APBRmetrics

The discussion of the analysis of basketball through objective evidence, especially basketball statistics.
It is currently Mon Dec 22, 2014 6:20 pm

All times are UTC




Post new topic Reply to topic  [ 8 posts ] 
Author Message
PostPosted: Mon Apr 09, 2012 3:20 pm 
Offline

Joined: Tue Mar 27, 2012 12:54 am
Posts: 7
The recent play of the 76ers has highlighted and potentially significant problem with SRS and other power ratings. Most methods look at a team's entire record to calculate all the power ratings. However, that has a potential problem in that assumes a team was playing at a similar level at all points in the season. A team that faced the 76ers at the start of the season when Hawes was playing great and the team was playing around +8 basketball is being considered the same as a team that is facing the 76ers now while they are in a death spiral.


Top
 Profile  
 
PostPosted: Mon Apr 09, 2012 4:15 pm 
Offline

Joined: Thu Apr 21, 2011 8:25 pm
Posts: 190
Location: Boone, NC
The Sixers are 5th in SRS, at +4.1.

They are tied for the 7th-best record. The difference there isn't very large.

If you count their 16 point loss at Orlando on Feb 15th as the start of their "death spiral", (20 losses and 9 wins) their point margin has been -0.92.

Their 10-game average has fluctuated from 18.1 (Jan 16) to -2.5 (Feb 22) back to 7.6 (March 11) and now down to -6.2.

Recency is somewhat important, but I think rationally they will probably still average out in their remaining games to +4 ish.


Top
 Profile  
 
PostPosted: Mon Apr 09, 2012 4:17 pm 
Offline

Joined: Mon Apr 18, 2011 10:09 am
Posts: 405
As far as I know a recency bias does not improve the predictive power.

The 76ers are a weird example, because they are playing really different against different strong opponents. They are pretty awesome against bad teams while being closer to the average against better teams. I can compensate for that by using a different way of adjusting for the strength of the opponents in my Power Ranking, that's why the 76ers are listed clearly lower (and never were among the Top3 during any point of the season). So, in the end it is not so much about a different point of the season as much more a problem of exactly determining the playing level, if a team like 76ers is constantly blowing out the bad teams while doing nothing really impressive against the good teams.

_________________
http://bbmetrics.wordpress.com/


Top
 Profile  
 
PostPosted: Mon Apr 09, 2012 4:49 pm 
Offline

Joined: Tue Mar 27, 2012 12:54 am
Posts: 7
This is not a recency issue. If it was just a recency issue, I could look at the last 20 or 30 games for each team and compare that rating to the seasonal average. Then I could do a subjective analysis of each team to see if I account for any significant changes and determine what was most likely going forward. That's the way I tend to handle these things now. I look for coaching changes, key injuries, trades, FA signings etc...

What I am worried about here is that the 76ers started out as an approximately +8 team and seem to be fading fast and furiously. So IMO the average does not reflect the difference in their level of performance from start to finish. Perhaps an adjustment for the extremes of performance against good vs. bad teams (their standard deviation of performance was the highest in the league last time I looked) would tend to smooth things out a little, but I think they are legitimately just playing a lot worse now. So the teams that played them early in the season are probably being underrated a hair and the teams that are playing them now are probably being overrated a hair.

This is the most extreme case, but there are other examples.

The Suns are playing better now than at the start of the season.

The Knicks are playing much better now (though it's easier to account for their change so I didn't bring it up).

Maybe when you net all these things out it doesn't make much difference, but intuitively it seems there is some potential for problems if you don't account for fluctuations in a team's performance from month to month (especially when you can account for it)


Top
 Profile  
 
PostPosted: Tue Apr 10, 2012 9:02 pm 
Offline

Joined: Thu Apr 21, 2011 8:25 pm
Posts: 190
Location: Boone, NC
@Statmandu

When you say they are playing terrible "now" -- what is your definition of "now" ??
Anyone can have a crap night, crap week, crap month.


Top
 Profile  
 
PostPosted: Tue Apr 10, 2012 9:20 pm 
Offline

Joined: Thu Apr 14, 2011 10:41 pm
Posts: 817
Location: Hotlanta
My Vegas ratings (or any Vegas ratings) definitely change over time. GSW is rated about 2 points lower since the trade deadline than the full season.

Assuming Vegas knows what they are doing, that probably means it has more predictive power to use the more recent rating.

_________________
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  
 
PostPosted: Wed Apr 11, 2012 7:14 am 
Offline

Joined: Mon Apr 18, 2011 10:09 am
Posts: 405
EvanZ wrote:
My Vegas ratings (or any Vegas ratings) definitely change over time. GSW is rated about 2 points lower since the trade deadline than the full season.

Assuming Vegas knows what they are doing, that probably means it has more predictive power to use the more recent rating.


Not quite sure what you are getting at, but I think your Vegas Rating has all games from the season included without any further recency bias. It should be obvious that the latest rating is better as predictive tool. Or do you value later games more than earlier games (which would be a recency bias).

_________________
http://bbmetrics.wordpress.com/


Top
 Profile  
 
PostPosted: Wed Apr 11, 2012 10:45 am 
Offline

Joined: Thu Apr 14, 2011 10:41 pm
Posts: 817
Location: Hotlanta
I'm talking about doing regression on only games since the trade deadline vs. the entire season (or some other subset of games). The ratings based on more recent games are in some cases significantly different from the full-season ratings. My assumption is the former are more predictive.

_________________
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  [ 8 posts ] 

All times are UTC


Who is online

Users browsing this forum: Google [Bot] and 4 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:  
Powered by phpBB® Forum Software © phpBB Group