What's with all these cites to "per 40 mins" stats I'm seeing lately?

[trinity92]

For the life of me, I can't see the relevance of a player's stats extrapolated out to what he would be averaging if he played 40 minutes a game. And I've been seeing references to that a LOT lately.

I get the point of such a calculation-- you're trying to show efficiency of a player who doesn't get starter's minutes. The only problem is absolutely no-one, on any team-- not even our beloved workhorse Jon Scheyer-- plays 40 minutes a game. As is pointed out regularly, there are many reasons a player can't play more minutes, whether it's because they're foul-prone, not in game-shape or languish behind other players in the rotation. The point is, if these players could really produce that much more if given additional time, they would be given that additional time.

I'm all for acknowledging a player's efficiency by keeping and referencing an X/minute stat, but I'd really like to see, if only on our boards, a retreat from all these fantasy calculations.

[OZZIE4DUKE]

The only problem is absolutely no-one, on any team-- not even our beloved workhorse Jon Scheyer-- plays 40 minutes a game.

We've had many players, star players, who routinely played 40 minutes a game, much to the chagrin of many on this board. Bobby Hurley comes first to mind for his endurance. In fact, after he played his 40 minutes, he would get on the stairmaster and do another hour or so on that. (OK, maybe it was only another 30 minutes, I don't remember.)

Shane, Jason and JJ were also monster minute machines, and all but JJ won National Championships without wearing out or running down. Admittedly, JJ did wear down (a bit).

[Qwerty]

For me the use of Per 40 Minutes stats is simply that it is the lowest common denominator among people who play different amounts. You have to start with the caveat of "he only plays 18 minutes, but..."

If, for instance, Ed Davis and Brian Zoubek are battling for rebounds, who wins? Davis has more RPG, so initially it seems he is the better rebounder. But Zoubs gets a higher number of rebounds per minute while he is in the game- so for those few minutes they are both in together, Zoubek would actually be more likely to get a given rebound. The fact that Davis will play more minutes and end up with a greater total by game-end is irrelevant in comparing them on a per-minute, or per-40-minutes, basis.

[trinity92]

For me the use of Per 40 Minutes stats is simply that it is the lowest common denominator among people who play different amounts. You have to start with the caveat of "he only plays 18 minutes, but..."

If, for instance, Ed Davis and Brian Zoubek are battling for rebounds, who wins? Davis has more RPG, so initially it seems he is the better rebounder. But Zoubs gets a higher number of rebounds per minute while he is in the game- so for those few minutes they are both in together, Zoubek would actually be more likely to get a given rebound. The fact that Davis will play more minutes and end up with a greater total by game-end is irrelevant in comparing them on a per-minute, or per-40-minutes, basis.

I guess you just don't agree with me. My whole point is there's a reason Zoubek can't play more minutes, which is the most relevant point in comparing which of two basketball players is better, or even better in a particular facet of the game. IMO, Zoubek's being a more efficient rebounder than Ed Davis in the limited time he's on the floor doesn't make him a better rebounder than Davis, and giving Z credit for minutes he might have played but didn't is irrelevant. It's not irrelevant to you, but to each his own.

BTW, without knocking Zoubek, I (and I would venture to guess, most posters here on DBR) would be willing to wager very little indeed that in a head-to-head battle, Z would "be more likely to get a given rebound," which tends to support my contention that the per-40-minute stat is meaningless.

[ncexnyc]

For the life of me, I can't see the relevance of a player's stats extrapolated out to what he would be averaging if he played 40 minutes a game. And I've been seeing references to that a LOT lately.

I get the point of such a calculation-- you're trying to show efficiency of a player who doesn't get starter's minutes. The only problem is absolutely no-one, on any team-- not even our beloved workhorse Jon Scheyer-- plays 40 minutes a game. As is pointed out regularly, there are many reasons a player can't play more minutes, whether it's because they're foul-prone, not in game-shape or languish behind other players in the rotation. The point is, if these players could really produce that much more if given additional time, they would be given that additional time.

I'm all for acknowledging a player's efficiency by keeping and referencing an X/minute stat, but I'd really like to see, if only on our boards, a retreat from all these fantasy calculations.

What's that old saying about statistics? Obviously you and I have been reading the same posts. Most of them stem from a person trying to show how Player A compares to Player B. In the majority of cases one of those players is markedly inferior, but the writer attempts to build-up his favorite player. As you've already stated, "The point is, if these players could really produce that much more if given additional time, they would be given that additional time."

[Dukeface88]

What's that old saying about statistics? Obviously you and I have been reading the same posts. Most of them stem from a person trying to show how Player A compares to Player B. In the majority of cases one of those players is markedly inferior, but the writer attempts to build-up his favorite player. As you've already stated, "The point is, if these players could really produce that much more if given additional time, they would be given that additional time."

Really? Because from what I've seen in my time on here so far, what happens is one poster says "Player A is bad because of his per game stats; we should play Player B instead" followed by the counterpoint "Player A's per minute numbers are better; he can't play a full game due to foul issues/fatigue/asthma(remember G's freshamn year?)/whatever but that doesn't mean he shouldn't be used when possible". The stat has meaning, but like most stats (+/- numbers, yards per carry, public opinion polls), it needs to be placed in its proper context and used for the proper things

[Kedsy]

For the life of me, I can't see the relevance of a player's stats extrapolated out to what he would be averaging if he played 40 minutes a game. And I've been seeing references to that a LOT lately.

I get the point of such a calculation-- you're trying to show efficiency of a player who doesn't get starter's minutes. The only problem is absolutely no-one, on any team-- not even our beloved workhorse Jon Scheyer-- plays 40 minutes a game. As is pointed out regularly, there are many reasons a player can't play more minutes, whether it's because they're foul-prone, not in game-shape or languish behind other players in the rotation. The point is, if these players could really produce that much more if given additional time, they would be given that additional time.

I'm all for acknowledging a player's efficiency by keeping and referencing an X/minute stat, but I'd really like to see, if only on our boards, a retreat from all these fantasy calculations.

I am one of the people who often cites per 40 stats in my posts here. FWIW, I assume you saw during the Penn game ESPN put up a graphic with per 40 numbers -- Z is 2nd in the nation in rebounding -- so it seems to have at least some acceptance in the world.

The reason people use per 40 numbers is not because we think the player would, could, or should play 40 minutes, but (as someone else said in this thread) because it shows relative contribution among players who play a different amount of minutes. It might be better to simply refer to contribution per minute, except that number would appear meaningless without a whole lot of context, and it's easier to get a feel for the per 40 numbers. As an example, which is easier to quickly evaluate: the fact that Z gets 0.46 rebounds per minute or the fact that he gets 18.4 rebounds per 40 minutes? (And if you agree that the per minute number is difficult to evaluate, then 40 is the figure that makes the most sense -- using per 20 minute stats or per 30 minute stats would lead to a similar problem as per 1 minutes stats.)

Another reason to use the per 40 numbers is when someone complains that a player is not contributing because he "only" scores 6 points with 3 rebounds in game where the player was on the court for 12 minutes. How much the player is on the court is generally not up to him, but what he does when he's out there is. My hypothetical person put up per 40 numbers of 20 pp40 and 10 rp40, and looking at those numbers suggests the player contributed plenty for the limited time he was on the court.

Do per 40 stats really tell us how well someone would rebound (or score or whatever) in 40 minutes? Do they really tell us which of Z or Ed Davis would grab a ball that was bouncing around between them? Of course not. Nor are they meant to. However, the biggest flaw in the stat is not that a player would get tired or foul out before getting to 40 but rather that not every minute is equal to every other minute. In other words, if you pad your stats at garbage time against the other team's walk-ons, your per 40 minutes will be overinflated.

But despite the fact that no statistic is perfect, in my opinion using per 40 statistics is a valuable way to compare players' contributions during the time they are on the court. Z's per 40 numbers have been put up over 12 games and 200 minutes, a sizeable sample, and he doesn't play that often in garbage time. Whether you think Z is any good or not, he's one of the best rebounders in the country. Would his per minute numbers go down if he played more minutes? Possibly. Does the fact that he plays fewer minutes mean he deserves fewer minutes? Absolutely not. There are many reasons why coaches play the players for the amount of minutes they do. Part of it is to optimize each player's performance and part of it is an evaluation of who on the team should play the most. But most of it is an optimization of the team's performance. And therefore your assertion that "if these players could really produce that much more if given additional time, they would be given that additional time" is incorrect.

[trinity92]

Some good points, but we just disagree on this. The very reason I started the thread is this method of statkeeping seems to have taken hold and gained acceptance. I don't like it or find it relevant, and you do.

Good point on the per-minute stats and their disconnect from the way we typically look at stats. I don't come away with much from a .48 rebounds per minute stat either.

I have to say your assertion "[t]here are many reasons why coaches play the players for the amount of minutes they do. Part of it is to optimize each player's performance and part of it is an evaluation of who on the team should play the most. But most of it is an optimization of the team's performance" is simply wrong. Every coach optimizes their team's performance by playing their best players more. Without a single exception. Duke, which is known as a team that plays its starters very heavily, is just about the opposite of Carolina, which is famous for spreading minutes around, yet the top 5 scorers from each team average a total 138.2 (Duke) and 125.8 (UNC) minutes, respectively. A 10% difference. So Roy, who is your best chance at finding a coach who might have a secret "reason" for not playing an amazing player for some reason other than having a better player available, doesn't do it much. I'm very comfortable sticking to what I said.

I expect a fan to be able to discern a quality player making quality contributions, no matter what amount of playing time they're getting. Resorting to this "per-40" calculation seems both contrived and likely to be misused. I'm worried that people will use this statkeeping method to do what a poster on this thread did-- come to the absurd conclusion from the relative "per-40" stats that a role player like Z would likely outrebound a starter like Davis playing head-to-head. Even if Z is shown to be a better per-minute rebounder than Davis, my sum reaction is "so what?"

[turnandburn55]

One method isn't "better" than another. It's just sometimes useful to normalize for another variable (which is why we look at FG% and PPG, for example, instead of just total points) when analyzing a player's performance. It's not a tell-all, just another stat that has strengths and weaknesses, like any other number we can crank out. Important thing is knowing what it tells us and where its usefulness breaks down...

[BlueintheFace]

It is a helpful tool in determining efficiency of a player.

[hughgs]

Some good points, but we just disagree on this. The very reason I started the thread is this method of statkeeping seems to have taken hold and gained acceptance. I don't like it or find it relevant, and you do.

Good point on the per-minute stats and their disconnect from the way we typically look at stats. I don't come away with much from a .48 rebounds per minute stat either.

I have to say your assertion "[t]here are many reasons why coaches play the players for the amount of minutes they do. Part of it is to optimize each player's performance and part of it is an evaluation of who on the team should play the most. But most of it is an optimization of the team's performance" is simply wrong. Every coach optimizes their team's performance by playing their best players more. Without a single exception. Duke, which is known as a team that plays its starters very heavily, is just about the opposite of Carolina, which is famous for spreading minutes around, yet the top 5 scorers from each team average a total 138.2 (Duke) and 125.8 (UNC) minutes, respectively. A 10% difference. So Roy, who is your best chance at finding a coach who might have a secret "reason" for not playing an amazing player for some reason other than having a better player available, doesn't do it much. I'm very comfortable sticking to what I said.

I expect a fan to be able to discern a quality player making quality contributions, no matter what amount of playing time they're getting. Resorting to this "per-40" calculation seems both contrived and likely to be misused. I'm worried that people will use this statkeeping method to do what a poster on this thread did-- come to the absurd conclusion from the relative "per-40" stats that a role player like Z would likely outrebound a starter like Davis playing head-to-head. Even if Z is shown to be a better per-minute rebounder than Davis, my sum reaction is "so what?"

Under this reasoning we should get rid of ERA in baseball.

[RoyalBlue08]

The thing about any statistic that makes it dangerous is that people use it to justify an opinion that is well beyond the facts contained in the statistic. To say that rebounds per 40 mins is irrelevant doesn't make much sense to me. It is a very good measure of a player's rebounding ability while they are on the floor. What makes equally little sense is to take a number like that an extrapolate to something like value to a team without considering 30-40 other relevant facts. Determining a player's value based on statistics in basketball or any other sport is very, very difficult; but many books worth of material have been written on the subject, so I will leave it at that. It would be interesting to see some sort of version of the PER ratings that ESPN is using for the NBA these days for Duke and other college players.

[turnandburn55]

The thing about any statistic that makes it dangerous is that people use it to justify an opinion that is well beyond the facts contained in the statistic. To say that rebounds per 40 mins is irrelevant doesn't make much sense to me. It is a very good measure of a player's rebounding ability while they are on the floor. What makes equally little sense is to take a number like that an extrapolate to something like value to a team without considering 30-40 other relevant facts. Determining a player's value based on statistics in basketball or any other sport is very, very difficult; but many books worth of material have been written on the subject, so I will leave it at that. It would be interesting to see some sort of version of the PER ratings that ESPN is using for the NBA these days for Duke and other college players.

Yeah... stats are useful tools, but only if you know what they're good for. Remember the "Randy Ratio"??? The rationale was that since Randy Moss averages 18 yards per catch (or whatever), imagine if he caught 100 passes! Or 120! He'd have 1800-2200 yards! Simple extrapolation, right??

Didn't quite work out that way, since YPC has an odd tendency to vary with the number of times you get the ball...

[Kedsy]

Some good points, but we just disagree on this. The very reason I started the thread is this method of statkeeping seems to have taken hold and gained acceptance. I don't like it or find it relevant, and you do.

Good point on the per-minute stats and their disconnect from the way we typically look at stats. I don't come away with much from a .48 rebounds per minute stat either.

I have to say your assertion "[t]here are many reasons why coaches play the players for the amount of minutes they do. Part of it is to optimize each player's performance and part of it is an evaluation of who on the team should play the most. But most of it is an optimization of the team's performance" is simply wrong. Every coach optimizes their team's performance by playing their best players more. Without a single exception. Duke, which is known as a team that plays its starters very heavily, is just about the opposite of Carolina, which is famous for spreading minutes around, yet the top 5 scorers from each team average a total 138.2 (Duke) and 125.8 (UNC) minutes, respectively. A 10% difference. So Roy, who is your best chance at finding a coach who might have a secret "reason" for not playing an amazing player for some reason other than having a better player available, doesn't do it much. I'm very comfortable sticking to what I said.

I'm happy to agree to disagree on the value of per 40 calculations. However, I see a problem with your idea that "[e]very coach optimizes their team's performance by playing their best players more," which is there is no objective way to determine who the "best" players are.

Of course the top 5 scorers get most of the minutes. If Olek Czyz had gotten 35 minutes a game he would have been one of Duke's top 5 scorers. So if the top five scorers are your definition of the guys who deserve the most minutes, it's a circular argument. But this illustrates my point. Olek could have scored 10+ points a game given 30+ minutes, but his defense was so poor the team's performance would have suffered. That's part of what I meant when I said the coach allocates minutes in order to optimize the team's performance, and you can say I'm "simply wrong," except you're simply wrong when you say it.

Also, depth at a particular position affects minutes but has nothing to do with who the best players are. If Duke had Zoubek plus a bunch of guards, you can bet he'd get as many minutes as he could handle before fouling out or collapsing in exhaustion. He'd probably also be leading the conference in rebounding and scoring double-figure points if the team were forced to play him so many minutes. But just because we have four other big men to play, and thus don't need to play Z so many minutes does not make him a worse player. That's why the per 40 minute calculations are helpful, at least to me.

[stickdog]

No statistic is all that meaningful outside of its context. Per 40 minute statistics have all of the problems you suggested. But that doesn't meaning they are meaningless statistics. When some people suggest, for example, that Brian Zoubek does not deserve any minutes in competitive games because he has never averaged more than 7 points or 8 rebounds per game, it is illustrative to point out to them that Brian Zoubek is 1st the NCAA in offensive rebounding percentage, 1st in the ACC and 8th in the NCAA in field goal percentage, 3rd in the ACC in offensive rebounds per game, 6th in the ACC in defensive rebounding percentage, 6th in the ACC in true shooting percentage, 9th in the ACC in floor percentage and is averaging 17 points, 18 rebounds, 9 offensive rebounds, 2 assists, 2 steals & 2 blocks per 40 minutes played.

Notice that all of the statistics in which Zoubek shines (except for offensive rebounds per game), are either percentage based or else somehow equalized for time on the floor. This is because the fact that Zoubek only plays 17 minutes per game obscures just how productive his season has been so far on a per minute basis. Does this mean that Zoubek could actually ever produce 17 points, 18 rebounds, 9 offensive rebounds, 2 assists, 2 steals & 2 blocks per game if he played every possible minute? Of course not. He would tire. He would foul out of many or most games and he would have to avoid contact due to foul trouble in most other games. However, appreciating these obvious limitations does not render these statistics meaningless.

[DevilWolf]

So Steve Johnson comes into a game with 2 minutes left against the other team's walk-ons, gets an offensive rebound and is fouled on the put-back and hits the two free throws.

Is it fair to say he is a 40 point, 20 rebound per 40 minutes player?

[Bob Green]

So Steve Johnson comes into a game with 2 minutes left against the other team's walk-ons, gets an offensive rebound and is fouled on the put-back and hits the two free throws.

Is it fair to say he is a 40 point, 20 rebound per 40 minutes player?

Multiple posters have already addressed your question in this thread.

No statistic is all that meaningful outside of its context.

The reason people use per 40 numbers is not because we think the player would, could, or should play 40 minutes, but (as someone else said in this thread) because it shows relative contribution among players who play a different amount of minutes.

The stat has meaning, but like most stats (+/- numbers, yards per carry, public opinion polls), it needs to be placed in its proper context and used for the proper things

For me the use of Per 40 Minutes stats is simply that it is the lowest common denominator among people who play different amounts.

[Acymetric]

So Steve Johnson comes into a game with 2 minutes left against the other team's walk-ons, gets an offensive rebound and is fouled on the put-back and hits the two free throws.

Is it fair to say he is a 40 point, 20 rebound per 40 minutes player?

Really? You honestly don't understand what the difference between per 40 stats for Johnson and per 40 for Z or any other player that gets significant minutes?

The closer the player is to 40 minutes played, the more accurate their per 40 is. No one is saying that if Z played 40 minutes he would put up those numbers, but in context, combined with other statistics and some critical thinking (something you should try sometime, you may like it) per 40 numbers can be informative. If someone who plays at least 10-15 minutes has good per 40 numbers, it means they probably deserve those minutes. It doesn't mean they should play 40, because they wouldn't be able to keep up that production the entire game.

What per 40 minutes do mean, though, is that certain players (Z) do more with their limited time in the game than some others. Thats where per 40 comes in.

[DevilWolf]

My post was extreme, like all those idiots who talk about somebody being on pace for 300+ home runs after opening day.

But I would be willing to bet that points per 40 minutes doesn't show up on any scouting report used by anyone anywhere. It's a stat designed for guys who don't get enough minutes to let their real stats speak for themselves, and they're usually used to make an argument that so-and-so should start/play more minutes.

In Zoubek's case, he doesn't even need this bogus calculation to get credit for the huge impact he's had on our season so far.

[COYS]

In Zoubek's case, he doesn't even need this bogus calculation to get credit for the huge impact he's had on our season so far.

No, he doesn't need per 40 minute stats to show that he's played well for us this year, but the stat does confirm that he makes a major contribution while on the floor.

I think you can look at per 40 minute numbers almost like you would look at shooting percentage or free throw percentage. If a technical is assessed to the other team, you would want to know which player on the court has the best chance of making the free throws. Who cares if that player only scores 2.9 ppg from the charity stripe if they have a free throw percentage of 90%? Same is true with Zoubek. Who cares if he can only play 20 minutes per game if he's going to grab lots of rebounds, block a few shots, and shoot a high percentage from the floor while he's in there. If we need rebounding, shot blocking, and a few easy put-backs (which we do in every game we play) the per 40 minute numbers tell us we can run zoubs out there and he'll be productive while he's on the court.