In depth team by team previews

by **TheMC5** on Sat Sep 15, 2007 8:55 am

Shannon, I refer you to this board.

Some of the problems you cite with statistics can actually be tracked, many can be estimated, and almost all can be inferred. Things like +/-, both on the team and individual level, defensive composite scores, opponent production, weighting things like team strength and replacement player production. 82 Games has even done some extreme team tracking, which certainly involved things like effective defense, passes out of double teams, shot charting, etc.

I understand your side of the argument, too, but the reality is that statistics can effectively reflect or imply productiveness and proficiency on almost every level.

by **Christopherson** on Sat Sep 15, 2007 9:05 am

I agree with most of your argument Benji, but how do you account for the fact that more often than not, Hinrich is guarding a better player than Duhon is? Since this is the case, Duhon could be playing at the same level defensivly as Hinrich but yielding more stops, misses, turnovers, etc.

by **---** on Sat Sep 15, 2007 9:18 am

No, you only think you do. You are not acquiring anywhere near enough objective data. You are taking your opinion, and reinforcing it by seeing what you want to see. You have a handful of select plays you remember of Hinrich playing good defense, out of around 3,000 defensive possessions by his man over the last three years..

I won't have as much as the statistic data, but while I may only recall a handful of good defensive plays, there are others I have forgotten that still contributed to my opinion that Kirk is a great defender. It's not just what you remember that forms your opinion.

Everytime Kirk Hinrich makes a defensive play you are placing it into subjective values: Good defense, or bad defense. Instead of gathering the data and then evaluating it, you are evaluating it as you are measuring it. Injecting an systematic measurement error to the data that can only grow larger with each input.

To be honest I don't exactly understand this post. Are you saying that when I watch games, I naturally have two catagories - bad defence and good defence?

If that's what you're saying, that's completely wrong. As I've been trying to point out, a play like "Scenario A" I would hold higher than "Scenario B" even though good defence was played in both.

But the stats record what matters, whether or not points are allowed.

Exactly, which shows statistics don't (and simply cannot) go in depth enough into what defence is and how it is played by different players.

No, it doesn't. What matters is whether or not the points are denied. Not that how you did it appears better.

How you denied those points is much more important for analysis than just "Did he deny those points?"

That way, Kirk could have a play like "Scenario A", and Duhon could play atrocious defence and the defender just misses the shot - and noth defensive plays would have equal statistical value in terms of rating a players defensive performance, which is wrong.

I removed the entire other player scoring. The data would show he played bad defense, BECAUSE HE DID, his man scored. Good defense = man not scoring, bad defense = man scoring. It is irrelevant if he gives it his all and does everything he can, the guy still scored. He does not get points for effort.

Are you saying that if you absolutely smother a guy with the best defence you've ever seen in your life and he manages to throw up a wild show and it somehow drops, you played bad defence? If that guy put a tiny bit less power on his shot he wuld have airballed it and the defender would of been praised for doing the exact same thing he was doing when it did drop.

Unless, whoever gets the ball scores. Then the first one is always better, as you have around a 70% chance of denying a point.

I forgot to mention that both scenarios result in a miss, which statistics will regard as equal, when they aren't.

But in both cases, the result is the same. He missed, points were denied.

It's not so clear cut. Sure, the results are the same, but the player obviously played better defence in the first scenario - which statistics wouldn't pick up.

Since I have some data, we don't have to say they were not scored in every single one. Which would mean they were equal in results, the rest is just evaluating how they looked while they did it. Which is irrelevant.

I disagree.

If all the results are equal, you look at how each player did it. If he played decent defence and the guy missed - then good, he did his job. But he didn't do his job to the level that the other player did when he played better defence by forcing a fadeaway at the shot clock buzzer.

(And although I prefer to count stops related to the team, not individuals, I am doing so here, I will do the other in an aside later.) Hinrich's more impressive defensive plays last year resulted in a stop on 48.9% of his counterparts possessions, Duhon's solid defense resulted in 53.4% of counterparts possessions being stops. Over 200 possessions, Hinrich is getting 98 stops, and Duhon is getting 107 stops. Assuming just two points per made basket, Hinrich is giving up 204 points, Duhon is giving up 186 points. (Using, the team relation, which I prefer. Hinrich got a stop on 12.26% of Bulls opponents possessions, Duhon on 13.20%, over 200 possessions, Hinrich racked up 25 stops and Duhon got 26. On 200 shots, Hinrich forces 17 misses, Duhon 16. But on 200 possessions, Hinrich forces just 7 turnovers to Duhon's 10)

And once again, if he stopped his man, how he did it is very important in measuring defensive performance.

You can't just say Duhon stopped his man 53.4% of the time and Hinrich stopped his man 48.9% of the time and say Duhon is a better defender. Look at how they defended. Did Duhon get some lucky bounces in his favor? Was Duhon going against lesser players in not-so-close games?

Statistics only look at the end result, it's like putting a 1 point win and a 60 point blowout in the same catagory and calling them even - they don't tell the full story.

EDIT:

Shannon, I refer you to this board.

Some of the problems you cite with statistics can actually be tracked, many can be estimated, and almost all can be inferred. Things like +/-, both on the team and individual level, defensive composite scores, opponent production, weighting things like team strength and replacement player production. 82 Games has even done some extreme team tracking, which certainly involved things like effective defense, passes out of double teams, shot charting, etc.

I understand your side of the argument, too, but the reality is that statistics can effectively reflect or imply productiveness and proficiency on almost every level.

I don't disregard statistics altogether. I know that stats are a very good way to look at how a player performs. But defence is alot harder to track and analyse than offence.

I don't agree with putting a good defender in Chris Duhon over an All-NBA caliber defensive player in Kirk Hinrich because of a slight statistical edge. I know Benji said he watched them both and his opinion is that Duhon is better, but I can't help but assume he thinks that way because of what he see's in stats - just like how he says I don't watch Hinrich objectively and have a slight bias towards him because I think he is the better defender.

by **benji** on Sat Sep 15, 2007 11:03 am

I agree with most of your argument Benji, but how do you account for the fact that more often than not, Hinrich is guarding a better player than Duhon is? Since this is the case, Duhon could be playing at the same level defensivly as Hinrich but yielding more stops, misses, turnovers, etc.

Well, I would not recommend looking at it, but that DCS guy accounted for time against starters (as well as other things) and still has Duhon out ahead. Take that for what you will.

I know Benji said he watched them both and his opinion is that Duhon is better, but I can't help but assume he thinks that way because of what he see's in stats - just like how he says I don't watch Hinrich objectively and have a slight bias towards him because I think he is the better defender.

I said that because of how fucking illogical your argument is. It was supposed to show you just how irrational you and Indy (and how many others) are being in claiming that because you SAW Hinrich appear better, he is.

I don't agree with putting a good defender in Chris Duhon over an All-NBA caliber defensive player in Kirk Hinrich because of a slight statistical edge.

I would say Kirk Hinrich is hardly a All-Defense team player. Wade, Iverson, Davis, Iguodala, Kobe, Paul, Andre Miller, Kidd, and Larry Hughes all racked up more stops per game for example. And most of those guys would beat him in total stops as well.

See what you are doing here? You are saying that because you have decided Kirk Hinrich is All-Defense (which is an opinion anyway) that it cannot be that Duhon could possibly be better.

I don't disregard statistics altogether.

seeing a player beats stats 10 times out of 10.

And once again, if he stopped his man, how he did it is very important in measuring defensive performance.

A good defense player stops his man more than a bad defensive player. How he appears when he does it is irrelevant.

You can't just say Duhon stopped his man 53.4% of the time and Hinrich stopped his man 48.9% of the time and say Duhon is a better defender.

I did not. But you are the one saying that because Hinrich appears to be better to you, that the available information that Duhon was more efficient on defense must be wrong and that there is no way Duhon could possibly be better. The data cannot be right, it doesn't agree with your opinion or what you have seen!

If all the results are equal, you look at how each player did it. If he played decent defence and the guy missed - then good, he did his job. But he didn't do his job to the level that the other player did when he played better defence by forcing a fadeaway at the shot clock buzzer.

His job is to prevent a made basket. Both players did their job equally well, they both prevented made baskets.

Sure, the results are the same, but the player obviously played better defence in the first scenario - which statistics wouldn't pick up.

How did he obviously do better? Because he moved more? Because he looked better doing it?

Are you saying that if you absolutely smother a guy with the best defence you've ever seen in your life and he manages to throw up a wild show and it somehow drops, you played bad defence?

You gave up a basket. Therefore, you failed on defense.

Actually you know what. Fuck all these singular examples, I am done with them. We are not talking about a single example. We want a major sample size, the bigger the better. One possession is meaningless.

Exactly, which shows statistics don't (and simply cannot) go in depth enough into what defence is

Defense is not allowing points. Points are a statistic. Defense measured.

To be honest I don't exactly understand this post. Are you saying that when I watch games, I naturally have two catagories - bad defence and good defence?

If that's what you're saying, that's completely wrong. As I've been trying to point out, a play like "Scenario A" I would hold higher than "Scenario B" even though good defence was played in both.

Fine, that is even more subjective and worse. There will be even more errors in your data.

I won't have as much as the statistic data, but while I may only recall a handful of good defensive plays, there are others I have forgotten that still contributed to my opinion that Kirk is a great defender. It's not just what you remember that forms your opinion.

You have to somehow quantify the data you have collected to decide Hinrich is the second best defender in the league. So you are gathering all this faulty data while watching then games. Then you forget some of that faulty data and only retain only bits. Then you are fitting the faulty data you have left, into a predetermined conclusion. Then taking this conclusion and the scraps of bad data left, and comparing it to your conclusions and scraps on other players, and reaching another conclusion. It is like dismissing all of Kobe's games except his 81 point game and declaring him best ever.

Think about this, you have absolutely no way to say Hinrich is better than Rondo or Baron Davis other than saying, "well, when I watch the games he appears to be better" even though you cannot compare all his possessions instantly to those two guys possessions. Your only evidence, is your own opinion. So you have no evidence.

Those of us who are not emperical-research adverse, can say, "hey, those guys are good defenders. I wonder how good?" Then we can look at the data and see that Hinrich gets a steal on 1.76% of opponents possessions, Rondo 3.57% and Davis 2.86%. We can see that Rondo forces a turnover on 4.2%, Davis on 3.8% and Hinrich on 3.7%. Force missed shots on 8.2%, 8.2% and 8.4%. And make stops on 14.4%, 13.4%, and 12.3%. We can say, well, Rondo looks like the best defender, but Davis and Hinrich played more minutes, who then had the bigger impact on the season? Well, Davis had 10 stops a game, Hinrich had 8.7 and Rondo had 6.6. But Hinrich played more games because Davis was injured so he beats out davis 697 total stops to 630. What's our conclusion then? If they played the same minutes, Rondo. Same games, Davis. Otherwise, Hinrich had the biggest defensive impact.

Now we can argue actual things with gobs of data in front of us. Someone can argue Davis was the best, and would have been if he had not had to sit out games. While someone else could argue that Hinrich because he played 17 more games, was more useful. The question of what constitutes "best" and which data points are most important is the argument. No longer is the argument "well I saw Hinrich do this and this" "well, I saw Steve Nash do this and this, and did not see Hinrich do this" We have actual quantifiable data instead of just "impressions" of the player.

by **---** on Sat Sep 15, 2007 11:18 am

Benji, could I get a link to all this data? I wanna check out the numbers.

I see what your point is, but I don't agree with it. Defence is not just a matter of wether the man you're gaurding scores or not. It's how you defend. You already let yourself down once if he managed to get the ball and twice if he manages to shoot it.

I you don't deny your ma, you're playing bad defence. Stats don't pick that up.

Statistics are not everything. I bet I could go through and find a bunch of not-so-great defenders up near the top when ranked statistically. That happens because statistics cannot possibly measure everything. If you watch two guys and one is much more impressive on how he denies his man the ball, rotates, get's position, pushes guys out of position, etc. but still has the same result on wether or not the ball goes in or forces a turnover, statistics hold them equal.

Nothing can compare to actually watching a player.

by **Indy** on Sat Sep 15, 2007 12:09 pm

Point Guard defense cannot be measured by any stat. Here are the 3 most important things for defending at the point guard position.

1.) Pressuring your man, wearing him out, and making it harder for him to get the offense moving later in the game because of fatigue.

2.) NOT gambling and going for steals. Point guards need to be solid and trustworthy, point guards that get a lot of steals are usually the ones that can be backed down or driven by easily.

3.) Starting the entire defensive rotation. After the point guard makes his first pass, your first step as the man guarding that player determines the rest of the play.

Defense is a team accomplishment. Everyone needs to do their job for a defense to succeed. That doesn't always happen, and the stats will always be skewed because of it. So the only way to measure it is by watching and analyzing.

Also, of course Kirk is going to give up more points then Chris Duhon, that's because Kirk is head and shoulders the better defender so he's going to guard the better player.

by **benji** on Sat Sep 15, 2007 12:25 pm

Nothing can compare to actually watching a player.

Right, nothing is as irrational.

I bet I could go through and find a bunch of not-so-great defenders up near the top when ranked statistically.

You will find players who are "not-so-great" in YOUR OPINION. Which is not fact, nor reality. If you disagree with the data, that does not make your opinion become fact.

I see what your point is, but I don't agree with it. Defence is not just a matter of wether the man you're gaurding scores or not.

I care about winning games. You win games by outscoring your opponent, not by looking better than they do.

Benji, could I get a link to all this data? I wanna check out the numbers.

MC5 already linked to 82games where some of the raw data comes from, the rest comes from nba.com. (Because I assume they have correct stats at nba.com.)

Indy wrote:Point Guard defense cannot be measured by any stat.

Wrong.

So the only way to measure it is by watching and analyzing.

Even more wrong. (However, also right. As all stats are derived from someone watching and recording what happens.)

Kirk is head and shoulders the better defender

Prove it.

Come on. Just try and prove to everyone Kirk Hinrich is the second best defensive player in the league without using stats.

by **Indy** on Sat Sep 15, 2007 12:33 pm

I never said Kirk was the second best defensive player in the league. I said he's the second best point guard defender.

It is now clear to me that you don't know what that means.

But I can't let you get away with responding to my posts by just saying wrong. Give some evidence. Show me some stats that measure the 3 things I listed.

by **rl564411** on Sat Sep 15, 2007 12:52 pm

Can I see a Washington Wizards one please?

by **benji** on Sat Sep 15, 2007 3:14 pm

Indy wrote:I never said Kirk was the second best defensive player in the league. I said he's the second best point guard defender.

I figured I didn't need to constantly say point guard as we were clearly only talking about point guards.

But I can't let you get away with responding to my posts by just saying wrong. Give some evidence.

Of Hinrich not being the second best defender in the league? I did.

Show me some stats that measure the 3 things I listed.

The results of it are in the data I provided before.

Those are what you think are the three most important things.

The BoP is on you:

Just try and prove to everyone Kirk Hinrich is the second best defensive [point guard, since you are looking for anything to get out of it] in the league without using stats.

I'm also interested in your explanation for why you said the Bulls were a slow down team, when they clearly are not.

by **Dean** on Sat Sep 15, 2007 4:06 pm

I'm also interested in your explanation for why you said the Bulls were a slow down team, when they clearly are not.

Well its not like theyre the Phoenix Suns.. They play a half court game, move the ball around and slash alot without the ball

by **Indy** on Sat Sep 15, 2007 5:29 pm

The Bulls get a lot of easy points that's generated from their full court press defense. When they aren't pressing full court, they are at least at half court or three quarters court.

However, their offensive strategy is based on a half court offense that relys on ball movement and slashing. They are not a run and gun team as other teams that have small ball lineups are.

by **benji** on Sat Sep 15, 2007 5:39 pm

You implied they were a slow down team in the manner of the Spurs. (And you also said Jazz, who were 15th, or average.) They are the opposite, only five teams had a faster pace last season.

by **TheMC5** on Sat Sep 15, 2007 6:41 pm

Indy, a good defensive stat that doesn't just measure points scored against player X by player Y, is on-court/off-court team +/-. You can even look it up with specific combinations of players. How does that not quantify individual contributions to team defense? And it is just one of many metrics in this manner.

And I don't think anyone, at least not myself, is advocating that stats can tell the entire story and makes watching games moot. There are quite a number of stats for some players that might look weird or contrary to what we see subjectively on the court, but a further investigation into the circumstances that produced the odd statistical behavior generally yields a solid explanation, one that I would assume more often than not confirms our suspicions. The main use of stats is to attempt to corroborate what we think we know, and to try to dispel common misconceptions.

If a player, for example, is renowned defensively, but shows up as absolutely terrible by almost every defensive metric, shouldn't that warrant a re-evaluation of the consensus regarding that player? If the data shows, for this hypothetical player, that his man scores well above average against him, his on/off numbers show the team to perform better with him on the bench, and this has all been weighted to account for strength of opponent and the player's replacement in the lineup, how could that be dismissed off-hand?

by **Sauru** on Sun Sep 16, 2007 11:30 am

this thread should be good, we got the "stats rule all" person in benji going up against the "overall picture" of indy(note i am with indy on this one as there is far more to the game than the stats that are shown). anyway i hope this argument continues, oh and i wanna see your thoughts on the celtics

by **benji** on Sun Sep 16, 2007 11:57 am

Sauru wrote:this thread should be good, we got the "stats rule all" person in benji going up against the "overall picture" of indy(note i am with indy on this one as there is far more to the game than the stats that are shown).

Um, no. Indy's position is "stats are meaningless, only my own personal observations matter" while mine is, "having solid data allows us to better analyze teams and players."

The opposite position could never be called "overall picture" as they are dismissing any data that runs counter to their own viewpoint. My position accepts all data initially, and evaluates it after collecting it.

by **jonthefon** on Sun Sep 16, 2007 3:06 pm

Pfft, I don't think it matters. Either way, the Bulls have two great defenders at point guard

by **JT_55** on Sun Sep 16, 2007 3:50 pm

^ Except this has turned into the age-old "stats vs. opinions" battle instead of comparing the defenders.

This is pretty interesting, the huge quote wars, but bottom line, stats can give a general idea of how good a player is, but it doesn't tell the whole story. Unless I missed something, the only stat in basketball (and maybe even all sports) that is 100% accurate is FT %, cause no outside factors (other than away crowd antics)contribute.

Now if I have enough time maybe I'll read all the posts in this dreadfully long topic; I guess looking forward to Indy's next update will be better.

by **J@3** on Sun Sep 16, 2007 4:14 pm

the only stat in basketball (and maybe even all sports) that is 100% accurate is FT %

well really I think they are all 100% accurate. If you score 2 points, you are given 2 points statistically... if you grab a rebound, you are given a rebound statistically etc.

by **benji** on Sun Sep 16, 2007 4:15 pm

JT_55 wrote:the only stat in basketball (and maybe even all sports) that is 100% accurate is FT %

All stats are 100% accurate. If Kobe Bryant scores 28 points, he scored 28 points. He did not score 27, or 29, he scored 28. That is 100% accurate is it not?

Except this has turned into the age-old "stats vs. opinions" battle instead of comparing the defenders.

It's only age old because some people continue to be on the opinion side of "facts vs. opinions" and refuse to even acknowledge counterpoints that question their beliefs.

the huge quote wars...dreadfully long topic;

Waaannh, keep the topics short and timid please! No arguments, just agreement. :cry:

by **JT_55** on Sun Sep 16, 2007 4:41 pm

Jae wrote:
the only stat in basketball (and maybe even all sports) that is 100% accurate is FT %

well really I think they are all 100% accurate. If you score 2 points, you are given 2 points statistically... if you grab a rebound, you are given a rebound statistically etc.

benji wrote:
JT_55 wrote: the only stat in basketball (and maybe even all sports) that is 100% accurate is FT %

All stats are 100% accurate. If Kobe Bryant scores 28 points, he scored 28 points. He did not score 27, or 29, he scored 28. That is 100% accurate is it not?

Aw, c'mon guys. You know what I mean. Okay, I should've said FT % was the only stat that 100% truly reflects on the person's skill on the subject.

Note that I too am also contributing to this topic's dreadful length

by **benji** on Sun Sep 16, 2007 4:47 pm

Length is good...short threads and posts are lame.

Does FT% really reflect the persons skill? Shaq is said to nail 75-80% in practice.

by **tempo** on Sun Sep 16, 2007 8:12 pm

^^really? i also hear ben wallace hits 50% of his 3's in practice! :roll:

assuming a players skill is based purley on stats is complete ludacris. bruce bowen dosent average a steal of block per game but he has been one of the premier defenders in the league for years, you would only ever garner this information having actually watched him in a game.

to suggest you can evaluate every player and every game from a box score is stupid, its the combination of stats and in actual game analysis that one could arive at a conclusion.

by **grusom** on Sun Sep 16, 2007 8:53 pm

tempo wrote:
to suggest you can evaluate every player and every game from a box score is stupid, its the combination of stats and in actual game analysis that one could arive at a conclusion.

Is this the first time you read a benji post? :-)

by **J@3** on Sun Sep 16, 2007 9:08 pm

assuming a players skill is based purley on stats is complete ludacris. bruce bowen dosent average a steal of block per game but he has been one of the premier defenders in the league for years, you would only ever garner this information having actually watched him in a game.

I don't buy into the mass stats thing either, but this comment is just stupid. Have you not read what is being posted? It's not based on just steals or blocks. I can't think of anyone who would use those as a pure statistical measure of how good a defender is, statistics have progressed ALOT over the last few years.

In depth team by team previews

Who is online