Hornets waive Mensah-Bonsu

[Chaplin]

Wow, now I see why you don't like per time unit stats... trying to put minutes from various games into blocks of 48 or whatever.

The bold part is quite wrong. You can use minutes per game and, for example, reb per 48 min, to compute reb per game exactly. Just multiply 'reb per 48' times 'minutes per game' and divide by 48. Actually try it once and you'll see for yourself. Using minutes per game, you can easily move back and forth between per 48/40/36 stats and per game stats.

I wonder if this is Chaplin's problem too. He talks about high per48 stats and terrible per game stats. There's nothing 'terrible' about them, they are exactly what you expect given the guy's minutes played per game. At least they are what you should expect... expecting anything else being rather silly.

The problem is having people here tout signing no-name scrubs simply because they can pull down a couple rebounds in garbage time of meaningless games. That alone makes per48 stats useless. The problem isn't the stats themselves, it's using them to make players SEEM a lot better than they really are. A guy may pull down 15 rebounds per 48 minutes, but he's NEVER going to play 48 minutes because he simply isn't good enough to play against good players. How is that so hard to understand?

 

[asudevil83]

The problem is having people here tout signing no-name scrubs simply because they can pull down a couple rebounds in garbage time of meaningless games. That alone makes per48 stats useless. The problem isn't the stats themselves, it's using them to make players SEEM a lot better than they really are. A guy may pull down 15 rebounds per 48 minutes, but he's NEVER going to play 48 minutes because he simply isn't good enough to play against good players. How is that so hard to understand?

or he'll be averaging 5 mins of garbage time throughout the year, and you'll take his numbers and give them a per 40 or 48 minutes and think "well he'll do just as well playing regular minutes." well that 40-48 mins is just a games worth of garbage time, which isnt at all realistic.

hell, Barron's per48 is close to 10/10/1. but he sucks.

Lopez's per48 is almost 12 boards per game, but he's a terrible rebounder.

 

[Errntknght]

The problem is having people here tout signing no-name scrubs simply because they can pull down a couple rebounds in garbage time of meaningless games. That alone makes per48 stats useless. The problem isn't the stats themselves, it's using them to make players SEEM a lot better than they really are. A guy may pull down 15 rebounds per 48 minutes, but he's NEVER going to play 48 minutes because he simply isn't good enough to play against good players. How is that so hard to understand?

I understand that there are people who use/misuse all kinds of stats with meager grasp of their meaning. On that basis, I suppose its a good thing that you eschew per48 stats.

The humorous part of all this is that your silly clamoring about per48 stats was triggered by me using them to point out that Barron's hot stint with the Knicks was most likely a fluke. Exactly the kind of thing they are useful for. You'd think an intelligent person would have figured it was not the ideal time to go on a rant.

 

[Errntknght]

or he'll be averaging 5 mins of garbage time throughout the year, and you'll take his numbers and give them a per 40 or 48 minutes and think "well he'll do just as well playing regular minutes." well that 40-48 mins is just a games worth of garbage time, which isnt at all realistic.

hell, Barron's per48 is close to 10/10/1. but he sucks.

Lopez's per48 is almost 12 boards per game, but he's a terrible rebounder.

Well, you might jump to such conclusions but I would not, so speak for yourself.

Lopez per48 was 11.3 the day I checked and the league average for C's was 12.5 so I would conclude that he's a below average rebounder.

Really guys, its not all that hard once you get used to using them correctly.

 

[Covert Rain]

The bold part is quite wrong. You can use minutes per game and, for example, reb per 48 min, to compute reb per game exactly. Just multiply 'reb per 48' times 'minutes per game' and divide by 48. Actually try it once and you'll see for yourself. Using minutes per game, you can easily move back and forth between per 48/40/36 stats and per game stats.

Simply not true. I can show you tons of examples of Per 48, Per 36 that don't match Per Game stats. That's why I always say you must look at a combo of stats and looking at just one type of stat never gives you the accurate picture.

I don't mind looking at "Pers" at all. I just don't think it's accurate by itself all the time. To me it's just one piece of the puzzle and I never look at "Per" as telling me what a person does per game. Another perfect example is +/-. By itself it's pretty meaningless. However, you look at a persons points per game, turnovers, assists and "per" I think you get a pretty good picture.

 

[mojorizen7]

 

[Chaplin]

I understand that there are people who use/misuse all kinds of stats with meager grasp of their meaning. On that basis, I suppose its a good thing that you eschew per48 stats.

The humorous part of all this is that your silly clamoring about per48 stats was triggered by me using them to point out that Barron's hot stint with the Knicks was most likely a fluke. Exactly the kind of thing they are useful for. You'd think an intelligent person would have figured it was not the ideal time to go on a rant.

A rant? How is putting forth a solid argument against worthless statistics a rant? I guess it's a rant only to those who disagree.

 

[asudevil83]

Well, you might jump to such conclusions but I would not, so speak for yourself.

Lopez per48 was 11.3 the day I checked and the league average for C's was 12.5 so I would conclude that he's a below average rebounder.

Really guys, its not all that hard once you get used to using them correctly.

by that logic, the following players are "below average rebounders"

Amare
Elton Brand
Al Jefferson
LeMarcus Aldridge
Bosh
Milsap
Nene
Marc Gasol
David West

my point is that per48 numbers can make good players look bad and bad players look decent....and therefore is a TERRIBLE stat to use to gauge a players skill, value, or potential production. sure it may work for some players, but i'd say the vast majority it does little to show their real worth ingame.

 

[elindholm]

by that logic, the following players are "below average rebounders"

Amare
Elton Brand
Al Jefferson
LeMarcus Aldridge
Bosh
Milsap
Nene
Marc Gasol
David West

Aren't they? Errntknght was comparing them to the pool of centers, and a lot of those are PFs -- but even so, heck, roughly half of all members of a population should be "below average" by whatever measure you want to use. Would you argue that Stoudemire or any of those others are above average rebounders as far as NBA centers go? Exactly average? There aren't that many options.

 

[Errntknght]

Simply not true. I can show you tons of examples of Per 48, Per 36 that don't match Per Game stats. That's why I always say you must look at a combo of stats and looking at just one type of stat never gives you the accurate picture.

I don't mind looking at "Pers" at all. I just don't think it's accurate by itself all the time. To me it's just one piece of the puzzle and I never look at "Per" as telling me what a person does per game. Another perfect example is +/-. By itself it's pretty meaningless. However, you look at a persons points per game, turnovers, assists and "per" I think you get a pretty good picture.

Where did I say those things 'matched'? I said: rbp48 * mgp / 48 equals rbpgm. Give just one example where that's not true. (Heck, I'll do the arithmetic for you if necessary.)

rbp48 means rebounds per 48 minutes of playing time
mpg means minutes of playing time per game
rbpgm means rebounds per game played
* means multiply
/ means divide

 

[Errntknght]

A rant? How is putting forth a solid argument against worthless statistics a rant? I guess it's a rant only to those who disagree.

Let me quote what you yourself said a few posts back: "The problem isn't the stats themselves, it's using them to make players SEEM a lot better than they really are."

Now here you are again claiming the stats are worthless when your only argument is that people misuse them - which I agreed with in the very quote of mine you gave.

I'd say make up your mind, but I'm not sure that applies.

 

[Errntknght]

by that logic, the following players are "below average rebounders"

Amare
Elton Brand
Al Jefferson
LeMarcus Aldridge
Bosh
Milsap
Nene
Marc Gasol
David West

my point is that per48 numbers can make good players look bad and bad players look decent....and therefore is a TERRIBLE stat to use to gauge a players skill, value, or potential production. sure it may work for some players, but i'd say the vast majority it does little to show their real worth ingame.

Except for Brand, who is above average among PFs, all of them are below average for their position on reb/48. Exactly what information do you have that indicates they are somehow above average? (Thanks for the assist, Eric, I have no more to add about that part.)


If you can't remember that reb/48 means total rebounds divided by minutes played times 48 - nothing more and nothing less - then you may well misinterpret them and thus mis-categorize some players. For you the stat would be terrible, but that is not the stats fault.

You might ask yourself, if the stat is so terrible, why do players per48 stats vary so little from year to year. Per game stats bounce around much more, and the reason is obvious - the playing times vary. Per48 stats vary so little because they measure something intrinsic in the player, though, of course other things impact his performance - injuries, for example. If you want to know how Amare is going to rebound, you can look at nothing more revealing than his per48 stats over his career. When exceptional numbers show up you can usually pinpoint the reason - for example, Amares assists per48 this year is 65% above his average and the reason is, as others have remarked, D'A has more of the offense running thru him and he's expected to make plays. He's adjusted to altered expectations, as people do. Is something amiss with per48? Of course not - it accurately depicts the before and after. It didn't help in predicting the change, but then neither would any other stat.

 

[Chaplin]

Let me quote what you yourself said a few posts back: "The problem isn't the stats themselves, it's using them to make players SEEM a lot better than they really are."

Now here you are again claiming the stats are worthless when your only argument is that people misuse them - which I agreed with in the very quote of mine you gave.

I'd say make up your mind, but I'm not sure that applies.

I was referring only to specific stats, in this case those per48 you love so much. And again, you have done nothing disprove that they make bad players look good. Earl Barron is a perfect example. It's hard to "misuse" per48 points for guys that get little or not time in the NBA. The one thing I will agree with when it comes to those stats is that they are easy to come by because with garbage time players, there isn't a very big sample size.

Garbage time players are still garbage time players, regardless of how you dress them up in meaningless statistics.

 

[Covert Rain]

Where did I say those things 'matched'? I said: rbp48 * mgp / 48 equals rbpgm. Give just one example where that's not true. (Heck, I'll do the arithmetic for you if necessary.)

rbp48 means rebounds per 48 minutes of playing time
mpg means minutes of playing time per game
rbpgm means rebounds per game played
* means multiply
/ means divide

No it doesn't. Just because 48 minutes is equal to a game does not equate to Per 48 being the equivalent of a game. As I stated previously, Per 48 could include minutes from multiple games. I think people have already posted plenty of examples were guys per is inflated or understated. When compared with per game statistics they don't match.

Maybe I misunderstood you.

 

[asudevil83]

Except for Brand, who is above average among PFs, all of them are below average for their position on reb/48. Exactly what information do you have that indicates they are somehow above average? (Thanks for the assist, Eric, I have no more to add about that part.)


If you can't remember that reb/48 means total rebounds divided by minutes played times 48 - nothing more and nothing less - then you may well misinterpret them and thus mis-categorize some players. For you the stat would be terrible, but that is not the stats fault.

You might ask yourself, if the stat is so terrible, why do players per48 stats vary so little from year to year. Per game stats bounce around much more, and the reason is obvious - the playing times vary. Per48 stats vary so little because they measure something intrinsic in the player, though, of course other things impact his performance - injuries, for example. If you want to know how Amare is going to rebound, you can look at nothing more revealing than his per48 stats over his career. When exceptional numbers show up you can usually pinpoint the reason - for example, Amares assists per48 this year is 65% above his average and the reason is, as others have remarked, D'A has more of the offense running thru him and he's expected to make plays. He's adjusted to altered expectations, as people do. Is something amiss with per48? Of course not - it accurately depicts the before and after. It didn't help in predicting the change, but then neither would any other stat.

the point that some are trying to make is that Per48 does not give a true representation of a players ability within 48 minutes of playing time. i'll use antonio mcdyess as an example.

he averages 4.6 points and 4.8 boards per game. his per48 is 13.2/13.8. he doesnt start, so the vast majority of his minutes are played against other bench players who are less skilled. you could infer that he would get these kind of numbers if it were bench players the entire game, but that's not the case. giving him starters minutes against starting calibur players would probably raise his averages, but his per would go down. his per puts him in the top 10 amoung PFs, but i'd hardly call a guy who plays less than 20 minutes per game and doesnt start a top10 rebounding PF

the problem with per48 is that it falsely brings guys like Marreese Speights and Nazr Mohammad in line with guys like Noah, Duncan, and Okafor.....and assumes that base-lining minutes then base-lines production and ability.

i will say though that comparing per48 stats between players with similar assumed playing conditions (such as minutes and starts) is fair.

Now individual +/- is a difficult stat because it is so dependent on the game context in which is is earned.

i dont see how you can support per48 because it falls into the same principle as +/-. the context in which the stats were earned is easy to skew.....especially when talking about players who average sub 20 minutes per game, and even more so with guys who average sub 10 with more DNPs than games played.

 

[jagu]

+/- Rulesss!!!

 

[Errntknght]

No it doesn't. Just because 48 minutes is equal to a game does not equate to Per 48 being the equivalent of a game. As I stated previously, Per 48 could include minutes from multiple games. I think people have already posted plenty of examples were guys per is inflated or understated. When compared with per game statistics they don't match.

Maybe I misunderstood you.

Yes, you have, yet again. Of course, per48 stats do not match per game stats - when I gave you the calculation for converting between them, I thought it would be perfectly clear that I didn't remotely think they matched. (If I thought they matched I'd have said the calculation was the pergame equals per48.) Does this help: if a guy plays 24 mpg his pergame stats match per24 stats, exactly half of per48 stats. Its precisely that simple, which is why I can't understand people can get into such a tizzy over them.

 

[Errntknght]

I was referring only to specific stats, in this case those per48 you love so much. And again, you have done nothing disprove that they make bad players look good. Earl Barron is a perfect example. It's hard to "misuse" per48 points for guys that get little or not time in the NBA. The one thing I will agree with when it comes to those stats is that they are easy to come by because with garbage time players, there isn't a very big sample size.

Garbage time players are still garbage time players, regardless of how you dress them up in meaningless statistics.

I was referring to the same stats you were.

How can I disprove that a stat or anything else makes a bad player look good - those are judgements people make and people make wrongheaded judgements for all kinds of reasons, including per48 stats.

Yes, Barron is a perfect example and I used him to show how useful per48 stats can be to keep from overestimating his probable value. Now, if you want to claim that his recent reb per48 of 10.5 is misleading to you, well you just aren't very bright - what can I say? Its right at what I expected from him given his history. If he signs with another team and plays some more I'd expect about that, too. Its probably a reasonably accurate measurement of his rebounding ability - the bottom 1/3 among all the league centers. It says absolutely nothing about the rest of his game, which was pretty horrible in his games for us.

I wonder about your use of terms like 'garbage time' player - does it inflate your ego to use demeaning terms about basketball players, who as poor as they may be are much better than you (or me). Teams sometimes do lose games in 'garbage time' when they have such disdain for the guys sent out to oppose them. It does go right along with your 'meaningless' stats - a demeaning name for something you've demonstrated scant knowledge of.

 

[Errntknght]

asudevil,

i'll use antonio mcdyess as an example.
he averages 4.6 points and 4.8 boards per game. his per48 is 13.2/13.8. he doesnt start, so the vast majority of his minutes are played against other bench players who are less skilled. you could infer that he would get these kind of numbers if it were bench players the entire game, but that's not the case. giving him starters minutes against starting calibur players would probably raise his averages, but his per would go down. his per puts him in the top 10 amoung PFs, but i'd hardly call a guy who plays less than 20 minutes per game and doesnt start a top10 rebounding PF

Good choice. 13.8 rp48 is right in line with Antonio's career average. Given his age and injury history I infer that Popovich is using him an appropriate amount. Kudos to Pops. No kudos to those who draw inappropriate conclusions from stats of any kind.(A popular form of entertainment in some circles.)

i dont see how you can support per48 because it falls into the same principle as +/-. the context in which the stats were earned is easy to skew.....especially when talking about players who average sub 20 minutes per game, and even more so with guys who average sub 10 with more DNPs than games played.

The reason that +/- is more context dependent than say reb per 48 is that someone can gather rebounds at his normal rate regardless of how badly his teammates are playing while his +/- is very dependent on them. It will exactly the same as theirs for the stretch they play together. You do know that +/- is a team (lineup) stat which is attributed to individual players without considering their individual merit, right?

 

[Bufalay]

This is the worst thread ever.

 

[Mainstreet]

With all the discussion about stats, I thought I had wandered into a baseball discussion.

 

[Errntknght]

A challenge for those of you who believe that 'garbage time' per48 stats are inflated. I wouldn't ask you to prove its true but I challenge you to gather some statistical evidence that supports it. A few players who played enough GT minutes to establish a basis for their 'inflated' stats and then by some stroke of luck get to play decent minutes in non-GT. If their average net change is significantly downward then you have some evidence, otherwise you don't and its still just an opinion. If the average net change is significantly upward then its probably a wrong opinion. Of course, you could cheat and select only 'downtrending' players but my guess is that you cannot do it even by cheating.

I'd suggest avoiding stls, blks and asts(except for PGs) because they're infrequent so require many more minutes for a basis - about five times as many as points scored or rebounds per48.

The reason I don't think you can find evidence is that the effect, if it exists at all is small (my opinion) and that most players who play GT minutes then non-GT minutes are young players who generally improve over time(not entirely an opinion) and that obliterates the effect you are trying to find.

The most shocking result of all would be anyone rising to this challenge. (Or changing a long cherished opinion - regarded as an established fact.)

 

[elindholm]

The most shocking result of all would be anyone rising to this challenge.

Actually, the most shocking thing would be anyone who has been on the other side of this debate with you actually possessing the mathematical prowess to attempt any sort of analysis. You're speaking a language that is simply foreign to most of the board.

 

[mojorizen7]

This is the worst thread ever.

You must be registered for see images

 

[Covert Rain]

Yes, you have, yet again. Of course, per48 stats do not match per game stats - when I gave you the calculation for converting between them, I thought it would be perfectly clear that I didn't remotely think they matched.

My bad. I thought you were making exactly the opposite argument. I went back and read what you said. I get what your trying to say. My argument was that you can't look at Per 48 stat line and say that is what he should do per game.