Teague is #12 in Assist/48 min in the entire NBA

[AHF]

To turn that question right back around, why in the world would anyone care about their per36 minute stats if they never get good enough to earn 36 minutes per game? If they never end up being good enough to earn a larger role, what exactly did their earlier extrapolated stats show?

Because they show how much productivity the player is providing in every minute of play. This is why Gortat got a nice contract. This is why Jermaine O'Neal averaged between 9 and 11 rebounds per 36 every season for the first 11 years of his career despite having mpg that varied from 9 to 38.

It let's you know that if you played Jermaine ONeal about 10 mpg he would give you about 2.8 rpg, if you played him 20 mpg he would give you about 5.6 rpg, if you played him 30 mpg he would give you about 8.3 rpg, etc.

That is useful information and can identify players like Chuck Hayes who are being underutilized and who can earn a larger role based on their productivity.

Every player you mentioned earned a larger role and did as well or better than their extrapolated stats predicted.

[exodus]

What about this guy in Boston?

PER 36:

15.3 ppg - 11.4 reb

Actual stats:

6.9 ppg - 5.1 reb ( 16 min/gm )

He's on his way out of the league though.

At least read the thread before trying to chime in.

I just noticed that Teague is 5th in the league in steals per 48 minutes. Granted it is a small sample size but it is encouraging.

There are plenty of examples which have shown time and again it is a reliable indicator assuming a legit sample size.

And lets not forget that Shelden's per minute rebounding numbers were pretty good when he was here.

[AHF]

What about this guy in Boston?

PER 36:

15.3 ppg - 11.4 reb

Actual stats:

6.9 ppg - 5.1 reb ( 16 min/gm )

He's on his way out of the league though.

That same player is performing slightly above career norms. He is doing slightly better than his per minute stats would tell you he would do with increased minutes. That player has averaged the following rebounds per 36 over his career:

10.3

9.5

10.8

11.4

So if you increase his minutes to 16.1 mpg (as Boston has done this season) you can expect about 5 rpg.

An 8 game sample size is not really large enough to rely on but this is a good indicator that this "anonymous" player can give you 6 and 5 on 16 minutes of play. Some coaches would value that from a low maintenance role player.

[exodus]

To turn that question right back around, why in the world would anyone care about their per36 minute stats if they never get good enough to earn 36 minutes per game? If they never end up being good enough to earn a larger role, what exactly did their earlier extrapolated stats show?

Now i know why you don't think they are a reliable indicator. you just don't understand them. It doesn't matter if it is per 48 or per 36 or per 20, the per minute numbers are the same. The reason people do per 36 is because it makes it easier to see how productive a player is relative to other players.

[niremetal]

Why would you expect someone's per minute productivity to vary when their minutes are increased? All the per minute productivity is supposed to tell you is what productivity you can reasonably expect to get out of those additional minutes based on consistent performance from the player on a per minute basis.

I'm sorry, but do you really not know the answer to your first question?

1) The more minutes someone plays, the more fatigued they become (both during each game and cumulatively over the course of a season). Most players have the stamina to put up the same numbers despite the increased fatigue. Some don't.

2) The more minutes someone plays, the higher the proportion of those minutes will be played against the other team's best players instead of against the other team's reserves (not to mention garbage time minutes). Some players rise to the occasion and produce the same numbers against starters as they would against reserves. But many don't. And I really don't see how you can argue against that with a straight face.

Most coaches realize that 1) and 2) are important considerations and recognize which players don't have the stamina/ability to play starters' minutes. And if a player doesn't show them that he has the stamina and/or skill level to log more minutes against better players, he won't give that player the shot. If you think that means nothing, then I guess you must have thought that Nazr Mohammed could have been a superstar if only a coach had played him 36 minutes per game. Because 17.5 and 12.1 are D-12-esque numbers.

That is useful information and can identify players like Chuck Hayes who are being underutilized and who can earn a larger role based on their productivity.

I never said it wasn't useful. I said it's a poor indicator. I'll modify that again to say "misleading in many cases" instead of "poor" for the sake of precision. In most cases, per36 minutes is a good predictor of what a player can become. But in many cases, it is not.

Every player you mentioned earned a larger role and did as well or better than their extrapolated stats predicted.

That is not true even on a per36 basis. And even if it were, why should I get excited if Jeff Teague is going to be the next Jacque Vaughn? After all, Vaughn was a late 1st round pick who put up per36 minute stats reminiscent of Teague's as a rookie, and his per36 minute stats didn't drop much until his 7th year in the league. That's my point when I say "why should I care?"

Edited November 10, 2009 by niremetal

[AHF]

That is not true even on a per36 basis. And even if it were, why should I get excited if Jeff Teague is going to be the next Jacque Vaughn? After all, Vaughn was a late 1st round pick who put up per36 minute stats reminiscent of Teague's as a rookie, and his per36 minute stats didn't drop much until his 7th year in the league. That's my point when I say "why should I care?"

Yes it is. Every player put up a higher PER with greater minutes and either higher pp36, ap36 or both. In any case, every one of those players did either better than expected or right about where expected based on their per minute productivity from the season before they got the increase in minutes and over the next two seasons of increased minutes.

In every case, the players performed better overall on a per minute basis with the increased minutes.

Re Jacque and Teague, Jacque Vaughn put up a 6.9 PER his rookie season.

Teague has put up a 13.9 PER so far.

How is that per minute productivity remotely comparable?

Edited November 10, 2009 by AHF

[niremetal]

Re Jacque and Teague, Jacque Vaughn put up a 6.9 PER his rookie season.

Teague has put up a 13.9 PER so far.

How is that per minute productivity remotely comparable?

I frankly don't care about PER and think it's just about the most bullsh!t thing ever dreamed up (the weighting of each stat is incredibly arbitrary, and is based on assumptions that are, in my view bogus). Their traditional stats per36 were/are very close (11.9pp36 and 7.2ap36 for Vaughn, 12.0 and 7.6 for Teague). I'm actually not a fan of traditional box score stats either, but I think they are less-bad than PER. Unless you really want to tell me that Paul Pierce was better than Steve Nash in 2005-2006 - something that anyone who looked at their stat lines would have chuckled at and anyone who actually watched their games would have died laughing at...but I won't write that essay here.

Every player put up a higher PER with greater minutes and either higher pp36, ap36 or both. In any case, every one of those players did either better than expected or right about where expected based on their per minute productivity from the season before they got the increase in minutes and over the next two seasons of increased minutes.

In every case, the players performed better overall on a per minute basis with the increased minutes.

Did you skip Nazr Mohammed? He never put up per36 stats remotely resembling the 17.5/12.1 he put up in his third year, when he averaged 15.7mpg. He certainly never put up per-game stats resembling that. Oh, and btw - his PER never got up to that level again either.

Rebraca's rookie year was his best by every measure I know of. Ekezie's solid per36 stats in his first two years didn't mean jack when he took on a "larger role" with us in the 13-69 campaign - his per36 stats that year (and, incidentally, his PER) were the worst of his career.

Oh, and about my point:

None of them ever became more than bit players.

I still want to hear you address that. Because that's the crux of my point - solid per36 minute numbers don't mean that these guys will ever be NBA starters (or even good role players). Because again - if Jeff Teague never becomes more than a bit player, then I couldn't care less what his per36 stats (and PER) are.

Actually, I've spent enough time on this. Feel free to talk amongst yourself, though.

[AHF]

I am not bothering to research every player in the NBA.

You gave 4 specific examples of players who, in your mind, made the reliability of per minute productivity numbers suspect and particularly from the point guard position.

I looked up those 4 players and saw that their performance was actually as good or better than expected in every case -- numbers which support the extrapolation of per minute productivity over larger minutes.

There are going to be exceptions to every trend but the projection of per minute productivity over different periods of time is pretty reliable.

The reason you should care for Teague is that most players improve their numbers over time from their rookie season and Teague is already performing better than Mike Bibby on a per minute basis. He isn't Jacque Vaugn who averaged 4.8 turnovers and 0.8 steals per 36 his rookie year. Teague is putting up 1.3 TO/36 and 2.7 steals per 36, in addition to marginally more points per 36 and more assists per 36.

BTW - While I understand you don't credit PER, Vaughn's PER was less than 70% of his career average his rookie year. If Teague kept up what he is doing and had a similar increase, he would be performing well above Joe Johnson's career high.

In short, there is no reason to see Teague as being a Jacque Vaugn level scrub at this point in time.

* * * *

I lied. I did look up Nazr's productivity as a UK guy. He performed well above his average during his shortened first season in Atlanta.

That said, his numbers were pretty consistent other than in scoring. He was 1 rp36 higher than his career norm, exactly on his career norm in bp36, etc. It is his scoring that was out of line. But drill down a little deeper and you can see the reason for that.

Atlanta and especially Philly gave him more FGA per minute than any other season in his career.

In 2000-01, he put up 14.6 FGA/36 which translated into 17.4 pp36.

Over Nazr's career, he put up 11.4 FGA/36 which translated into 13.5 pp36.

In other words, he was used differently those two seasons - not just his role on the team but his role within the offense. He averaged 1.19 ppFGA in 2000-01 and 1.18 ppFGA over his career. This is less than a 1% difference than what you would expect if he got similar FG opportunities over his career.

The whole premise behind the per 36 number is that you take the person's productivity and project it into the same role over normalized minutes. Nazr's productivity in 2000-01 was very slightly above what you would expect on a per minute basis. However, the pp36 difference wasn't explained by anything other than a bigger role in the offense.

The same thing would be seen if you took Caron Butler and fed him 30 shots per game or took Steve Nash and parked him in the post. When you change the player's role you expect the numbers to change on a per minute basis.

When the role stays the same but the minutes change, you expect reasonable similarity in per minute productivity (of course there will be exceptions).

But the 4 players mentioned in your original post all support the reliability of per minute productivity as applied to a player in a similar role who is getting significantly different minutes. In fact, all the factors you mentioned that would apply to increased minutes were not seen. The players did not suffer drops in per minute productivity due to fatigue or harder opponents but actually improved their per minute productivity as the minutes increased.

Edited November 10, 2009 by AHF

[niremetal]

I am not bothering to research every player in the NBA.

You gave 4 specific examples of players who, in your mind, made the reliability of per minute productivity numbers suspect and particularly from the point guard position.

Well, you said "none of the players you cited" after I had made a post where I had cited more than those 4 players. I cited Nazr, Rebraca, Ekezie, and several others. So it's not like I was bringing them up for the first time.

C-ya.

Edited November 10, 2009 by niremetal

[AHF]

Well, you said "none of the players you cited" after I had made a post where I had cited more than those 4 players. I cited Nazr, Rebraca, Ekezie, and several others. So it's not like I was bringing them up for the first time.

C-ya.

Understood but you are picking out a new set of examples to try to prove the point that per minute figures are unreliable after your first set actually made the case for their reliability. The only one I looked at was Nazr Mohammed and I don't see anything from him that casts the use of per minute productivity into doubt. I am sure there are examples where people have legitimately had large decreases in numbers when their minutes increased and their role stayed the same but those also won't mean that the per minute productivity numbers aren't a useful tool all the same. Obviously the people with the best eyes for scouting talent are useful in making draft decisions and evaluations but none of them are perfect. I look at the per minute productivity in a similar light, although it has been substantially more predictable than scouts.

For fun, however, I will look at the other examples.

(1) Nazr. Addressed above.

(2) Zelijko Rebraca. He never had a significant variation in minutes in his career, always between 14 - 16 mpg every season of his career where he played 25 or more games. He also didn't show huge variation:

Career highs: 15.7 pp36, 8.9 rbp36, 1.2 ap36, 2.2 bp36

Career averages: 14.0 pp36, 7.6 rbp36, 1.0 ap36, 1.8 bp36

The guy generally gave about what you would expect based on his mpg.

(3) Obinna Ekezie. How is he your example?

Season before bump in minutes: 12.8 pp36, 8.1 rbp36, 11.0 PER

Season after bump in minutes: 11.3 pp36, 8.8 rbp36, 10.4 PER

Career averages -----------------: 12.1 pp36, 9.1 rbp36, 11.2 PER

(4) Rodney White.

Season before bump in minutes: 15.6 pp36, 5.0 rbp36, 3.3 ap36, 2.5 sp36, .6 bp36

Season after bump in minutes: 15.0 pp36, 4.9 rbp36, 2.8 ap36, 1.0 sp36, .7 bp36

Career averages ------------------: 16.5 pp36, 5.2 rbp36, 2.5 ap36, 1.3 sp36, .7 bp36

Not much variation at all.

(5) Brian Cook.

Again, not a huge jump in minutes anywhere but I will assume you are talking about the jump from 15 mpg to 19mpg, the season Cook actually started half his games.

Season before bump in minutes: 15.2 pp36, 7.2 rbp36, 1.2 ap36, .8 sp36, .9 bp36, 14.1 PER

Season after bump in minutes: 15.1 pp36, 6.4 rbp36, 1.7 ap36, .9 sp36, .8 bp36, 15.6 PER

Career averages -----------------: 14.9 pp36, 7.0 rbp36, 1.7 ap36, .9 sp36, .9 bp36, 13.8 PER

(6) Stanislav Medvendenko

His out of line season he played a total of 39 minutes. That is not statistically significant.

All these guys performed within the range of where you would expect them to perform except the one season where Nazr's role on offense changed and then only with respect to pp36. The rest are pretty close to career norms.

If these are the most extreme examples of variations you can come up with, I am very satisfied with this cursory review of the reliability of per minute statistics.

Edited November 10, 2009 by AHF