View previous topic :: View next topic |
Author |
Message |
coach_dbop
Joined: 30 Mar 2007 Posts: 44 Location: California
|
Posted: Sun Apr 08, 2007 10:54 pm Post subject: Players Minutes and how they are calculated |
|
|
I throw this question out to fellow stat followers and basketball fans.
Since the 2005-06 season players minutes played have been calculated differently. If you visit basketball websites (NBA.com, ESPN.com) or any other websites that contain stats on the NBA you might notice differences in a players minutes for the season and possibly for certain games. This problem has bothered me since I keep daily stats on every NBA team and every NBA player. I compare my results with the results on NBA.com and ESPN.com and it always upsets me when there is a slight difference in the information because when it comes to stats I like to be able to have the exact number for reference purposes.
To give you an example if you look up Joe Johnson's 2006-07 stats on NBA.com and ESPN.com you will that NBA.com has him at 2359 minutes played and ESPN.com has him at 2360 minutes played. I went on Hoopstats.com and was curious what they had for him under minutes played. Their results came to 2357 minutes played.
I then look at his individual game logs on all 3 websites comparing them side by side together and found that in at least 4 games there is a difference of 1 minute for his game stats.
Prior to the 2005-06 season NBA.com on their boxscores had players minutes in a round whole number. The past 2 years they have posted a players minutes in minutes and seconds form. An example of this would be (43:32) saying that player played 42 minutes and 32 seconds. I have assumed that when dealing with that if they played higher than 30 seconds you round up the nearest whole number and if they played less than 30 seconds you round down.
If anyone takes a look at the examples above and is able to give me some insight on this slight problem I would gladly appreciate it. I am in the process of setting up a basketball history website as well as publishing a yearly statistics book and this one little problem would make it difficult to give people who ready my stuff the most accurate data. _________________ Greg Eggleston
Lakewood Lancers Assistant Basketball Coach
Director of Basketball Operations |
|
Back to top |
|
|
Ryan J. Parker
Joined: 23 Mar 2007 Posts: 707 Location: Raleigh, NC
|
Posted: Mon Apr 09, 2007 12:10 am Post subject: |
|
|
Unfortunately the answer is probably find the most accurate website and use those calculations for everyone. |
|
Back to top |
|
|
coach_dbop
Joined: 30 Mar 2007 Posts: 44 Location: California
|
Posted: Mon Apr 09, 2007 1:56 pm Post subject: |
|
|
Well it is my understanding after doing some research that websites get their stats from STATS LLC which i've heard is the statistics company for sports or something like that. What I would like to know if anyone has any insight or information is why did the NBA change the way they posted a players minutes played in the box score? If you take a look at a 2002-03 box score the minutes are all a whole number with no seconds attached and if you added them all up it equals 240. Now take a box score from the 2005-06 or this season and take a team and add the minutes played using both minutes and seconds and it will not equal 240. I've calculated some box scores for teams that they have a total of 242 minutes played. That can't be possible with the game being 48 minutes or 480 for stats purposes. Both teams in a regulation game produce 240 minutes played. This is something that would factor into a players PER and affect the actual PER for that player. _________________ Greg Eggleston
Lakewood Lancers Assistant Basketball Coach
Director of Basketball Operations |
|
Back to top |
|
|
kjb
Joined: 03 Jan 2005 Posts: 860 Location: Washington, DC
|
Posted: Mon Apr 09, 2007 2:04 pm Post subject: |
|
|
My guess is that the NBA went for what they thought would be greater precision -- instead of rounding to the nearest minute and "forcing" the total to be 240, now we have playing time to the second. BUT, the totals always should equal 240. That sometimes it doesn't suggests there's a glitch in the system somewhere. Kinda surprising they haven't addressed it yet.
And no, you're not the only person to notice this. I've seen it a few times, and always figured it was some kind of user error (data inputting error). My guess is that nba.com would have the most accurate box score stats. |
|
Back to top |
|
|
coach_dbop
Joined: 30 Mar 2007 Posts: 44 Location: California
|
Posted: Mon Apr 09, 2007 3:34 pm Post subject: |
|
|
Yes, I agree. NBA.com being the official site I would think that since its their teams playing in their games their box scores would be accurate. Looks like the statisticians at the game have yet to find this error. If there is anyone out there who knows a way to contact a person working on the NBA.com website perhaps mention this to them and see if it can be corrected. _________________ Greg Eggleston
Lakewood Lancers Assistant Basketball Coach
Director of Basketball Operations |
|
Back to top |
|
|
mtamada
Joined: 28 Jan 2005 Posts: 375
|
Posted: Mon Apr 09, 2007 5:57 pm Post subject: |
|
|
Sounds like rounding error. Until two years ago, it's evident that NBA statskeepers were told that the minutes had to add up to exactly 240 minutes (for non-overtime games). (And also that every player who played had to be given at least 1 minute of reported playing time, no "0 minutes" as we see nowadays).
Nowadays they're probably rounding each player to the nearest minute, which can result in rounding error. (Or have been given the option of doing that rounding. Or maybe the NBA hasn't developed an algorithm for apportioning minutes so they add up to 240.)
E.g.: suppose Luke Ridnour, Earl Watson, and Mike Wilks collectively play 48 minutes at PG for the Sonics tonight. Ridnour and Watson each play 21:40, and Wilks plays 4:40.
How would you report each player's minutes played, in integers? Most of us I think would instinctively report Ridnour and Watson at 22, and Wilks at 5. Voila, a combined 49 minutes played. |
|
Back to top |
|
|
coach_dbop
Joined: 30 Mar 2007 Posts: 44 Location: California
|
Posted: Mon Apr 09, 2007 6:47 pm Post subject: |
|
|
That is correct. That is the conclusion I came to. In reality any player that has played in the NBA from the 2005-06 season to present date will not have accurate stats until they fix that problem. _________________ Greg Eggleston
Lakewood Lancers Assistant Basketball Coach
Director of Basketball Operations |
|
Back to top |
|
|
Ryan J. Parker
Joined: 23 Mar 2007 Posts: 707 Location: Raleigh, NC
|
Posted: Mon Apr 09, 2007 6:50 pm Post subject: |
|
|
Correct me if I'm wrong, but this should really only affect players that don't play that much.
I figure a player is just as likely to be rounded up as they are rounded down, so over an extended period of time this sort of thing works itself out (they get rounded both ways an almost even number of times). |
|
Back to top |
|
|
mtamada
Joined: 28 Jan 2005 Posts: 375
|
Posted: Mon Apr 09, 2007 7:13 pm Post subject: |
|
|
Ryan J. Parker wrote: | Correct me if I'm wrong, but this should really only affect players that don't play that much.
I figure a player is just as likely to be rounded up as they are rounded down, so over an extended period of time this sort of thing works itself out (they get rounded both ways an almost even number of times). |
That's my reaction too. One individual game where a player played only 10 minutes could have his per minute stats skewed by up to 5% by rounding error, and all players' stats might be skewed by almost 1% if their collective team minutes add up to 242 instead of 240. But over a season, I would expect this imprecision to have minimal impact.
But it'd be itneresting to look at the actual numbers, if there were a way to re-create or simulate the old, ingtegral, sum-to-240 NBA minutes for 2007, and compare them to what the NBA actually reported for this year. |
|
Back to top |
|
|
coach_dbop
Joined: 30 Mar 2007 Posts: 44 Location: California
|
Posted: Tue Apr 10, 2007 6:57 am Post subject: |
|
|
Ok, this is where it gets interesting. Earlier I mentioned if that in 2005-06, Alex Scales played 1 game for the San Antonio Spurs. It was the Spurs 10th game of the season and they were playing Phoenix. Now on both NBA.com and ESPN.com they have him officially down as 1 game played and 0 minutes. It is also published the the NBA's 2006-07 NBA Register. Now if you go to Basketballreference.com and look up Alex Scales they have him marked down that he played 1 game for the Spurs and that he played 1 minute. Now I also have compared a few things and its kind of interesting. I compared Kobe Bryant's minutes this season from NBA.com and ESPN.com and compared those to my excel spredsheets that I keep a daily total of and this is the interesting thing. Both NBA.com and ESPN.com match with his minutes in every game. However NBA.com's total minutes for him are 3 minutes less than on ESPN.com. I added up the minutes from both websites and the total that is on ESPN.com is what i got. So the question for the NBA.com people is where did those 3 minutes go? _________________ Greg Eggleston
Lakewood Lancers Assistant Basketball Coach
Director of Basketball Operations |
|
Back to top |
|
|
cherokee_ACB
Joined: 22 Mar 2006 Posts: 157
|
Posted: Tue Apr 10, 2007 7:28 am Post subject: |
|
|
IMO, NBA.com is doing the right thing. I suspect they are adding minutes and seconds for all games, while other sites just add game minutes. That means NBA.com numbers are the most accurate here.
Want a proof? Take James Augustine, for instance. He's just played two games this season
http://www.nba.com/games/20070202/NJNORL/boxscore.html
http://www.nba.com/games/20070214/PORORL/boxscore.html
His playing time in those games is 4:40 and 2:39, which adds to 7:19. NBA.com has him as having played 7 minutes; all other sources give him an extra minute, since 4:40 is rounded up to 5, and 2:39 becomes 3.
Last edited by cherokee_ACB on Wed Apr 11, 2007 2:01 am; edited 1 time in total |
|
Back to top |
|
|
coach_dbop
Joined: 30 Mar 2007 Posts: 44 Location: California
|
Posted: Tue Apr 10, 2007 7:51 am Post subject: |
|
|
Actually if you were to look at James Augstine's player profile on NBA.com and then go to the game log tab and look at the his game log you will see the two games that you are referring to. I just happen to be looking at it right now and it says in clear writing that NBA.com rounded his minutes to both 5 and 3. Now it says that he played in 2 games and averaged 3.5 minutes a game which would total 7 minutes, but where my problem lies is that these types of errors or mistakes should be not happen. If I'm looking at going ok James Augustine averaged 3.5 minutes per game and then went to look at his game logs and saw that minutes played didnt match his average minutes that would raise a question about the accuracy of the stat. _________________ Greg Eggleston
Lakewood Lancers Assistant Basketball Coach
Director of Basketball Operations |
|
Back to top |
|
|
cherokee_ACB
Joined: 22 Mar 2006 Posts: 157
|
Posted: Tue Apr 10, 2007 9:18 am Post subject: |
|
|
coach_dbop wrote: | If I'm looking at going ok James Augustine averaged 3.5 minutes per game and then went to look at his game logs and saw that minutes played didnt match his average minutes that would raise a question about the accuracy of the stat. |
But that stat is accurate. Your complain seems to be with the precision of player minutes in the game log. Fine. Ask NBA.com to insert seconds there. But if I have to choose between accuracy and apparent consistency, I'll go with the former. |
|
Back to top |
|
|
coach_dbop
Joined: 30 Mar 2007 Posts: 44 Location: California
|
Posted: Tue Apr 10, 2007 9:41 am Post subject: |
|
|
Ok my question to you is how is that stat accurate? Are you telling me that if they have 5 minutes and 3 minutes played in his game log and you add those up to get 8 minutes played and then you wanted to know his minutes averaged so you divide minutes played by games played (in this scenerio 8 / 2). I'm not a math major, but that does not come out to a 3.5 minutes per game average. This situation is making my point. In the box score they mark him down for 4:40 then they round it up to make it a whole number and insert that into his game log. So if someone is using box scores as a reference to track stats they will never be accurate because taking those numbers and rounding them up or down you will run into conflicts with players minutes for a team in a game not adding up to 240. As long as I am getting the accurate information so I can set up my database I am fine with the way the do things, but when it comes to stats and what I'm trying to accomplish with putting together a website and yearly stats book I am very picky about my information being accurate. _________________ Greg Eggleston
Lakewood Lancers Assistant Basketball Coach
Director of Basketball Operations |
|
Back to top |
|
|
cherokee_ACB
Joined: 22 Mar 2006 Posts: 157
|
Posted: Tue Apr 10, 2007 10:49 am Post subject: |
|
|
What I'm saying is that NBA.com does the right thing by using seconds to get to their numbers, only that they don't show seconds but round to minutes for presentation purposes. But they don't carry the game rounding errors to the total player minutes, which is good.
Now, we statheads know that 5 minute is not really 5 minutes, but something between 4:30 and 5:30. So we can't be sure that 5+3 is always 8. If nba.com tells me it's 7, I'm inclined to trust them. If you don't, well, just grab the player times with seconds from boxscores, and check that they add up to 240 minutes per game and team, as they should. That will give you greater precision in player minutes and get the figures you want for team minutes. |
|
Back to top |
|
|
|