught on in the mainstream, but most coaches intuitively think of the game in those terms.
Thanks. I used the numbers from basketball-reference.com this time (which also includes team turnovers) and added pace to the analysis. It's still true, even if you look at per possession data, that good offensive teams play at a somewhat faster pace and good defensive teams play at a somewhat lower pace. But it's not a strict rule, and clearly the correlation isn't as high as it would be for the per game stats. Here's the correlations table, season by season, for the last 5 years: Total (last five seasons) Code: off/poss def/poss off/g def/g pace win% ------------------------------------------------------- off/poss 1.000 def/poss -0.039 1.000 off/g 0.846 -0.158 1.000 def/g -0.173 0.831 -0.522 1.000 pace 0.224 -0.202 0.697 -0.701 1.000 win% 0.684 0.641 0.490 0.441 0.005 1.000 2006 Code: off/poss def/poss off/g def/g pace win% ------------------------------------------------------- off/poss 1.000 def/poss -0.034 1.000 off/g 0.776 -0.171 1.000 def/g -0.165 0.805 -0.594 1.000 pace 0.167 -0.171 0.747 -0.716 1.000 win% 0.587 0.663 0.292 0.516 -0.122 1.000 2005 Code: off/poss def/poss off/g def/g pace win% ------------------------------------------------------- off/poss 1.000 def/poss -0.001 1.000 off/g 0.816 -0.088 1.000 def/g -0.138 0.776 -0.529 1.000 pace 0.202 -0.141 0.723 -0.726 1.000 win% 0.723 0.658 0.559 0.384 0.097 1.000 2004 Code: off/poss def/poss off/g def/g pace win% ------------------------------------------------------- off/poss 1.000 def/poss -0.240 1.000 off/g 0.880 -0.427 1.000 def/g -0.319 0.918 -0.632 1.000 pace 0.332 -0.438 0.731 -0.755 1.000 win% 0.557 0.642 0.287 0.545 -0.181 1.000 2003 Code: off/poss def/poss off/g def/g pace win% ------------------------------------------------------- off/poss 1.000 def/poss -0.013 1.000 off/g 0.872 -0.144 1.000 def/g -0.118 0.844 -0.458 1.000 pace 0.129 -0.243 0.592 -0.718 1.000 win% 0.785 0.575 0.628 0.377 -0.003 1.000 2002 Code: off/poss def/poss off/g def/g pace win% ------------------------------------------------------- off/poss 1.000 def/poss 0.092 1.000 off/g 0.888 0.039 1.000 def/g -0.127 0.815 -0.393 1.000 pace 0.292 -0.021 0.691 -0.587 1.000 win% 0.765 0.673 0.669 0.395 0.221 1.000
Exactly. Per possession stats are almost undoubtedly some of the most useful. But I think you have to look at a lot of statistics, in combination. You can start with per possession and delve further, or start with FG% and delve further or whatever. I think this is our main area of disagreement. You're talking about the point of a good offense, or a good defense. I'm talking about it in isolation. In isolation, I think team's can be good offensively, or defensively, and still suck, because they've ignored other aspects of the game. Why look at something in isolation? I don't know. To get a measure of a team's strengths and weaknesses. If all we're trying to do is measure how good a team is overall, then you can either (1) look at standings, or (2) look at point differential.
Here's the difference. FG% only describes a specific aspect of a team's defense, where as points allowed per possession encapsulates pretty much everything a team is trying to achieve on the defensive end. The ultimate goal for any team's defense is to limit the opposing team's points per possession. How they go about doing that could depend on pace or other strategies (man-to-man or zone, how they defend pick and roll, etc.). But the ultimate goal is to limit the other team's per-possession efficiency. There really is only two aspects to the game -- offense and defense (the per-possession metrics I'm talking about incorporates rebounding into these). If you're a very good offensive team or a very good defensive team, you'll probably be a good team overall. Not necessarily, but there's a good chance you'll be pretty good. Take a look at some of my posts above dealing with how these different measurements correlate with eachother. Perhaps that will convince you. Exactly. But what you're doing is not looking at things in isolation. You're doing the opposite. By considering per-game stats over per-possession stats, you're actually including factors that have nothing to do with good defense. That's why, as I've shown above: (1) per game offense and per game defense tend to be correlated (i.e. not independent), and (2) per game offense/defense does not correlate as well with winning as per possession offense/defense.
