Articles on statistical +/- (Crow. 2009)
Posted: Mon Apr 18, 2011 5:08 am
Author Message
Crow
Joined: 20 Jan 2009
Posts: 797
PostPosted: Tue Jun 09, 2009 2:17 am Post subject: Articles on statistical +/- Reply with quote
Had missed these articles til now.
Interesting stuff Scott S.
fyi
http://basketball-statistics.com/sereda ... rtone.html
http://basketball-statistics.com/sereda ... rttwo.html
Back to top
View user's profile Send private message
deepak
Joined: 26 Apr 2006
Posts: 664
PostPosted: Tue Jun 09, 2009 1:30 pm Post subject: Reply with quote
I had not until now seen in the adjusted defensive +/- numbers for the past season. Are these numbers available for all players anywhere? It looks like he just selected the players with a defensive reputation.
Back to top
View user's profile Send private message
Ben
Joined: 13 Jan 2005
Posts: 264
Location: Iowa City
PostPosted: Tue Jun 09, 2009 3:35 pm Post subject: Reply with quote
Very interesting. Thanks for posting.
Back to top
View user's profile Send private message
DLew
Joined: 13 Nov 2006
Posts: 222
PostPosted: Wed Jun 10, 2009 11:08 am Post subject: Reply with quote
This is interesting work. Scott, you may want to publicize this a little more, I'm not sure how anyone would have known about it.
Back to top
View user's profile Send private message
Neil Paine
Joined: 13 Oct 2005
Posts: 774
Location: Atlanta, GA
PostPosted: Wed Jun 10, 2009 11:21 am Post subject: Reply with quote
Did you run the regressions on unadjusted offensive and defensive +/-? Because as far as I know, there's not an online source for adjusted OAPM & DAPM beyond what Eli did for 2007-08 (and maybe some of Dan Rosenbaum's early work).
Back to top
View user's profile Send private message Visit poster's website
Jacob S.
Joined: 24 Jun 2008
Posts: 2
PostPosted: Wed Jun 10, 2009 12:41 pm Post subject: Reply with quote
I agree that it's very interesting work, and I had completely missed it until now. It deserves wider viewership, IMO.
Have you looked at the collinearity between the different variables?
Back to top
View user's profile Send private message
Mike G
Joined: 14 Jan 2005
Posts: 3528
Location: Hendersonville, NC
PostPosted: Wed Jun 10, 2009 1:02 pm Post subject: Reply with quote
Whenever a site pops up that is redundant with APBRmetrics, I wonder how the proprietors hope to reach a bigger or better audience than by posting right here. Unless there's some functionality not available here? I don't see any place to put a response to the posts there.
"Basketball-Statistics.com | Innovative Stats and Analysis "
Is that not what APBRMetrics is, already? Is Scott Sereday a member and a poster? If he isn't (any longer), you have to email him. Seems like a detour or two.
Maybe people feel their submissions are more secure elsewhere. Or they don't really want any feedback?
_________________
`
36% of all statistics are wrong
Back to top
View user's profile Send private message Send e-mail
Scott S
Joined: 10 Feb 2008
Posts: 46
Location: East Rutherford, NJ
PostPosted: Wed Jun 10, 2009 9:31 pm Post subject: Reply with quote
Sorry I didn't bring this up before guys. I had planned on starting a post with the link (I feel I can get too wordy for a post at times). I was just hoping to finish some more analysis first and got busy.
To answer a few of your questions, I chose the top several players and then additional players based on reputation.
In order to obtain the adjusted offensive and defensive numbers, I used the method described by Eli W as follows:
http://www.countthebasket.com/blog/2008 ... lus-minus/
I was trying to use a method that was more similar to that done by Ilardi and Barzilai in http://www.82games.com/ilardi2.htm, but R couldn't seem to handle it.
You can see the some of the effects of collinearity between the different variables when their coefficients change after selecting different variables. I ran several different combinations to find the lowest standard errors and most reasonable interactions, etc.
I hope to post again soon guys.
Let me know if you have any other comments/suggestions/questions.
Back to top
View user's profile Send private message Visit poster's website
deepak
Joined: 26 Apr 2006
Posts: 664
PostPosted: Wed Jun 10, 2009 11:12 pm Post subject: Reply with quote
Scott S wrote:
I hope to post again soon guys.
Let me know if you have any other comments/suggestions/questions.
If possible, I'd appreciate a link to your adjusted +/- numbers for all players, instead of just a select few. Thanks for the work.
Back to top
View user's profile Send private message
Crow
Joined: 20 Jan 2009
Posts: 797
PostPosted: Thu Jun 11, 2009 3:02 am Post subject: Reply with quote
What if you split up missed field goals by location in the model?
Does the value of an offensive rebound vary much by shot location?
Not that the data is there but would a penalty for the passer to the shooter if the shot is missed be appropriate? Conceptually I think it would.
Would be very helpful to see a full team or the full league. I get the impression it is tough to end up positive. What % of players do? 20-30%?
A player that went 5-10 from field composed of 2 of 3 from inside, 2-4 mid-range (1 assisted 1 not) and 1-3 from 3 point range(assisted) with 3 assists 4 rebounds (1 offensive) 1 steal 2 turnovers and 3-4 from line is close to zero, I think... except for True Shot Attempts X (Height - 60)?
Is that True Shot Attempts X (Height - 60) thing right? Isn't that a huge factor or am I misreading?
On the defensive side steals are highly valued, with the cost of steal attempts hard to see? Ideally I'd be interested in a points allowed / steal attempt factor. Height (by itself) not that important on defense (way less than offense??)
Back to top
View user's profile Send private message
Scott S
Joined: 10 Feb 2008
Posts: 46
Location: East Rutherford, NJ
PostPosted: Thu Jun 11, 2009 7:48 pm Post subject: Reply with quote
Quote:
If possible, I'd appreciate a link to your adjusted +/- numbers for all players, instead of just a select few. Thanks for the work.
I will work on summarizing all of the results (probably after the draft)
Quote:
What if you split up missed field goals by location in the model?
I split field goals by close shots, long 2's and 3 pointers. Any other locations might be difficult to compile. Maybe in another study? I know there is some data available on nba.com, but where would I find data in a form easier to extract?
Quote:
Does the value of an offensive rebound vary much by shot location?
I think the value of an offensive rebound is strongly tied to the probability of the opposing team obtaining a rebound had it not been collected by that player. There was a study on 82 games.com that showed rebounding percentages didn't differ greatly from the three different shot "locations" I chose. I do think there are different types of shots that produce varying rebounding expectations and that these can be extracted somehow statistically, but I haven't looked into this extensively yet.
Quote:
Not that the data is there but would a penalty for the passer to the shooter if the shot is missed be appropriate? Conceptually I think it would.
This data is hard to come by. I do think that "potential assists" could help shed some light on the actual value of each assists by a player, but I don't know that assisted misses should be negatively attributed to the passer in most cases. The passer did his part by delivering the ball without a turnover. If the shooter decides not to shoot, the most significant negative value of the pass that I can think of is the time running off the shot clock to force a tougher shot later. I tried an interaction with late shot clock assists, but the play by play seems sketchy with the shot clock and I didn't get enough data. A passer should ideally be evaluated by whether his passes increase or decrease the teams scoring expectations. A good estimate should reflect this. This is why close assists are so much more valuable.
Quote:
Would be very helpful to see a full team or the full league. I get the impression it is tough to end up positive. What % of players do? 20-30%?
A player that went 5-10 from field composed of 2 of 3 from inside, 2-4 mid-range (1 assisted 1 not) and 1-3 from 3 point range(assisted) with 3 assists 4 rebounds (1 offensive) 1 steal 2 turnovers and 3-4 from line is close to zero, I think... except for True Shot Attempts X (Height - 60)?
similar percentages as regular adjusted plus/minus and probably statistics like PER. Since the best players play the most minutes, they have a higher relative impact on the average so the vast majority of players are below "average". If I had to guess, a third are above average?
Quote:
Is that True Shot Attempts X (Height - 60) thing right? Isn't that a huge factor or am I misreading?
Actually, it is a typo. It should be [TSA(ht-60)]/100. So a 7 footer and 6 footer are statistically identical and average 20 true shot attempts, this indicates we would expect the 7 footer to have a value of about 1.5 points more than the 6 footer. I initially wanted to see if there was any value for a post player shooting outside and drawing the shot blocker away from the basket, but discovered it was just more valuable in general for a big man to shoot the ball, all else being equal.
Quote:
On the defensive side steals are highly valued, with the cost of steal attempts hard to see? Ideally I'd be interested in a points allowed / steal attempt factor. Height (by itself) not that important on defense (way less than offense??)
Steals are very valuable. The high value of their coefficient can likely be attributed to a correlation between good defensive pressure and steals. This would include good defensive pressure that either forces a turnover or a lower percentage shot. The negative impact of "steals attempted" and failed is included in both Opponent PER Plus/Minus and possibly a few more variables. One factor that should be considered is that a missed steal has a limited detrimental impact because there are 4 more players available to help. I do think that someone who does it too often is clearly still hurting to the team, but haven't found any way to value this statistically yet.
The interaction between height and blocks seems to account for almost all of the value of blocks and height. I guess this means if you are very short and block shots, those blocks don't help your team very much. Similarly, it seems to mean if you are tall and no threat to block shots, your height is not very useful at all defensively, after accounting for defensive rebounding and the other variables in the model.
Back to top
View user's profile Send private message Visit poster's website
YaoPau
Joined: 28 Jan 2009
Posts: 28
PostPosted: Fri Jun 12, 2009 10:14 pm Post subject: Reply with quote
Scott, are you sure those Off/Def APM numbers are correct? You've got Dwight Howard at +1.9 offensively and +8.4 defensively, but basketballvalue has his overall APM this year at just over +1.
I tried running APM using Eli's method and got -3.59 offensively and +5.17 defensively for Howard. It's hard to say for sure whether my results are accurate, but I followed the method closely and the DiffOD results seem to match up to the on/off 82games splits I've looked at so far.
Back to top
View user's profile Send private message Visit poster's website
Scott S
Joined: 10 Feb 2008
Posts: 46
Location: East Rutherford, NJ
PostPosted: Sat Jun 13, 2009 1:27 am Post subject: Reply with quote
I ended up selecting roughly the top 71% of the players by minutes in my regression model, which very likely differs from your model. I chose the players who basketballvalue used in their model based on the file available for download. However, several players (such as Martell Webster) had a rating with just limited minutes played (5.25 minutes for him and his rating doesn't match his two year rating on the actual site). I eliminated these records in my run. Note that these ratings were not used in the estimates of the coefficients for the statistics of my model, so I didn't check them as extensively as I checked the other years. I really only posted them as a reference. So, basically, both of our models could be right or both could be wrong, etc. but I would recommend using the more consensus model in any studies that depend on these values.
Back to top
View user's profile Send private message Visit poster's website
YaoPau
Joined: 28 Jan 2009
Posts: 28
PostPosted: Sat Jun 13, 2009 1:59 am Post subject: Reply with quote
Hmm I did the same thing - using the players basketballvalue listed an APM for in the playerstats spreadsheet, although I didn't delete anybody beyond that. Why the change in Howard's APM? You've got Kobe at +13 too, when bv has him at +7.87.
Back to top
View user's profile Send private message Visit poster's website
Jon Nichols
Joined: 18 Aug 2005
Posts: 370
PostPosted: Sat Jun 13, 2009 10:08 am Post subject: Reply with quote
Mike G wrote:
Whenever a site pops up that is redundant with APBRmetrics, I wonder how the proprietors hope to reach a bigger or better audience than by posting right here. Unless there's some functionality not available here? I don't see any place to put a response to the posts there.
"Basketball-Statistics.com | Innovative Stats and Analysis "
Is that not what APBRMetrics is, already? Is Scott Sereday a member and a poster? If he isn't (any longer), you have to email him. Seems like a detour or two.
Maybe people feel their submissions are more secure elsewhere. Or they don't really want any feedback?
As the creator of Basketball-Statistics.com, I guess I might as well respond to this post and a few of the others talking about a lack of recognition. First, Scott's ideas are great. In order to maximize his exposure, I tried to set a block of time on the site where I didn't post any new entries. In retrospect, I should have posted links on this space. I think it was simply a matter of me being forgetful.
As for your thoughts about having my own web site, I understand how you can see things that way but I strongly disagree. As a matter of fact, nearly every big project I work on (Composite Score, Box Score Prediction System, PAC, etc.) is posted on this board and I take any and all comments. Every stat I have created has incorporated (usually to a considerable extent) ideas from posters here. Composite Score was almost entirely based on those ideas.
My site doesn't get a ton of recognition, but that's not from a lack of trying. I've written for NBADraft.net, RealGM.com, and now Hardwood Paroxysm with the intent of drawing attention to my work and the work of others such as Scott who have posted on my site. It's just not that easy to find a lot of people interested in this stuff.
I'm glad Scott's work has gained attention. He clearly spent a lot of time on this project. I will be sure to post any new updates from him on this board.
-Jon Nichols
Author Message
Mike G
Joined: 14 Jan 2005
Posts: 3613
Location: Hendersonville, NC
PostPosted: Sat Jun 13, 2009 11:35 am Post subject: Reply with quote
Jon,
With all sincerity, how can you 'strongly disagree' with a few questions (and an 'I wonder')?
No one has suggested it's better to not have your own site. But if worthwhile research is happening, I'm just asking if it can be expected to get better exposure elsewhere.
It's not hard to toggle between a batch of information on one site and a question/answer area elsewhere. But it does make it easier for a 'spinoff' site to lose touch, as you are aware.
_________________
`
36% of all statistics are wrong
Back to top
View user's profile Send private message Send e-mail
Jon Nichols
Joined: 18 Aug 2005
Posts: 370
PostPosted: Sat Jun 13, 2009 11:38 am Post subject: Reply with quote
Mike G wrote:
Jon,
With all sincerity, how can you 'strongly disagree' with a few questions (and an 'I wonder')?
No one has suggested it's better to not have your own site. But if worthwhile research is happening, I'm just asking if it can be expected to get better exposure elsewhere.
It's not hard to toggle between a batch of information on one site and a question/answer area elsewhere. But it does make it easier for a 'spinoff' site to lose touch, as you are aware.
Mike,
That wasn't supposed to sound harsh. I was just stating my preference for having my own site. And as you said, it is easier for a spinoff site to lose touch, but at the same time I can't post every update I make on this board. It would simply be too much. Creating a separate entity allows for the best of both worlds.
I didn't mean to get this discussion off track. I think there were some good questions...
-Jon
Crow
Joined: 20 Jan 2009
Posts: 797
PostPosted: Tue Jun 09, 2009 2:17 am Post subject: Articles on statistical +/- Reply with quote
Had missed these articles til now.
Interesting stuff Scott S.
fyi
http://basketball-statistics.com/sereda ... rtone.html
http://basketball-statistics.com/sereda ... rttwo.html
Back to top
View user's profile Send private message
deepak
Joined: 26 Apr 2006
Posts: 664
PostPosted: Tue Jun 09, 2009 1:30 pm Post subject: Reply with quote
I had not until now seen in the adjusted defensive +/- numbers for the past season. Are these numbers available for all players anywhere? It looks like he just selected the players with a defensive reputation.
Back to top
View user's profile Send private message
Ben
Joined: 13 Jan 2005
Posts: 264
Location: Iowa City
PostPosted: Tue Jun 09, 2009 3:35 pm Post subject: Reply with quote
Very interesting. Thanks for posting.
Back to top
View user's profile Send private message
DLew
Joined: 13 Nov 2006
Posts: 222
PostPosted: Wed Jun 10, 2009 11:08 am Post subject: Reply with quote
This is interesting work. Scott, you may want to publicize this a little more, I'm not sure how anyone would have known about it.
Back to top
View user's profile Send private message
Neil Paine
Joined: 13 Oct 2005
Posts: 774
Location: Atlanta, GA
PostPosted: Wed Jun 10, 2009 11:21 am Post subject: Reply with quote
Did you run the regressions on unadjusted offensive and defensive +/-? Because as far as I know, there's not an online source for adjusted OAPM & DAPM beyond what Eli did for 2007-08 (and maybe some of Dan Rosenbaum's early work).
Back to top
View user's profile Send private message Visit poster's website
Jacob S.
Joined: 24 Jun 2008
Posts: 2
PostPosted: Wed Jun 10, 2009 12:41 pm Post subject: Reply with quote
I agree that it's very interesting work, and I had completely missed it until now. It deserves wider viewership, IMO.
Have you looked at the collinearity between the different variables?
Back to top
View user's profile Send private message
Mike G
Joined: 14 Jan 2005
Posts: 3528
Location: Hendersonville, NC
PostPosted: Wed Jun 10, 2009 1:02 pm Post subject: Reply with quote
Whenever a site pops up that is redundant with APBRmetrics, I wonder how the proprietors hope to reach a bigger or better audience than by posting right here. Unless there's some functionality not available here? I don't see any place to put a response to the posts there.
"Basketball-Statistics.com | Innovative Stats and Analysis "
Is that not what APBRMetrics is, already? Is Scott Sereday a member and a poster? If he isn't (any longer), you have to email him. Seems like a detour or two.
Maybe people feel their submissions are more secure elsewhere. Or they don't really want any feedback?
_________________
`
36% of all statistics are wrong
Back to top
View user's profile Send private message Send e-mail
Scott S
Joined: 10 Feb 2008
Posts: 46
Location: East Rutherford, NJ
PostPosted: Wed Jun 10, 2009 9:31 pm Post subject: Reply with quote
Sorry I didn't bring this up before guys. I had planned on starting a post with the link (I feel I can get too wordy for a post at times). I was just hoping to finish some more analysis first and got busy.
To answer a few of your questions, I chose the top several players and then additional players based on reputation.
In order to obtain the adjusted offensive and defensive numbers, I used the method described by Eli W as follows:
http://www.countthebasket.com/blog/2008 ... lus-minus/
I was trying to use a method that was more similar to that done by Ilardi and Barzilai in http://www.82games.com/ilardi2.htm, but R couldn't seem to handle it.
You can see the some of the effects of collinearity between the different variables when their coefficients change after selecting different variables. I ran several different combinations to find the lowest standard errors and most reasonable interactions, etc.
I hope to post again soon guys.
Let me know if you have any other comments/suggestions/questions.
Back to top
View user's profile Send private message Visit poster's website
deepak
Joined: 26 Apr 2006
Posts: 664
PostPosted: Wed Jun 10, 2009 11:12 pm Post subject: Reply with quote
Scott S wrote:
I hope to post again soon guys.
Let me know if you have any other comments/suggestions/questions.
If possible, I'd appreciate a link to your adjusted +/- numbers for all players, instead of just a select few. Thanks for the work.
Back to top
View user's profile Send private message
Crow
Joined: 20 Jan 2009
Posts: 797
PostPosted: Thu Jun 11, 2009 3:02 am Post subject: Reply with quote
What if you split up missed field goals by location in the model?
Does the value of an offensive rebound vary much by shot location?
Not that the data is there but would a penalty for the passer to the shooter if the shot is missed be appropriate? Conceptually I think it would.
Would be very helpful to see a full team or the full league. I get the impression it is tough to end up positive. What % of players do? 20-30%?
A player that went 5-10 from field composed of 2 of 3 from inside, 2-4 mid-range (1 assisted 1 not) and 1-3 from 3 point range(assisted) with 3 assists 4 rebounds (1 offensive) 1 steal 2 turnovers and 3-4 from line is close to zero, I think... except for True Shot Attempts X (Height - 60)?
Is that True Shot Attempts X (Height - 60) thing right? Isn't that a huge factor or am I misreading?
On the defensive side steals are highly valued, with the cost of steal attempts hard to see? Ideally I'd be interested in a points allowed / steal attempt factor. Height (by itself) not that important on defense (way less than offense??)
Back to top
View user's profile Send private message
Scott S
Joined: 10 Feb 2008
Posts: 46
Location: East Rutherford, NJ
PostPosted: Thu Jun 11, 2009 7:48 pm Post subject: Reply with quote
Quote:
If possible, I'd appreciate a link to your adjusted +/- numbers for all players, instead of just a select few. Thanks for the work.
I will work on summarizing all of the results (probably after the draft)
Quote:
What if you split up missed field goals by location in the model?
I split field goals by close shots, long 2's and 3 pointers. Any other locations might be difficult to compile. Maybe in another study? I know there is some data available on nba.com, but where would I find data in a form easier to extract?
Quote:
Does the value of an offensive rebound vary much by shot location?
I think the value of an offensive rebound is strongly tied to the probability of the opposing team obtaining a rebound had it not been collected by that player. There was a study on 82 games.com that showed rebounding percentages didn't differ greatly from the three different shot "locations" I chose. I do think there are different types of shots that produce varying rebounding expectations and that these can be extracted somehow statistically, but I haven't looked into this extensively yet.
Quote:
Not that the data is there but would a penalty for the passer to the shooter if the shot is missed be appropriate? Conceptually I think it would.
This data is hard to come by. I do think that "potential assists" could help shed some light on the actual value of each assists by a player, but I don't know that assisted misses should be negatively attributed to the passer in most cases. The passer did his part by delivering the ball without a turnover. If the shooter decides not to shoot, the most significant negative value of the pass that I can think of is the time running off the shot clock to force a tougher shot later. I tried an interaction with late shot clock assists, but the play by play seems sketchy with the shot clock and I didn't get enough data. A passer should ideally be evaluated by whether his passes increase or decrease the teams scoring expectations. A good estimate should reflect this. This is why close assists are so much more valuable.
Quote:
Would be very helpful to see a full team or the full league. I get the impression it is tough to end up positive. What % of players do? 20-30%?
A player that went 5-10 from field composed of 2 of 3 from inside, 2-4 mid-range (1 assisted 1 not) and 1-3 from 3 point range(assisted) with 3 assists 4 rebounds (1 offensive) 1 steal 2 turnovers and 3-4 from line is close to zero, I think... except for True Shot Attempts X (Height - 60)?
similar percentages as regular adjusted plus/minus and probably statistics like PER. Since the best players play the most minutes, they have a higher relative impact on the average so the vast majority of players are below "average". If I had to guess, a third are above average?
Quote:
Is that True Shot Attempts X (Height - 60) thing right? Isn't that a huge factor or am I misreading?
Actually, it is a typo. It should be [TSA(ht-60)]/100. So a 7 footer and 6 footer are statistically identical and average 20 true shot attempts, this indicates we would expect the 7 footer to have a value of about 1.5 points more than the 6 footer. I initially wanted to see if there was any value for a post player shooting outside and drawing the shot blocker away from the basket, but discovered it was just more valuable in general for a big man to shoot the ball, all else being equal.
Quote:
On the defensive side steals are highly valued, with the cost of steal attempts hard to see? Ideally I'd be interested in a points allowed / steal attempt factor. Height (by itself) not that important on defense (way less than offense??)
Steals are very valuable. The high value of their coefficient can likely be attributed to a correlation between good defensive pressure and steals. This would include good defensive pressure that either forces a turnover or a lower percentage shot. The negative impact of "steals attempted" and failed is included in both Opponent PER Plus/Minus and possibly a few more variables. One factor that should be considered is that a missed steal has a limited detrimental impact because there are 4 more players available to help. I do think that someone who does it too often is clearly still hurting to the team, but haven't found any way to value this statistically yet.
The interaction between height and blocks seems to account for almost all of the value of blocks and height. I guess this means if you are very short and block shots, those blocks don't help your team very much. Similarly, it seems to mean if you are tall and no threat to block shots, your height is not very useful at all defensively, after accounting for defensive rebounding and the other variables in the model.
Back to top
View user's profile Send private message Visit poster's website
YaoPau
Joined: 28 Jan 2009
Posts: 28
PostPosted: Fri Jun 12, 2009 10:14 pm Post subject: Reply with quote
Scott, are you sure those Off/Def APM numbers are correct? You've got Dwight Howard at +1.9 offensively and +8.4 defensively, but basketballvalue has his overall APM this year at just over +1.
I tried running APM using Eli's method and got -3.59 offensively and +5.17 defensively for Howard. It's hard to say for sure whether my results are accurate, but I followed the method closely and the DiffOD results seem to match up to the on/off 82games splits I've looked at so far.
Back to top
View user's profile Send private message Visit poster's website
Scott S
Joined: 10 Feb 2008
Posts: 46
Location: East Rutherford, NJ
PostPosted: Sat Jun 13, 2009 1:27 am Post subject: Reply with quote
I ended up selecting roughly the top 71% of the players by minutes in my regression model, which very likely differs from your model. I chose the players who basketballvalue used in their model based on the file available for download. However, several players (such as Martell Webster) had a rating with just limited minutes played (5.25 minutes for him and his rating doesn't match his two year rating on the actual site). I eliminated these records in my run. Note that these ratings were not used in the estimates of the coefficients for the statistics of my model, so I didn't check them as extensively as I checked the other years. I really only posted them as a reference. So, basically, both of our models could be right or both could be wrong, etc. but I would recommend using the more consensus model in any studies that depend on these values.
Back to top
View user's profile Send private message Visit poster's website
YaoPau
Joined: 28 Jan 2009
Posts: 28
PostPosted: Sat Jun 13, 2009 1:59 am Post subject: Reply with quote
Hmm I did the same thing - using the players basketballvalue listed an APM for in the playerstats spreadsheet, although I didn't delete anybody beyond that. Why the change in Howard's APM? You've got Kobe at +13 too, when bv has him at +7.87.
Back to top
View user's profile Send private message Visit poster's website
Jon Nichols
Joined: 18 Aug 2005
Posts: 370
PostPosted: Sat Jun 13, 2009 10:08 am Post subject: Reply with quote
Mike G wrote:
Whenever a site pops up that is redundant with APBRmetrics, I wonder how the proprietors hope to reach a bigger or better audience than by posting right here. Unless there's some functionality not available here? I don't see any place to put a response to the posts there.
"Basketball-Statistics.com | Innovative Stats and Analysis "
Is that not what APBRMetrics is, already? Is Scott Sereday a member and a poster? If he isn't (any longer), you have to email him. Seems like a detour or two.
Maybe people feel their submissions are more secure elsewhere. Or they don't really want any feedback?
As the creator of Basketball-Statistics.com, I guess I might as well respond to this post and a few of the others talking about a lack of recognition. First, Scott's ideas are great. In order to maximize his exposure, I tried to set a block of time on the site where I didn't post any new entries. In retrospect, I should have posted links on this space. I think it was simply a matter of me being forgetful.
As for your thoughts about having my own web site, I understand how you can see things that way but I strongly disagree. As a matter of fact, nearly every big project I work on (Composite Score, Box Score Prediction System, PAC, etc.) is posted on this board and I take any and all comments. Every stat I have created has incorporated (usually to a considerable extent) ideas from posters here. Composite Score was almost entirely based on those ideas.
My site doesn't get a ton of recognition, but that's not from a lack of trying. I've written for NBADraft.net, RealGM.com, and now Hardwood Paroxysm with the intent of drawing attention to my work and the work of others such as Scott who have posted on my site. It's just not that easy to find a lot of people interested in this stuff.
I'm glad Scott's work has gained attention. He clearly spent a lot of time on this project. I will be sure to post any new updates from him on this board.
-Jon Nichols
Author Message
Mike G
Joined: 14 Jan 2005
Posts: 3613
Location: Hendersonville, NC
PostPosted: Sat Jun 13, 2009 11:35 am Post subject: Reply with quote
Jon,
With all sincerity, how can you 'strongly disagree' with a few questions (and an 'I wonder')?
No one has suggested it's better to not have your own site. But if worthwhile research is happening, I'm just asking if it can be expected to get better exposure elsewhere.
It's not hard to toggle between a batch of information on one site and a question/answer area elsewhere. But it does make it easier for a 'spinoff' site to lose touch, as you are aware.
_________________
`
36% of all statistics are wrong
Back to top
View user's profile Send private message Send e-mail
Jon Nichols
Joined: 18 Aug 2005
Posts: 370
PostPosted: Sat Jun 13, 2009 11:38 am Post subject: Reply with quote
Mike G wrote:
Jon,
With all sincerity, how can you 'strongly disagree' with a few questions (and an 'I wonder')?
No one has suggested it's better to not have your own site. But if worthwhile research is happening, I'm just asking if it can be expected to get better exposure elsewhere.
It's not hard to toggle between a batch of information on one site and a question/answer area elsewhere. But it does make it easier for a 'spinoff' site to lose touch, as you are aware.
Mike,
That wasn't supposed to sound harsh. I was just stating my preference for having my own site. And as you said, it is easier for a spinoff site to lose touch, but at the same time I can't post every update I make on this board. It would simply be too much. Creating a separate entity allows for the best of both worlds.
I didn't mean to get this discussion off track. I think there were some good questions...
-Jon