APBRmetrics

The discussion of the analysis of basketball through objective evidence, especially basketball statistics.
It is currently Wed Oct 01, 2014 6:47 am

All times are UTC




Post new topic Reply to topic  [ 234 posts ]  Go to page 1, 2, 3, 4, 5 ... 16  Next
Author Message
 Post subject: Power Ranking
PostPosted: Tue Dec 27, 2011 4:53 pm 
Offline

Joined: Fri Apr 15, 2011 8:28 am
Posts: 635
I've put Power Rankings (done with Ridge Regression) on my site.

http://stats-for-the-nba.appspot.com/power_rankings

These ratings have a lower out-of-sample MSE than bbr's SRS or straight up point differential.

Obviously, you won't be able to draw any meaningful conclusions from those, yet (except that Dallas had a bad start)

I'll try to update them at least every second day



I'll try to add power rankings based on player RAPM and minute allocation of the last x games in a couple of weeks

_________________
http://stats-for-the-nba.appspot.com/


Top
 Profile  
 
 Post subject: Re: Power Ranking
PostPosted: Tue Dec 27, 2011 8:33 pm 
Offline

Joined: Thu Apr 21, 2011 8:25 pm
Posts: 179
Location: Boone, NC
Wouldn't your multiyear rAPM (using projected minutes%) be better at predicting future games?

I was thinking about semi-automatically-updating these myself.


EDIT: Nevermind. You addressed this! Sweet. I'm gonna go ahead and check it out for myself anyways.

DOUBLE EDIT: Are you going to add rookies in soon?


Top
 Profile  
 
 Post subject: Re: Power Ranking
PostPosted: Tue Dec 27, 2011 10:54 pm 
Offline

Joined: Thu Apr 14, 2011 10:41 pm
Posts: 815
Location: Hotlanta
Well, looks like I wasted a few hours working on this. Good news is now I don't need to spend more time on it. :)

_________________
The City: A Golden State Warriors-Centric NBA Blog
NBA WOWY: Find statistics of your favorite team with any arbitrary combination of players on or off the court.


Top
 Profile  
 
 Post subject: Re: Power Ranking
PostPosted: Wed Dec 28, 2011 10:35 pm 
Offline

Joined: Fri Apr 15, 2011 8:28 am
Posts: 635
Now with team strength estimation based on RAPM player values and playing time allocation of recent games.
Minutes of more recent games will be weighed more heavily (with so few game played it doesn't matter, yet)
I don't have the time to research the perfect weighing scheme, so I'll use an arbitrary one. Once there are more games I want to use the last ~6 games and weigh them 60-50-40-30-20-10. If someone has a better idea, let me hear it

http://stats-for-the-nba.appspot.com/power_rankings
I'm sorry Evan.. (and other GSW fans)

_________________
http://stats-for-the-nba.appspot.com/


Top
 Profile  
 
 Post subject: Re: Power Ranking
PostPosted: Thu Dec 29, 2011 1:05 am 
Offline

Joined: Thu Apr 14, 2011 10:41 pm
Posts: 815
Location: Hotlanta
J.E. wrote:

http://stats-for-the-nba.appspot.com/power_rankings
I'm sorry Evan.. (and other GSW fans)


GSW is actually playing quite a bit better than I was expecting. I think Rush, Kwame, and Biedrins are playing better than their RAPM ratings. We'll see how long that lasts. I guess the positive spin on this may be that Jackson is getting more productivity than would be expected from the lineups he's throwing out there.

_________________
The City: A Golden State Warriors-Centric NBA Blog
NBA WOWY: Find statistics of your favorite team with any arbitrary combination of players on or off the court.


Top
 Profile  
 
 Post subject: Re: Power Ranking
PostPosted: Thu Dec 29, 2011 1:17 am 
Offline

Joined: Thu Apr 14, 2011 10:41 pm
Posts: 815
Location: Hotlanta
I count only 21 teams on your list, btw.

_________________
The City: A Golden State Warriors-Centric NBA Blog
NBA WOWY: Find statistics of your favorite team with any arbitrary combination of players on or off the court.


Top
 Profile  
 
 Post subject: Re: Power Ranking
PostPosted: Thu Dec 29, 2011 11:57 am 
Offline

Joined: Fri Apr 15, 2011 8:28 am
Posts: 635
EvanZ wrote:
GSW is actually playing quite a bit better than I was expecting. I think Rush, Kwame, and Biedrins are playing better than their RAPM ratings. We'll see how long that lasts. I guess the positive spin on this may be that Jackson is getting more productivity than would be expected from the lineups he's throwing out there.

I wasn't being completely serious. With so few games to predict minute allocation from, the ratings derived from player RAPM will go all over the place in the first week.

Seems like Boston is missing Paul Pierce
Denver is rolling just like at the end of last season
Atlanta has a +27 Point Differential, but probably had the easiest schedule (NJN, WAS)
NOH is at 2-0 and has 3 extremely easy games coming up, they could be 5-0 after those. They'll have a brutal stretch after that though

EDIT: Looking at the team ratings derived from RAPM I would say that it is already mostly in agreement with what various predictions said about the teams. Miami, Denver, Chicago, OKC, Boston, LAC, SAS, ORL, NYK, MIL, LAL, (MEM), PHX, NOH, MIN, DET, CHA. All their ratings are at least in proximity of where people predicted them to be. Houston and the Pacers look a little off(all too low), and DAL and GSW are way off..
Probably not a good sign for GSW

_________________
http://stats-for-the-nba.appspot.com/


Top
 Profile  
 
 Post subject: Re: Power Ranking
PostPosted: Thu Dec 29, 2011 1:51 pm 
Offline

Joined: Thu Apr 14, 2011 10:41 pm
Posts: 815
Location: Hotlanta
Jerry, why is there only 21 teams in your ranking?

_________________
The City: A Golden State Warriors-Centric NBA Blog
NBA WOWY: Find statistics of your favorite team with any arbitrary combination of players on or off the court.


Top
 Profile  
 
 Post subject: Re: Power Ranking
PostPosted: Thu Dec 29, 2011 2:05 pm 
Offline

Joined: Fri Apr 15, 2011 8:28 am
Posts: 635
EvanZ wrote:
Jerry, why is there only 21 teams in your ranking?

Because, until a couple of hours ago, not all teams were in the bbv data set for 2012. With no information about minute allocation I left those teams out for now. Should be fixed when I upload new rankings, which should be in ~6 hours

_________________
http://stats-for-the-nba.appspot.com/


Top
 Profile  
 
 Post subject: Re: Power Ranking
PostPosted: Thu Dec 29, 2011 8:23 pm 
Offline

Joined: Fri Apr 15, 2011 8:28 am
Posts: 635
http://stats-for-the-nba.appspot.com/power_rankings

Updated (now with all teams!) and fixed a bug that made all RAPM derived ratings twice as big as they should have been

Delonte West plays lots of minutes for Dallas, dragging their RAPM derived rating down. I looked it up because I found it weird that RAPM gave them a rating of -0.1. He's probably not the (only) problem though.

According to RAPM, Indiana, New Orleans, Cleveland, Toronto, Minnesota, Charlotte, Golden State, Sacramento have overachieved. Miami, Oklahoma, Chicago, Boston and Dallas are the biggest underachievers

Teams to look out for are Atlanta, currently at #1, Philly at #7, Denver at #4. I guess Philly and Denver might be able to hold their spots

_________________
http://stats-for-the-nba.appspot.com/


Top
 Profile  
 
 Post subject: Re: Power Ranking
PostPosted: Thu Dec 29, 2011 8:35 pm 
Offline

Joined: Thu Apr 21, 2011 8:25 pm
Posts: 179
Location: Boone, NC
For my own benefit, I will probably look at (GP/66)*TeamRating + (1-GP/66)*rAPMRating


Top
 Profile  
 
 Post subject: Re: Power Ranking
PostPosted: Thu Dec 29, 2011 8:46 pm 
Offline

Joined: Fri Apr 15, 2011 8:28 am
Posts: 635
bbstats wrote:
For my own benefit, I will probably look at (GP/66)*TeamRating + (1-GP/66)*rAPMRating

I like the RAPM rating for it's ability to quickly adjust after a trade or injury of a key player. If a team makes no trades and is not hit by major injuries, then the simple ridge regressed team rating is probably the way to go for that team. But in such a case, I'm pretty sure, the two ratings won't differ much anyway (after a healthy amount of games, anyway).

There are ways to incorporate recency into the ridge regressed team ratings, like Evan has done with bootstrapping, but I don't have the time for researching this right now

_________________
http://stats-for-the-nba.appspot.com/


Top
 Profile  
 
 Post subject: Re: Power Ranking
PostPosted: Thu Dec 29, 2011 9:07 pm 
Offline

Joined: Thu Apr 14, 2011 10:41 pm
Posts: 815
Location: Hotlanta
J.E. wrote:

According to RAPM, Indiana, New Orleans, Cleveland, Toronto, Minnesota, Charlotte, Golden State, Sacramento have overachieved. Miami, Oklahoma, Chicago, Boston and Dallas are the biggest underachievers

Teams to look out for are Atlanta, currently at #1, Philly at #7, Denver at #4. I guess Philly and Denver might be able to hold their spots


I'm just looking at GSW as an example. They're power rating (1) is -0.4 and their RAPM-minutes rating (2) or whatever you're calling it is -3.8. Doesn't the fact that the first rating is higher than the second one mean they are *overachieving*? If not, maybe I'm not understanding this properly.

On bootstrapping...man, that might take a while with a full season of samples.

_________________
The City: A Golden State Warriors-Centric NBA Blog
NBA WOWY: Find statistics of your favorite team with any arbitrary combination of players on or off the court.


Top
 Profile  
 
 Post subject: Re: Power Ranking
PostPosted: Thu Dec 29, 2011 10:11 pm 
Offline

Joined: Fri Apr 15, 2011 2:05 am
Posts: 95
I've been trying to compile an on-the-fly read of how the markets are rating the teams based on day-to-day pointspreads. An interesting exercise given the volatility of personnel changes and back-to-back challenges. Is it okay if I post that in this thread as a way to compare it to other power ratings? I could make it a separate thread if you guys preferred that.

I've aligned these to line up with the format in the link from the first post...with neutral around 0 and plus or minus ranging from that. To project a pointspread, take the distance between teams then add in 3 points for home court advantage...and 1-2 points for a b2b edge (one team rested, the other's in a b2b) depending on what you feel the travel or team depth challenges might be. I think the market's trying to get a read on that themselves.

+8 Miami
+6 Oklahoma City
+5 Chicago, San Antonio
+3 LA Clippers
+2 Orlando, Boston, New York, Denver, Dallas (with an anvil), LAL with Bynum (estimating)
+1 Portland, Memphis
0 Atlanta, Philadelphia, Indiana
-1 Houston
-2 New Orleans
-3 Milwaukee, Golden State
-4 Minnesota
-5 Phoenix, Charlotte
-6 Sacramento, Detroit
-7 Toronto, Cleveland, New Jersey, Washington, Utah

Tough for me down at the very bottom because so few of the cellar dwellers have lined up against each other. Also, I'm flying blind with guesses on how the market may react this weekend for extreme results from Wednesday Night (Clips played way below market value, Golden State and Charlotte way above for example).

Interested in hearing any needed corrections from those of you who follow the markets. Again, these are attempts to assess how markets are rating the teams rather than my personal assessments of those teams. So...I may think Boston is bettor or worse than +2...but that's kind of a composite of what they're lines were vs. New York-Miami-New Orleans based on where those teams stand above.

To the degree the market is respected as an evaluation tool, this may be a way to keep market ratings in the discussion as the season progresses.

_________________
Blogging basketball at http://www.statintelligence.blogspot.com/


Top
 Profile  
 
 Post subject: Re: Power Ranking
PostPosted: Thu Dec 29, 2011 10:43 pm 
Offline

Joined: Thu Apr 14, 2011 10:41 pm
Posts: 815
Location: Hotlanta
There's a NFL power rating that does regression with HFA for each team. That might be even more useful in basketball. Today I looked at ORTG and DRTG for teams last season at home and away. This is the HCA calculated as the difference between ORTG-DRTG at home and away:

Code:
TEAM   HCA
DEN   11.90
WAS   10.93
GSW   10.33
MEM   10.26
IND   9.27
POR   9.04
SAS   9.02
LAC   8.84
PHI   7.65
TOR   7.62
ORL   7.57
CHA   7.23
CLE   6.96
DET   6.76
MIN   6.70
NJN   6.58
MIL   6.29
CHI   6.25
BOS   6.06
NOH   5.87
UTA   5.53
LAL   5.29
HOU   5.24
OKC   5.17
DAL   4.44
SAC   3.80
PHX   3.59
MIA   3.58
NYK   2.36
ATL   -0.12


Seems to me calculating HCA might make prediction stronger given the variance here.

_________________
The City: A Golden State Warriors-Centric NBA Blog
NBA WOWY: Find statistics of your favorite team with any arbitrary combination of players on or off the court.


Top
 Profile  
 
Display posts from previous:  Sort by  
Post new topic Reply to topic  [ 234 posts ]  Go to page 1, 2, 3, 4, 5 ... 16  Next

All times are UTC


Who is online

Users browsing this forum: Yahoo [Bot] and 3 guests


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

Search for:
Jump to:  
cron
Powered by phpBB® Forum Software © phpBB Group