APBRmetrics

The discussion of the analysis of basketball through objective evidence, especially basketball statistics.
It is currently Wed Jul 27, 2016 11:34 am

All times are UTC




Post new topic Reply to topic  [ 13 posts ] 
Author Message
PostPosted: Wed Jul 30, 2014 1:00 pm 
Offline

Joined: Fri Apr 15, 2011 8:28 am
Posts: 780
With the goal to create a good team power ranking without using player specific information I, once again, looked at the following effects on team performance:
- effect of rest
- effect of location of the last game (when there's 1 or less days of rest between the last game and this game)
- effect of the last game having been an OT game (and you're on a B2B)
- team specific home hourt advantage

I'm using regular season data from '02 in one large (Ridge) Regression, so that I have a decent sample size to estimate the effects. (there's one variable per season and team, though). Game outcome is adjusted for pace. Ridge penalty, found through crossvalidation, was 7.5

First, team specific home court effects
Code:
╔═════════════════════════════════════╦═══════════╗
║                Team                 ║ Extra HCA ║
╠═════════════════════════════════════╬═══════════╣
║ Utah Jazz                           ║ 2.4       ║
║ Denver Nuggets                      ║ 2.1       ║
║ Charlotte Bobcats                   ║ 2         ║
║ Cleveland Cavaliers                 ║ 1.7       ║
║ Indiana Pacers                      ║ 1.7       ║
║ Washington Wizards                  ║ 1.5       ║
║ Sacramento Kings                    ║ 1.4       ║
║ New Orleans Pelicans                ║ 1.1       ║
║ Golden State Warriors               ║ 0.9       ║
║ New Jersey Nets                     ║ 0.6       ║
║ Portland Trail Blazers              ║ 0.5       ║
║ Los Angeles Clippers                ║ 0.5       ║
║ New Orleans/Oklahoma City Hornets   ║ 0.5       ║
║ Atlanta Hawks                       ║ 0.4       ║
║ Chicago Bulls                       ║ 0.2       ║
║ Los Angeles Lakers                  ║ 0.1       ║
║ Toronto Raptors                     ║ 0         ║
║ Seattle SuperSonics                 ║ -0        ║
║ Milwaukee Bucks                     ║ 0         ║
║ Minnesota Timberwolves              ║ -0.3      ║
║ Orlando Magic                       ║ -0.3      ║
║ Memphis Grizzlies                   ║ -0.4      ║
║ Phoenix Suns                        ║ -0.5      ║
║ Houston Rockets                     ║ -0.7      ║
║ Miami Heat                          ║ -0.7      ║
║ New York Knicks                     ║ -0.8      ║
║ New Orleans Hornets                 ║ -0.8      ║
║ San Antonio Spurs                   ║ -1        ║
║ Boston Celtics                      ║ -1.3      ║
║ Detroit Pistons                     ║ -1.3      ║
║ Dallas Mavericks                    ║ -1.3      ║
║ Philadelphia 76ers                  ║ -1.4      ║
║ Oklahoma City Thunder               ║ -1.5      ║
║ Brooklyn Nets                       ║ -2.2      ║
║ Charlotte Hornets                   ║ -3.1      ║
╚═════════════════════════════════════╩═══════════╝
Having a high/low number here is neither good nor bad. A higher value just means you're better at home than the average home team, but you're worse at away games, and vice versa. For example, with average HCA being 3.2, an average Jazz team against an average team X would be favored by 3.2+2.4 = 5.6 in Utah, but team X would also be favored by 5.6 when playing at X.

and the effects of rest/ot/location
Code:
╔══════╦════════════════╦═══════════════╦═══════════════╦═════════════════════╗
║ Rest ║ OT (last game) ║ last location ║ this location ║ Effect for awayteam ║
╠══════╬════════════════╬═══════════════╬═══════════════╬═════════════════════╣
║ 1d   ║                ║ home          ║ away          ║ 1.3                 ║
║ 1d   ║                ║ away          ║ away          ║ 1.3                 ║
║ b2b  ║                ║ away          ║ away          ║ -0.2                ║
║ b2b  ║                ║ home          ║ away          ║ -1                  ║
║ b2b  ║ OT             ║ home          ║ away          ║ -1.2                ║
║ b2b  ║ OT             ║ away          ║ away          ║ -2.1                ║
╚══════╩════════════════╩═══════════════╩═══════════════╩═════════════════════╝
╔══════╦════════════════╦═══════════════╦═══════════════╦═════════════════════╗
║ Rest ║ OT (last game) ║ last location ║ this location ║ Effect for hometeam ║
╠══════╬════════════════╬═══════════════╬═══════════════╬═════════════════════╣
║ 1d   ║                ║ home          ║ home          ║ 1.7                 ║
║ 1d   ║                ║ away          ║ home          ║ 1.5                 ║
║ b2b  ║                ║ home          ║ home          ║ 0.8                 ║
║ b2b  ║                ║ away          ║ home          ║ 0.1                 ║
║ b2b  ║ OT             ║ away          ║ home          ║ -0.8                ║
║ b2b  ║ OT             ║ home          ║ home          ║ -3.9*               ║
╚══════╩════════════════╩═══════════════╩═══════════════╩═════════════════════╝
╔══════╦════════════════╦═══════════════╦═══════════════╦════════╗
║ Rest ║ OT (last game) ║ last location ║ this location ║ Effect ║
╠══════╬════════════════╬═══════════════╬═══════════════╬════════╣
║ 2d   ║                ║               ║               ║ 1.2    ║
║ 3d+  ║                ║               ║               ║ 1.2    ║
╚══════╩════════════════╩═══════════════╩═══════════════╩════════╝
(*) denotes small sample size.
According to this, teams are playing best when having had 1 day of rest, their last game was at home and this game is also at home. Having more than 1 day of rest leads to worse performance than having just 1 day of rest. Having to play B2Bs is (obviously) not good for team performance, and being on a B2B with the first game going into OT is obviously not good, either

It's important to note that this analysis was not done on the lineup level, but on the team level. Teams might also be playing worse in B2Bs because they give less minutes to their key players, not just because their players are underperforming due to exhaustion

And, just for fun, here are the top and bottom 10 regular season teams since '02 (using all the aforementioned adjustments)
Code:
╔═════════════════════════╦══════╦════════╗
║          Team           ║ Year ║ Rating ║
╠═════════════════════════╬══════╬════════╣
║ Cleveland Cavaliers     ║ 2009 ║ 9.1    ║
║ Boston Celtics          ║ 2008 ║ 8.6    ║
║ Sacramento Kings        ║ 2002 ║ 7.6    ║
║ Oklahoma City Thunder   ║ 2012 ║ 7.6    ║
║ Oklahoma City Thunder   ║ 2013 ║ 7.6    ║
║ San Antonio Spurs       ║ 2005 ║ 7.3    ║
║ San Antonio Spurs       ║ 2007 ║ 7.3    ║
║ Utah Jazz               ║ 2008 ║ 7.2    ║
║ San Antonio Spurs       ║ 2004 ║ 7.1    ║
║ Los Angeles Lakers      ║ 2002 ║ 6.9    ║
║ ..                      ║      ║        ║
║ Los Angeles Clippers    ║ 2009 ║ -7.8   ║
║ Milwaukee Bucks         ║ 2014 ║ -7.8   ║
║ New Jersey Nets         ║ 2010 ║ -7.9   ║
║ Minnesota Timberwolves  ║ 2010 ║ -8.1   ║
║ Chicago Bulls           ║ 2002 ║ -8.2   ║
║ Cleveland Cavaliers     ║ 2003 ║ -8.3   ║
║ Miami Heat              ║ 2008 ║ -8.5   ║
║ Portland Trail Blazers  ║ 2006 ║ -8.6   ║
║ Atlanta Hawks           ║ 2005 ║ -8.9   ║
║ Philadelphia 76ers      ║ 2014 ║ -9.9   ║
╚═════════════════════════╩══════╩════════╝

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


Top
 Profile  
 
PostPosted: Wed Jul 30, 2014 2:06 pm 
Offline

Joined: Fri Aug 31, 2012 1:52 am
Posts: 293
Does RAPM and xRAPM take into account HCA?

If you run an RAPM using those adjustments, I'm not sure it would right to use those HCA values. How much of it is simply randomness rather than the team having a good HCA. Both Utah and Denver definitely have a legitimate HCA due to distance and/or altitude. I'm fine with an adjustment for that. But, I don't think Road players should get an extra 2 point adjustment because they played in Cleveland or Charlotte.


In your final list of worst teams, where is the 7-59 Bobcats at. Are those adjustments really enough to make them not one of the worst teams since 2002?


Top
 Profile  
 
PostPosted: Wed Jul 30, 2014 3:17 pm 
Offline

Joined: Thu Apr 03, 2014 4:40 pm
Posts: 2
Team Specific Home advantage should mainly be an effect of Altitude and Location.
Having Charlotte at +2 and -2.2 looks a bit strange, and shows you are probably lacking a bit of data on these teams ?


Top
 Profile  
 
PostPosted: Wed Jul 30, 2014 3:48 pm 
Offline

Joined: Fri Apr 15, 2011 8:28 am
Posts: 780
Fablepongiste wrote:
Team Specific Home advantage should mainly be an effect of Altitude and Location.
Having Charlotte at +2 and -2.2 looks a bit strange, and shows you are probably lacking a bit of data on these teams ?
Yes. Lacking data for the Charlotte Hornets (just one season)

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


Top
 Profile  
 
PostPosted: Thu Jul 31, 2014 11:41 am 
Offline

Joined: Fri Apr 15, 2011 8:28 am
Posts: 780
Looked into 4-in-5 and 3-in-4. Here, I chose to count only those games as 3-in-4 that were not also 4-in-5, and where game 3 and 4 were B2B (so X0XX, but not XX0X)

It seems we need to subtract an additional point from team strength when the team is playing its' 4th game in 5 nights. However, playing a 3rd game in 4 nights (with G2 and G3 being B2B) doesn't seem to have an additional negative effect (when already accounting for the fact that you're on a B2B)

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


Last edited by J.E. on Thu Jul 31, 2014 12:35 pm, edited 1 time in total.

Top
 Profile  
 
PostPosted: Thu Jul 31, 2014 12:26 pm 
Offline

Joined: Thu Apr 14, 2011 11:18 pm
Posts: 776
Location: Maine
J.E. wrote:
Looked into 4-in-5 and 3-in-4. Here, I chose to count only those games as 3-in-4 that were not also 4-in-5, and where game 3 and 4 were B2B (so X0XX, but not XX0X)

It seems we need to subtract an additional point from team strength when the team is playing its' 4th game in 5 nights. However, playing a 3rd game in 4 nights (with G3 and G4 being B2B) doesn't seem to have an additional negative effect (when already accounting for the fact that you're on a B2B)


Nice! Did you also look at "first game of year" and "all star break" effects?

Also, have you broken the results into offense, defense, and perhaps pace?

_________________
APBRmetrics Forum Administrator
GodismyJudgeOK.com/DStats/
Twitter.com/DSMok1


Top
 Profile  
 
PostPosted: Thu Jul 31, 2014 1:05 pm 
Offline

Joined: Fri Apr 15, 2011 8:28 am
Posts: 780
DSMok1 wrote:
Nice! Did you also look at "first game of year" and "all star break" effects?
Not yet (and probably won't)
Quote:
Also, have you broken the results into offense, defense, and perhaps pace?
I'm primarily interested in effect on overall strength and don't care too much about where (offense vs. defense) the effect is coming from. Also, the dataset I originally chose to do this with has no info beyond final score for each team (and date/OT)

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


Top
 Profile  
 
PostPosted: Mon Aug 04, 2014 8:35 am 
Offline

Joined: Fri Apr 15, 2011 8:28 am
Posts: 780
Accidentally used the same data for '12 and '13.
Yes, the '12 Bobcats were the worst team since '02 with a -12.6 rating (expecting a slight regression to the mean had the season continued)
The '12 Bulls are ranked as the 8th best team since '02

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


Top
 Profile  
 
PostPosted: Mon Aug 04, 2014 10:45 am 
Offline

Joined: Mon Apr 18, 2011 10:09 am
Posts: 458
J.E. wrote:
Also, the dataset I originally chose to do this with has no info beyond final score for each team (and date/OT)


Just a quick question in regard to the interpretation: The results are points per game differences not per 100 possession then?

_________________
http://bbmetrics.wordpress.com/


Top
 Profile  
 
PostPosted: Mon Aug 04, 2014 5:15 pm 
Offline

Joined: Fri Apr 15, 2011 8:28 am
Posts: 780
mystic wrote:
Just a quick question in regard to the interpretation: The results are points per game differences not per 100 possession then?
Good question. The results are expected score difference after normalizing it by (home_score+away_score)/195.

For forecasting, if you have reason to believe that the total score for the game in question will be different from 195 you'll have to multiply the expected score difference (given by the power rating) with the above adjustment

Maybe I'll do a power rating that splits each teams' rating into Points_for and Points_against. I don't even need pace data for that

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


Top
 Profile  
 
PostPosted: Tue Aug 05, 2014 1:56 pm 
Offline

Joined: Mon Apr 18, 2011 10:09 am
Posts: 458
Thanks for the response and giving such great information on that matter in this thread!

_________________
http://bbmetrics.wordpress.com/


Top
 Profile  
 
PostPosted: Mon Dec 15, 2014 8:48 pm 
Offline

Joined: Tue Nov 27, 2012 7:04 pm
Posts: 308
@J.E.

Thank you so much for the numbers. Are these numbers up to date with your possible new findings?


Top
 Profile  
 
PostPosted: Mon Dec 15, 2014 9:16 pm 
Offline

Joined: Thu Oct 23, 2014 9:59 pm
Posts: 63
Location: Las Vegas
I looked at the impact of rest days on team performance against the spread and totals, and came to similar conclusions as JE, specifically the "teams are playing best when having had 1 day of rest, their last game was at home and this game is also at home. Having more than 1 day of rest leads to worse performance than having just 1 day of rest.".

Full Analysis here:
http://bigleagueinsights.com/impact-res ... ad-totals/


Top
 Profile  
 
Display posts from previous:  Sort by  
Post new topic Reply to topic  [ 13 posts ] 

All times are UTC


Who is online

Users browsing this forum: Bing [Bot], Google [Bot] and 5 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