Home Court Advantage

[Eternal]

Rediscovering no doubt well-known facts. Team home court advantage over and above the baseline home court advantage; higher is better.

Assumed constant across years.

Code: Select all

 team | home  
-----+-------
 UTA  | 1.019
 DEN  | 1.014
 IND  | 1.009
 SAC  | 1.009
 GSW  | 1.008
 LAL  | 1.008
 CLE  | 1.008
 POR  | 1.005
 SAS  | 1.004
 ATL  | 1.002
 ORL  | 1.002
 PHO  | 1.002
 WAS  | 1.002
 CHA  | 1.001
 MIL  | 1.001
 DAL  | 1.001
 NOK  | 1.000
 MEM  | 1.000
 MIA  | 0.999
 BRK  | 0.999
 SEA  | 0.998
 CHI  | 0.998
 LAC  | 0.996
 TOR  | 0.996
 HOU  | 0.996
 NJN  | 0.995
 OKC  | 0.995
 DET  | 0.993
 CHH  | 0.993
 NOH  | 0.992
 MIN  | 0.990
 BOS  | 0.990
 NYK  | 0.988
 PHI  | 0.988

[kjb]

Utah and Denver with the strongest home court advantages. Both cities are at higher altitudes.

From what seasons are these numbers calculated?

[DSMok1]

What is standard home court advantage, and what are the units here?

[Crow]

Several sets of recent findings have needed more labeling and explanation. Would appreciate a fuller guide.

[Eternal]

My model is multiplicative, so roughly:

points scored = home_court_advantage * offense * defense

HCA is modeled separately for offense and defense, depending on whether the offense or defense is home.

HCA itself is composed of two factors, an average factor across all teams, and and individual factor for each team.

HCA = home_court * team_home_court_offense (or team_home_court_defense for defense)

Here home_court is constant across all teams, and the product of all the team_home_court factors equals 1.

The overall impact on a team will be team_home_court_offense/team_home_court_defense.

Here are the separate offensive and defensive HCA values for each team. Above 1.0 is positive for offense (you score more points at home than a typical team) and below 1.0 is positive for defense (you allow fewer points at home than a typical team).

Code: Select all

 team | hca_offense | hca_defense 
-----+-------------+-------------
 ATL  |       1.007 |       1.005
 BOS  |       0.990 |       0.999
 BRK  |       1.000 |       1.001
 CHA  |       0.999 |       0.997
 CHH  |       1.001 |       1.008
 CHI  |       0.991 |       0.993
 CLE  |       1.001 |       0.992
 DAL  |       1.006 |       1.005
 DEN  |       1.011 |       0.997
 DET  |       0.994 |       1.001
 GSW  |       1.014 |       1.006
 HOU  |       0.988 |       0.992
 IND  |       1.003 |       0.994
 LAC  |       0.991 |       0.995
 LAL  |       1.006 |       0.998
 MEM  |       1.003 |       1.003
 MIA  |       0.998 |       0.999
 MIL  |       1.005 |       1.004
 MIN  |       0.994 |       1.004
 NJN  |       0.996 |       1.001
 NOH  |       0.993 |       1.001
 NOK  |       0.998 |       0.998
 NYK  |       0.996 |       1.008
 OKC  |       0.997 |       1.002
 ORL  |       1.000 |       0.998
 PHI  |       0.988 |       1.000
 PHO  |       1.005 |       1.004
 POR  |       1.008 |       1.004
 SAC  |       1.011 |       1.002
 SAS  |       1.002 |       0.998
 SEA  |       0.996 |       0.998
 TOR  |       0.999 |       1.003
 UTA  |       1.005 |       0.987
 WAS  |       1.004 |       1.002

-Chris

[DSMok1]

What is the value of home_court itself?

[Eternal]

Base HCA for all teams is 1.0167 for offense and 0.9835 for defense. So to get overall HCA for offense multiply that value by the team HCA for offense - the same goes for defense.

-Chris

[v-zero]

What is the value of home_court itself?

It should be about 1.035....

[Eternal]

Yep, 1.0167/0.9835 = 1.034.

The Utah Jazz have an enormous advantage over other teams.

-Chris

[DSMok1]

Yep, 1.0167/0.9835 = 1.034.

The Utah Jazz have an enormous advantage over other teams.

-Chris

Do you think the sample size is large enough that these numbers have stabilized?

Utah has large travel distances for basically all opponents, and is at an altitude advantage. Denver should be similar.

[Eternal]

I'm pooling data across 2002-2013; I should have made that clear. Of course, home court advantage could shift from year to year - this is a topic worthy of much more detailed study, but it's clear there's a gigantic impact.

I could easily geocode arena locations and include distance traveled in my model, but I'd need a list of arena locations by team by year.

-Chris

[Crow]

Thanks for the additional explanation.

[Mike G]

Utah has large travel distances for basically all opponents, and is at an altitude advantage. Denver should be similar.

Teams should also have a "road court disadvantage" that could be similarly calculated.
If so, then Utah and Denver may not have an overall advantage. They may be especially sucky on the road.

In Denver vs the Jazz, the Nuggets are 46-32 in regular season games.
In Utah, the Nuggs are 14-65.

[Eternal]

Road disadvantage would be confounded by strength of team offense and defense in the presence of home advantage factors. You'd need to escalate to team strength estimates based on players in order to estimate road disadvantage.

-Chris

[Mike G]

You mean it would be the same as, or indistinguishable from HCA ?