 |
APBRmetrics
The statistical revolution will not be televised.
|
| View previous topic :: View next topic |
| Author |
Message |
Ori
Joined: 07 Feb 2010
Posts: 9
|
 Posted: Mon Feb 08, 2010 12:45 pm Post subject: Calculating %FG percentage effect on team with weight? |
 |
|
Hi and look forward to posting on this board a lot!
I am trying to find how out big of an effect a player's shooting percentage has compared to other people in the league represented as how many deviations away from average they are. Finding the standard deviation of a players FG% compared to rest of the league is a simple excel command, however the issue arises when I try to take into account how many shot attempts they have. I have FGA and FGM data for the league, but am not sure how to do this. I have tried everything yet it appears no luck.
I know this is not the most useful stat for basketball analysis, but this is directed towards fantasy basketball and would be very helpful. Thanks.
Ori |
|
| Back to top |
|
 |
deepak
Joined: 26 Apr 2006
Posts: 655
|
| Posted: Mon Feb 08, 2010 1:53 pm Post subject: Re: Calculating %FG percentage effect on team with weight? |
|
|
| Ori wrote: | Hi and look forward to posting on this board a lot!
I am trying to find how out big of an effect a player's shooting percentage has compared to other people in the league represented as how many deviations away from average they are. Finding the standard deviation of a players FG% compared to rest of the league is a simple excel command, however the issue arises when I try to take into account how many shot attempts they have. I have FGA and FGM data for the league, but am not sure how to do this. I have tried everything yet it appears no luck.
I know this is not the most useful stat for basketball analysis, but this is directed towards fantasy basketball and would be very helpful. Thanks.
Ori |
When you say you want to take into account FGA, do you mean FGA/min or FGA/game or total FGA?
Perhaps you could standardize the FG% and the FGA, apply weights as you see fit, and just take the sum?
Something like:
| Code: |
FG% - average(all FG%) FGA - average(all FGA)
X ----------------------- + Y -----------------------
stdev(all FG%) stdev(all FGA)
|
Here is the top 20 by such a measure when I treat FGA as field goal attempts per minute, and set X = Y:
| Code: |
Player FGA_std FG%_std RTG
bryant,kobe 2.9 0.1 3.0
anthony,carmelo 2.8 0.2 3.0
wade,dwyane 2.7 0.2 2.8
james,lebron 2.2 0.6 2.8
ellis,monta 2.4 0.1 2.5
durant,kevin 2.1 0.4 2.5
bosh,chris 1.6 0.9 2.4
duncan,tim 1.5 0.9 2.4
stoudemire,amare 1.1 1.2 2.3
lee,david 1.0 1.2 2.3
nowitzki,dirk 1.9 0.3 2.2
kaman,chris 1.6 0.5 2.2
rose,derrick 1.9 0.2 2.1
jefferson,al 1.6 0.5 2.1
johnson,joe 1.9 0.1 2.1
boozer,carlos 1.0 1.1 2.1
landry,carl 0.9 1.1 2.0
randolph,zach 1.4 0.5 1.9
mohammed,nazr 0.5 1.4 1.9
arenas,gilbert 2.3 -0.4 1.9
|
Wow, a stat where LeBron doesn't lead the league. Maybe I should rethink it.  |
|
| Back to top |
|
|
Ori
Joined: 07 Feb 2010
Posts: 9
|
| Posted: Mon Feb 08, 2010 3:51 pm Post subject: |
|
|
Thank you for the reply!
However, I think I worded my original question poorly. Let me compare what I am looking for in terms of another stat, rebounds. For rebounds I simply found the stdev and average of the league for reb/game for each player. Then took the per game average, subtracted from the average and divided by stdev to find the score. Now what I'm looking for is a comparable score in terms of FG and FT percentage. These scores are summed with the other scores (Reb,ast,stl,etc) to form the players overall score, from which I determine ranking. I am having trouble determining how to form a rating for the player similar to the other ones.
The key issue is I need a rating that takes into account how good their FG percentage is and how often they shoot each game.
I am not entirely sure if they use stdev, but I'm not sure there is another way to keep FG and FT percentage equal to the rest of the categories (ast,stl,blk,etc) in terms of overall rating.
For 2010 Season, here is the top 5:
D. Howard 3.76
D. Lee 3.75
A. Stoude 3.57
M. Gasol 3.43
K. Perkins 3.28
For 2010 Average, here is the top 5:
D. Lee 3.25
D. Howard 3.12
A. Stoude 3.00
M. Gasol 2.90
K. Perkins 2.79
I know this analysis is completely different, but I am trying to apply a advanced stats mini revolution to fantasy basketball but need to first fully replicate rankings for personal use. Thanks for reading this far if you've made it,
Ori
P.S. I tried using [code] and it seems it just makes the code smaller font, but for you guys the columns are always aligned. What gives!  |
|
| Back to top |
|
|
deepak
Joined: 26 Apr 2006
Posts: 655
|
| Posted: Mon Feb 08, 2010 5:35 pm Post subject: |
|
|
| Ori wrote: | The key issue is I need a rating that takes into account how good their FG percentage is and how often they shoot each game.
I am not entirely sure if they use stdev, but I'm not sure there is another way to keep FG and FT percentage equal to the rest of the categories (ast,stl,blk,etc) in terms of overall rating.
For 2010 Season, here is the top 5:
D. Howard 3.76
D. Lee 3.75
A. Stoude 3.57
M. Gasol 3.43
K. Perkins 3.28
For 2010 Average, here is the top 5:
D. Lee 3.25
D. Howard 3.12
A. Stoude 3.00
M. Gasol 2.90
K. Perkins 2.79
I know this analysis is completely different, but I am trying to apply a advanced stats mini revolution to fantasy basketball but need to first fully replicate rankings for personal use. Thanks for reading this far if you've made it,
Ori
P.S. I tried using [code] and it seems it just makes the code smaller font, but for you guys the columns are always aligned. What gives! |
Maybe this would help (I found through google):
http://www.sportscomet.com/Fantasy-Sports/65214.htm
To replicate the Yahoo fantasy ratings, someone found this to be accurate:
Fantasy production (Note: All cats use standard deviation; not actual stat!!) = 0.43FG% + 0.46 FT% + 1.13(3ptm) + 0.98pts + 0.91reb + 1.10ast + 0.90stl + 1.18blk - 1.00to
I'm not sure why the code tag doesn't work for you. |
|
| Back to top |
|
|
Ori
Joined: 07 Feb 2010
Posts: 9
|
| Posted: Mon Feb 08, 2010 9:57 pm Post subject: |
|
|
| deepak_e wrote: | | Ori wrote: | The key issue is I need a rating that takes into account how good their FG percentage is and how often they shoot each game.
I am not entirely sure if they use stdev, but I'm not sure there is another way to keep FG and FT percentage equal to the rest of the categories (ast,stl,blk,etc) in terms of overall rating.
For 2010 Season, here is the top 5:
D. Howard 3.76
D. Lee 3.75
A. Stoude 3.57
M. Gasol 3.43
K. Perkins 3.28
For 2010 Average, here is the top 5:
D. Lee 3.25
D. Howard 3.12
A. Stoude 3.00
M. Gasol 2.90
K. Perkins 2.79
I know this analysis is completely different, but I am trying to apply a advanced stats mini revolution to fantasy basketball but need to first fully replicate rankings for personal use. Thanks for reading this far if you've made it,
Ori
P.S. I tried using [code] and it seems it just makes the code smaller font, but for you guys the columns are always aligned. What gives! |
Maybe this would help (I found through google):
http://www.sportscomet.com/Fantasy-Sports/65214.htm
To replicate the Yahoo fantasy ratings, someone found this to be accurate:
Fantasy production (Note: All cats use standard deviation; not actual stat!!) = 0.43FG% + 0.46 FT% + 1.13(3ptm) + 0.98pts + 0.91reb + 1.10ast + 0.90stl + 1.18blk - 1.00to
I'm not sure why the code tag doesn't work for you. |
Thank you, appreciate it! It seems to have gotten a little closer, but I would still like to find out if its possible to find a stdev rating that takes into account shot attempts and fg percentage because that number would be extremely useful in comparing values of players. I'm not even sure its possible to represent something like this through standard deviation but it would be very much appreciated if possible. thanks.
Ori |
|
| Back to top |
|
|
deepak
Joined: 26 Apr 2006
Posts: 655
|
| Posted: Mon Feb 08, 2010 11:20 pm Post subject: |
|
|
| Ori wrote: | Thank you, appreciate it! It seems to have gotten a little closer, but I would still like to find out if its possible to find a stdev rating that takes into account shot attempts and fg percentage because that number would be extremely useful in comparing values of players. I'm not even sure its possible to represent something like this through standard deviation but it would be very much appreciated if possible. thanks.
Ori |
Why do you think its necessary to take into account shot attempts? Its not clear what exactly you are looking for.
I mean, the simplest way to combine the two is to just look at FGM (FG%*FGA). |
|
| Back to top |
|
|
Ori
Joined: 07 Feb 2010
Posts: 9
|
| Posted: Mon Feb 08, 2010 11:54 pm Post subject: |
|
|
| deepak_e wrote: | | Ori wrote: | Thank you, appreciate it! It seems to have gotten a little closer, but I would still like to find out if its possible to find a stdev rating that takes into account shot attempts and fg percentage because that number would be extremely useful in comparing values of players. I'm not even sure its possible to represent something like this through standard deviation but it would be very much appreciated if possible. thanks.
Ori |
Why do you think its necessary to take into account shot attempts? Its not clear what exactly you are looking for.
I mean, the simplest way to combine the two is to just look at FGM (FG%*FGA). |
I am trying to quantify how good a player is at raising his fantasy teams shooting percentage.
I assume this means a player with a fg%=80 on 2 FGA is not as valuable as a player who shoots 60% on 30 FGA. I am not sure how to make the latter player shooting 60 percent appear as more valuable than the player shooting 80 percent.
In every other fantasy category, standard deviation is used to represent a players value in that category. Preferably, this new FG% value would be a deviation so it would be comparable to the rest of the stats. Thanks again. |
|
| Back to top |
|
|
deepak
Joined: 26 Apr 2006
Posts: 655
|
| Posted: Tue Feb 09, 2010 1:14 am Post subject: |
|
|
| Ori wrote: |
I am trying to quantify how good a player is at raising his fantasy teams shooting percentage. |
I see. That's going to depend on the FG% and FGA of the rest of the players on the fantasy team.
Suppose your player shoots fg% on fga attempts. And the rest of the players on the team shoot others_FG% on others_FGA attempts. Improvement would be:
improvement% = fg% * X - others_FG% * X
where X = fga / (fga + others_FGA)
You could come up with a generic formula by considering, on average, what others_FG% and others_FGA would be. |
|
| Back to top |
|
|
Ori
Joined: 07 Feb 2010
Posts: 9
|
| Posted: Tue Feb 09, 2010 3:51 pm Post subject: |
|
|
| deepak_e wrote: | | Ori wrote: |
I am trying to quantify how good a player is at raising his fantasy teams shooting percentage. |
I see. That's going to depend on the FG% and FGA of the rest of the players on the fantasy team.
Suppose your player shoots fg% on fga attempts. And the rest of the players on the team shoot others_FG% on others_FGA attempts. Improvement would be:
improvement% = fg% * X - others_FG% * X
where X = fga / (fga + others_FGA)
You could come up with a generic formula by considering, on average, what others_FG% and others_FGA would be. |
I tried this and it gives me data that is close to how ESPN analyses FG% and FT% (they give numerical rating I assume I need to match). However, I've been looking at your formula and tell me if this issue makes sense:
Let's assume the league shoots 20 FGA and makes 10 FGM for 50% shooting.
Player A shoots 10 percent above average at 60% shooting on 0.6/1.0 FGA.
Player B shoots 5 percent above average at 55% shooting but shoots twice as often at 1.1/2.0 shots a game.
New Shooting with Player A = 0.5047
New Shooting with Player B = 0.504545
Therefore, player A has a bigger impact on the team. However, your formula can be reduced to (fg%-fg%avg)(fga/total fga).
If we plug in the numbers above, they should have the exact same value, right? But above we saw that player A would have a better, if only slightly, effect on the team.
Here are the top 5 fantasy FG% players according to ESPN with your top 5 to the right.
| Code: |
ESPN deepak_e
D. Lee D. Lee
D. Howard A. Stoudemire
A. Stoudemire D. Howard
M. Gasol C. Boozer
K. Perkins M. Gasol
|
Obviously ESPN can be wrong, and I may be totally wrong (if yours is right, I apologize for all this), but it appears your formula gives too much weight to high FGA as Boozer and Amare are overly high on yours and Perkins, Gasol, and Howard (top 3 in league FG%) are lower.
This makes sense as discussed before since we saw the player shooting 10 percentage points better on 1 FGA is more useful than 5 percentage points better on 2 FGA, but your formula rates them evenly, giving the advantage to lower FG% and higher FGA players.
Appreciate the response and input and thanks for the help deepak_e. |
|
| Back to top |
|
|
deepak
Joined: 26 Apr 2006
Posts: 655
|
| Posted: Tue Feb 09, 2010 6:43 pm Post subject: |
|
|
| Ori wrote: |
Let's assume the league shoots 20 FGA and makes 10 FGM for 50% shooting.
Player A shoots 10 percent above average at 60% shooting on 0.6/1.0 FGA.
Player B shoots 5 percent above average at 55% shooting but shoots twice as often at 1.1/2.0 shots a game.
New Shooting with Player A = 0.5047
New Shooting with Player B = 0.504545
Therefore, player A has a bigger impact on the team. However, your formula can be reduced to (fg%-fg%avg)(fga/total fga).
If we plug in the numbers above, they should have the exact same value, right? But above we saw that player A would have a better, if only slightly, effect on the team. |
If you plug in the numbers into that formula, you don't get the same result for both players. Your assuming "total fga" is the same, but actually its not because Player A shoots 1 FGA less than Player B. |
|
| Back to top |
|
|
Ori
Joined: 07 Feb 2010
Posts: 9
|
| Posted: Tue Feb 09, 2010 7:23 pm Post subject: |
|
|
| deepak_e wrote: | | Ori wrote: |
Let's assume the league shoots 20 FGA and makes 10 FGM for 50% shooting.
Player A shoots 10 percent above average at 60% shooting on 0.6/1.0 FGA.
Player B shoots 5 percent above average at 55% shooting but shoots twice as often at 1.1/2.0 shots a game.
New Shooting with Player A = 0.5047
New Shooting with Player B = 0.504545
Therefore, player A has a bigger impact on the team. However, your formula can be reduced to (fg%-fg%avg)(fga/total fga).
If we plug in the numbers above, they should have the exact same value, right? But above we saw that player A would have a better, if only slightly, effect on the team. |
If you plug in the numbers into that formula, you don't get the same result for both players. Your assuming "total fga" is the same, but actually its not because Player A shoots 1 FGA less than Player B. |
I think I realized I may have misunderstood a part of the formula and thus oversimplified. I need others_fg% and (fga+others_FGA). others_FG% is the FG% average of the rest of the league including the player (what ive done) or excluding the player (i just realized)? The other thing, (fga+others_FGA), is that player A's FGA + others_FGA/game or player A's FGA+others_FGA. In the former, the parenthesis changes with each person (what I just realized) whereas in the latter the parenthesis becomes static since there is always the same amount of FGA as far as my data goes and then total doesnt change for me. |
|
| Back to top |
|
|
findingneema
Joined: 25 Feb 2008
Posts: 34
Location: Atlanta, GA
|
| Posted: Thu Feb 11, 2010 1:34 am Post subject: |
|
|
This is why last season I wrote a different sort of fantasy rankings algorithm - one that simulates putting teams together and actually calculating a VORP for each player. Worked pretty well, and was very similar to overall yahoo rankings.
Here's what I would suggest:
1) Come up with an average fantasy team.
2) Add the FG and FT stats for player in question.
3) Repeat for all players.
4) Now you have a mean and a stdev.
For future reference, you could just say you're calculating the z-score for each player in a category (normalizing the score by stdev).
If you wanted to take this further, generate a bunch of fantasy teams and repeat the calcs for each. You could even break down how each player contributes based on team % and volume. |
|
| Back to top |
|
|
|
|
You cannot post new topics in this forum
You cannot reply to topics in this forum
You cannot edit your posts in this forum
You cannot delete your posts in this forum
You cannot vote in polls in this forum
|
Powered by phpBB © 2001, 2005 phpBB Group
|
FAQ
Search
Memberlist
Usergroups
Register
Profile
Log in to check your private messages
Log in
