Newbie on basketball statistics

Home for all your discussion of basketball statistical analysis.
josecarlos
Posts: 98
Joined: Sun Oct 08, 2017 10:11 pm
Location: Spain
Contact:

Re: Newbie on basketball statistics

Post by josecarlos »

Hi everybody!!!

I'm here again with my doubts!!! :) I'm making calculations about Offensive Rating and Defensive Rating aplying Dean Oliver's formulas. With Offensive Rating I haven't any doubt but with Defensive Rating ... I've got strange results in some cases. I'm applying the formula just only for a game right now, so ... ¿is it possible that I've got strange results because I'm making calulations for only one game? The strange result is that I've got a negative defensive rating with a player that only play two minutes. I have revised the formula several times, and all the intermedia calculations are right but the negative result occurs when I calculate the last step:

DRtg = Team_Defensive_Rating + 0.2 * (100 * D_Pts_per_ScPoss * (1 - Stop%) - Team_Defensive_Rating

So, what can I do in this cases? I mean, what is the more elegant solution to show in this cases? Show a negative value? Convert negative values to zero? Don't show anything?

And my second question about Offensive Rating and Defensive Rating is ... When I will make calculations to get the Offensive Rating or Defensive Rating about 5, 10, 20 games or all a regular season. What is better ... to sum all the Offensive Rating and Defensive Rating of each game and get the average or apply the formulas to all the games? I guess is the second option, but I would like to know how do you work in this cases if you previously have all the Offensive Rating and Defensive Rating previously calculated by game.

That's all!!! Thanks in advance for your appreciated help!!!
eminence
Posts: 223
Joined: Sun Sep 10, 2017 8:20 pm

Re: Newbie on basketball statistics

Post by eminence »

Could you share the inputs for the player getting the negative result?
josecarlos
Posts: 98
Joined: Sun Oct 08, 2017 10:11 pm
Location: Spain
Contact:

Re: Newbie on basketball statistics

Post by josecarlos »

eminence wrote:Could you share the inputs for the player getting the negative result?
Yes, of course!!! Here is the data ...

Code: Select all

Player
======
STL = 2
BLK = 2
DRB = 1
MP = 2
PF = 1

Team
=====
Team_DRB = 22
Team_BLK = 2
Team_MP = 200
Team_STL = 8
Team_PF = 19
Team_Possessions = 71

Opponent Team
==============
Opponent_ORB = 8
Opponent_FGM = 23
Opponent_FGA = 53
Opponent_TOV = 20
Opponent_FTA = 16
Opponent_FTM = 12
Opponent_PTS = 61

Intermedia calculations
=========================
DOR%: 0.26666666666666666
DFG%: 0.4339622641509434
FMwt: 0.6782841823056301
Stops1: 2.3217158176943697
Stops2: 0.2767818183528527
Stops: 2.5984976360472225
Stop%: 3.659855825418623
Team_Defensive_Rating: 85.91549295774648
D_Pts_per_ScPoss: 2.103448275862069

Final Result
===============
DRtg: -43.164988634172474
josecarlos
Posts: 98
Joined: Sun Oct 08, 2017 10:11 pm
Location: Spain
Contact:

Re: Newbie on basketball statistics

Post by josecarlos »

josecarlos wrote:Hi everybody!!!

I'm here again with my doubts!!! :) I'm making calculations about Offensive Rating and Defensive Rating aplying Dean Oliver's formulas. With Offensive Rating I haven't any doubt but with Defensive Rating ... I've got strange results in some cases. I'm applying the formula just only for a game right now, so ... ¿is it possible that I've got strange results because I'm making calulations for only one game? The strange result is that I've got a negative defensive rating with a player that only play two minutes. I have revised the formula several times, and all the intermedia calculations are right but the negative result occurs when I calculate the last step:

DRtg = Team_Defensive_Rating + 0.2 * (100 * D_Pts_per_ScPoss * (1 - Stop%) - Team_Defensive_Rating

So, what can I do in this cases? I mean, what is the more elegant solution to show in this cases? Show a negative value? Convert negative values to zero? Don't show anything?

And my second question about Offensive Rating and Defensive Rating is ... When I will make calculations to get the Offensive Rating or Defensive Rating about 5, 10, 20 games or all a regular season. What is better ... to sum all the Offensive Rating and Defensive Rating of each game and get the average or apply the formulas to all the games? I guess is the second option, but I would like to know how do you work in this cases if you previously have all the Offensive Rating and Defensive Rating previously calculated by game.

That's all!!! Thanks in advance for your appreciated help!!!
Anyboby has any answer about my questions?
Crow
Posts: 10536
Joined: Thu Apr 14, 2011 11:10 pm

Re: Newbie on basketball statistics

Post by Crow »

"is it possible that I've got strange results because I'm making calulations for only one game?"

Yes. And for only 2 minutes. 2 blocks and 2 steals in 2 minutes is just tremendous low sample size results and sorta breaks the normal scale. Accept that as a relative result and don't worry about it being negative.

"What is better ... to sum all the Offensive Rating and Defensive Rating of each game and get the average or apply the formulas to all the games?" I am not sure there is a certain "better" answer. Applying the formula to all games at once is most common but players actually play in and impact individual games. If it is easier for you, do the sum and average. In long run the difference will probably be small. The same method is applied to everyone. Might make slight difference between most consistent player and most inconsistent (consistent with themselves game to game or consistent with team within games) but I don't think many cases would be affected much. If you did it both ways you could find out and probably be the only one who knows.

Your questions show thoroughness, attention to detail. Asking questions is good but won't always work. Trust your own judgments. Lay out your methods and if someone offers a suggestion or opinion, listen to it & evaluate it. Go where you want with your analysis. Certainly don't let non-response to questions stop you for long. Do your best and listen. Asking twice shows persistence and persistence is good but even that won't always work. If you can't get an answer here, try reddit, real gm or use twitter to ask someone good at analysis. Not many post here anymore. Read here? Some still do, some don't.
josecarlos
Posts: 98
Joined: Sun Oct 08, 2017 10:11 pm
Location: Spain
Contact:

Re: Newbie on basketball statistics

Post by josecarlos »

Crow wrote:"is it possible that I've got strange results because I'm making calulations for only one game?"

Yes. And for only 2 minutes. 2 blocks and 2 steals in 2 minutes is just tremendous low sample size results and sorta breaks the normal scale. Accept that as a relative result and don't worry about it being negative.

"What is better ... to sum all the Offensive Rating and Defensive Rating of each game and get the average or apply the formulas to all the games?" I am not sure there is a certain "better" answer. Applying the formula to all games at once is most common but players actually play in and impact individual games. If it is easier for you, do the sum and average. In long run the difference will probably be small. The same method is applied to everyone. Might make slight difference between most consistent player and most inconsistent (consistent with themselves game to game or consistent with team within games) but I don't think many cases would be affected much. If you did it both ways you could find out and probably be the only one who knows.

Your questions show thoroughness, attention to detail. Asking questions is good but won't always work. Trust your own judgments. Lay out your methods and if someone offers a suggestion or opinion, listen to it & evaluate it. Go where you want with your analysis. Certainly don't let non-response to questions stop you for long. Do your best and listen. Asking twice shows persistence and persistence is good but even that won't always work. If you can't get an answer here, try reddit, real gm or use twitter to ask someone good at analysis. Not many post here anymore. Read here? Some still do, some don't.
Thank you so much Crow for your help again!!! You are very kind with me and my doubts!!! :)))

Do you have any suggestion about any thread in reddit?
Crow
Posts: 10536
Joined: Thu Apr 14, 2011 11:10 pm

Re: Newbie on basketball statistics

Post by Crow »

I don't know much about reddit but you could start here and explore https://www.reddit.com/r/NBAanalytics/

At realGM this is the place https://forums.realgm.com/boards/viewforum.php?f=344
josecarlos
Posts: 98
Joined: Sun Oct 08, 2017 10:11 pm
Location: Spain
Contact:

Re: Newbie on basketball statistics

Post by josecarlos »

Crow wrote:I don't know much about reddit but you could start here and explore https://www.reddit.com/r/NBAanalytics/

At realGM this is the place https://forums.realgm.com/boards/viewforum.php?f=344
Thank you!!!
josecarlos
Posts: 98
Joined: Sun Oct 08, 2017 10:11 pm
Location: Spain
Contact:

Re: Newbie on basketball statistics

Post by josecarlos »

Crow wrote: Mon Apr 16, 2018 4:52 pm "is it possible that I've got strange results because I'm making calulations for only one game?"

Yes. And for only 2 minutes. 2 blocks and 2 steals in 2 minutes is just tremendous low sample size results and sorta breaks the normal scale. Accept that as a relative result and don't worry about it being negative.

"What is better ... to sum all the Offensive Rating and Defensive Rating of each game and get the average or apply the formulas to all the games?" I am not sure there is a certain "better" answer. Applying the formula to all games at once is most common but players actually play in and impact individual games. If it is easier for you, do the sum and average. In long run the difference will probably be small. The same method is applied to everyone. Might make slight difference between most consistent player and most inconsistent (consistent with themselves game to game or consistent with team within games) but I don't think many cases would be affected much. If you did it both ways you could find out and probably be the only one who knows.

Your questions show thoroughness, attention to detail. Asking questions is good but won't always work. Trust your own judgments. Lay out your methods and if someone offers a suggestion or opinion, listen to it & evaluate it. Go where you want with your analysis. Certainly don't let non-response to questions stop you for long. Do your best and listen. Asking twice shows persistence and persistence is good but even that won't always work. If you can't get an answer here, try reddit, real gm or use twitter to ask someone good at analysis. Not many post here anymore. Read here? Some still do, some don't.
Hi!!!

Several years ago I made a question about what is better to make calculus for a whole season. I did it because a player or a team can play different phases or competitions in the same season. For example, in USA you can play, regular-season, quarter-finals play-offs , semi-finals play-offs, final conference and then NBA final. In Europe, we've got regular season, cups of each country, quarter-finals play-offs, semi-finals play-offs and finals. So, you can get totals about advanced stats associated to each phase of the competition. But, If you want to know the advanced stats of a whole season from a team or player, you've got two ways to do it.

First, sum up all the stats of each phase and then divide by the number of phases played or recalculate all the advanced stats from all the games. And my question was ... Which method is better?

I have tested that If you decide to sump yo all the stats of each phase and then divide by the number of phases played you can have differences with the data that you've get if you recalculate the advanced stats considering all the games. The differences oscilating between 0,5 or 0,7 points in some stats until 5 or 7 points of difference in other advanced stats.

Therefore, in my humble opinion that is better to recalculate all the stats than to make the average of the data got it in each phase of the competition.
josecarlos
Posts: 98
Joined: Sun Oct 08, 2017 10:11 pm
Location: Spain
Contact:

Re: Newbie on basketball statistics

Post by josecarlos »

Several years ago too, I asked about if anyone had experience or knowledge about how to make a statistical report for a team.

Before to start, Crow recommended me to contact with Ben Falk. I wrote him and ... I never got any response about him :( It doesn't matter!!! I appreciated the help from Crow always!!!

Ok, I would like to share with you right now, which reports you can make with data from boxscore and data. from shot chart. I have no access to any other kind of data :(

I have checked that teams usually want to have a pre-game report, a post-game report and a monthly report.

The structure of the reports are quite similar with some little differences. An example of pre-game report would be:

1. Comparison stats from boxscore of both teams (total shoots, shoots per game, total rebounds, rebounds per game, total of other stats and other stats per game)

2. Comparison stats of advanced stats of both teams:
  • 4 factors, possessions
  • Pace, points by possession, points by attempt, distribution of possesssions,
  • Efficience (ortg, drtg, nrtg)
  • Advanced stats of shots
3. Shot chart of opponent team. In that shot chart you've got the field divided by zones and stats in each zone. This stats would be:
  • % of shots made in this zone
  • % of shots scored in this zone
  • % of shot scored in this zone by all the players of the league
4. Comparative graphics of the two teams
  • Distribution of possessions
  • Distribution of points
  • Pace, points per attempt, point per possession
  • Rebounds per game
5. For each player of the opponent team:
  • Stats from boxscore
  • total shoots,
  • shoots per game,
  • total rebounds,
  • rebounds per game,
  • total of other stats
  • other stats per game
  • Advanced stats
  • Shot chart of the player
6. A graphic with the distrubution of points of players from the opponent team.
7. A glosary with all the statistical terms

I'm sure that is possible to improve it so any suggestion would be welcome, of course!!!

I hope it helps!!!
Crow
Posts: 10536
Joined: Thu Apr 14, 2011 11:10 pm

Re: Newbie on basketball statistics

Post by Crow »

Prepare what you like for an article (published wherever you can / think makes sense) or maybe find a lower level team (not in NBA) who is interested in customized reports. Many teams may already have what they want in reports though. Some may not be as serious / thorough as perhaps they should be. You have to decide what is worth your time doing or trying to do. But if it interests you and builds skills, do as you wish.

Help may come or not for various reasons. Use what you can find and carry on.
Crow
Posts: 10536
Joined: Thu Apr 14, 2011 11:10 pm

Re: Newbie on basketball statistics

Post by Crow »

Ways to make team contacts: Coaches on twitter, coaching clinics, coaching associations, analytic conferences, email at team sites, etc.
josecarlos
Posts: 98
Joined: Sun Oct 08, 2017 10:11 pm
Location: Spain
Contact:

Re: Newbie on basketball statistics

Post by josecarlos »

Crow wrote: Mon Aug 10, 2020 11:36 pm Ways to make team contacts: Coaches on twitter, coaching clinics, coaching associations, analytic conferences, email at team sites, etc.
My experience trying to contact with coaches or people who works in analytics is different.

I have contacted with coaches. My experience is that they are more opened to hear and ask what they want than people who works in analitycs for teams.

I tried to contact with two people who are working in analytics for NBA team. One of them never responsed to me, the other responsed to me but she couldn't help me beucause she doesn't dedicated to analyze the game through the stats. But, at least, she answered me!!! :)
SuperDuper
Posts: 2
Joined: Thu Apr 02, 2020 2:59 pm

Re: Newbie on basketball statistics

Post by SuperDuper »

This is great. Analytics needs coaches though. They need to implement what they see in the numbers on the court.

Also how do stats explain the odds on Lebron's 10th NBA finals appearance?

https://www.sportsbetting3.com/nba/nba-betting#futures

Only Boston/Miami coaching superiority can stop Lakers but I doubt that.
Mike G
Posts: 6144
Joined: Fri Apr 15, 2011 12:02 am
Location: Asheville, NC

Re: Newbie on basketball statistics

Post by Mike G »

error
no odds yet
??

b-r.com gives LA a .922 chance of reaching the Finals and .585 of winning it.
https://www.basketball-reference.com/fr ... _prob.html
So IF they get there, it's a (585/922) .634 chance they win.
Post Reply