Quantifying home court advantage among ACC teams
With all the talk about Duke Basketball’s student section, the Cameron Crazies, losing their luster I sought out a couple of weeks ago to use statistics to prove that Duke has the best home court advantage in the ACC. That I know of, there is no formula out there that can be pointed to to determine this intangible, although many people have tried to attribute some sort of numerical rating to how and why a team performs better at home than they do away from it. This article in particular was striking to me in its thoroughness and unique take on the subject. Written by Duke graduate Thomas H. Giedgowd, he chose the top 20 teams from the 2000-2001 season and examined the difference between home winning percentage and away winning percentage. Through the course of his study, new theories popped up and Giedgowd applied some constraints to the results to adjust for strength of schedule home vs away, game attendance, and size of local alumni base. His findings show a number of conclusions, but most of all that schedule strength home vs away is the most important factor impacting home court advantage.
I took a different approach, and tried to emulate the Four Factors established by Dean Oliver in Basketball On Paper and calculate five stats that would tell a story of one particular ACC program having a defined home court advantage over the other 11 schools. While I didn’t go with Oliver’s numbers (eFG%, TO%, OR%, FT Rate), I picked five that I thought would capture a team’s ability to inflict their will on opponents in a setting they were comfortable with. Honestly, I’d love it if somebody were to do a similar study looking at the Four Factors and I may just do that after this season.
Every study I’ve seen done on this subject considers away games as any game away from their home gym, and the website I used (TeamRankings.com) to gather my data also adheres to this statistical restriction. All of you “neutral court games don’t count as away games” fanboys and girls out there will have to pipe down for now.
Methodology
Using data culled from TeamRankings.com, I evaluated the following statistics for all 12 teams in the ACC (Boston College, Clemson, Duke, Georgia Tech, Maryland, North Carolina, North Carolina State, Florida State, Virginia, Virginia Tech, Miami, Wake Forest).
- Home Win Percentage vs Away Win Percentage
- Home Points Per Game vs Away Points Per Game
- Home Defensive Points Per Game vs Away Defensive Points Per Game
- Home Free Throw Percentage vs Away Free Throw Percentage
- Home Defensive Free Throw Percentage vs Away Defensive Free Throw Percentage
At first glance, these stats are ones that I wouldn’t even begin to use to tell a story of how a basketball team was performing. Stats guys like John Gasaway and Ken Pomeroy have done a good job of explaining why points per game is a useless stat and defensive free throw percentage is a number that really can’t be affected by the team not shooting the ball (barring a player or coach screaming while the other team is shooting). However, we’re not trying quantify what team has been better in a particular season or how a team is likely to perform in the NCAA Tournament. This study is only attempting to quantify home court advantage in a relatively meaningful way.
In this study, I think it is ok to use points per game because that stat alone does not tell any story at all in relation to home court advantage. What does tell a story with PPG is the differential between home and away. Same goes for defensive points per game and winning percentage. When looking at free throw percentage and defensive free throw percentage, one would hypothesize that teams would shoot a better number at home, where their fans are quiet in support than on the road, with any number of distractions. The numbers will tell the story.
Win Percentage Differential
This is the stat that I figured would tell the greatest story going into this study. Nearly every other researcher, including Giedgowd, looked to win percentage differential as the holy grail number to show that a team performs better in friendly confines than they do away.
Just looking at the differential, one would think that Virginia performs so much better at home than they do away that you could consider John Paul Jones Arena (or University Hall prior to 2006) the toughest place to play in the conference. However, looking at the numbers for away and home win percentage, this differential is more likely attributed to the Cavaliers’ horrible road play during that time (a conference low 26.93%) compared to a somewhat more respectable home win percentage (3rd lowest in ACC at 72.10%). Whether this can be attributed to a strong home court environment is a matter of debate. University Hall was always known as a tough place to get a win and John Paul Jones Arena is an inhospitable place to play for road teams, especially now that Tony Bennett is building a strong program there.
While differential between home and away percentages is likely to tell the greatest story when attempting to quantify home court advantage, I don’t believe that teams should be punished for consistency (both in winning and in losing). In revision of these numbers, perhaps some weight should be attributed to teams that were so much better (or worse) than the other teams in the league. I’m not a statistician, so I don’t know how to do this. If you have any ideas, please drop comment down below this post and let me know.
I know this is a Duke site, so I’ll probably get accused of some homerism here, but it needs to be pointed out just how good Duke has been in relation to the rest of the conference over this ten year period. Duke’s home win percentage of 94.72% is a full 12 points better than the next best team (UNC at 82.40%) and 14 points better than the 3rd place team (Wake Forest at 80.05%). Even more ridiculous is that the 94.72% number is nearly 31 points higher than the conference average of 63.80% over this time period. The Blue Devils’ away win percentage of 71.14% is even more striking in its separation from the pack as the conference average sits at 36.68%. Away from Cameron, Duke is over 34 percentage points better than the ACC average over the last ten years.
Points Per Game Differential
This stat is one that will likely be disregarded by many statisticians, but I felt by looking at data from a ten-year period this study could show teams that tend to play better offensively at home versus on the road. Although there is certainly research that shows that teams tend to play weaker teams at home, I hoped that this number would prove an ability to force one’s will upon opponents in the friendlier confines of their home courts rather than being at the mercy of potential influences from travel, road crowds, and unfamiliar environments. When I originally set out to begin this study, this was the only offensive stat I measured. After looking at the results, I decided to go back and chart effective field goal percentage. I may come back around and do the same for offensive efficiency, as that would tell a truer picture of the differential between home and away. Look for a further study to do just that. On to the results.
Looking at the numbers, UNC has the greatest points per game differential between home and away games of any team in the conference over this ten year period. At a difference of 9.3 PPG, the Heels scored an average of 85.41 points in their games at the Dean Dome and 76.94 points in their games away from Chapel Hill. UVA comes up just short at 8.3 PPG Diff while Maryland is the only other team over 8, coming in at 8.13 PPG Diff.
What does this tell us? A number of things, but clearly UNC is able to impose their high-paced tempo on opponents much easier at home (9.3 PPG is a lot of points in a college basketball game) than they can when they take their show on the road. Maryland’s Comcast Center is well-documented as a tough place to play due to their rowdy fanbase and the fact that Gary Williams always put strong teams on the court.
Here’s another factoid that might not jump right out at you though. These stats not only include true road games, but also games away from each teams’ home court. That means these numbers also include the ACC and NCAA Tournament. In the opening rounds of the tournament, UNC likely faced inferior teams where they were able to run up the score, thus inflating their numbers. This makes what the Tar Heels have been able to do on their home court scoring-wise even more impressive.
While the Duke Blue Devils come in at a 6.84 PPG Diff, they should again be given high praise for the fact that they sport the league’s second-highest away points per game mark at 76.79, trailing only UNC at 76.94. (You’ll be able to see all teams’ numbers in subsequent posts. Below is a look at Home and Away Points Per Game over the last ten years to get a better idea of how each team has done.
Let’s look at defensive points per game differential and see if we come up with some different results.
Sporting the conference’s 3rd best defensive points per game margin at 65.2, the Georgia Tech Yellow Jackets come in nearly two points ahead of Florida State in defensive points per game differential. They were 10.14 points better defensively at the Thriller Dome than they did on the road. While Paul Hewitt’s teams always had athletic players, his teams do not immediately strike me as ones that were able to enforce their will on other squads. The atmosphere in the dome however was always intimidating and unfortunately is not so in their temporary home, Philips Arena.
Florida State under Leonard Hamilton has always been known as a defense first ballclub, so it is a bit surprising to me that their margin would be this high. I figured that their consistency would deflate these numbers but there is clearly something to playing as the Donald Tucker Center as they kept their opponents to a league-low 63.13 points per game from 2001-2011. Their margin was second best at 8.32 PPG Diff.
While FSU has the reputation as being tough without the ball, Duke came in only .1 points per game less effective defensively over this ten year period. The reason the Blue Devils suffered in margin is because they were the only team in the league to hold their opponents under 70 points per game on the road. That 68.82 mark was over one point better than the next best team defensively on the road, Boston College. Here’s a look at all the teams’ performances over the last ten years:
The next stat I wanted to take a look at was free throw differential, both offensive and defensive. What is free throw defense you ask? It’s Speedo Guy, it’s Wild Bill, it’s the Rock Chalk Flash Mob. Free throw defense is the one time when college basketball fans have an impact on the performance of the opposing team while the players have none.
Some very interesting numbers produced on this graph. Most striking is Georgia Tech’s negative FT% Diff at home compared to their performance away. While they have been good at keeping teams from scoring points at home as shown above, the Yellow Jackets for one reason or another shoot nearly 2% worse at home than they do away. The Thriller Dome is not a good place to shoot free throws apparently. They also hold the league’s lowest defensive free throw percentage at 66.51%. Miami, Clemson, and UNC also shoot worse at home than they do on the road. Maybe these fans need to yell like the fans are on the road to make the players feel more comfortable instead of being silent?
Not surprising considering the above Speedo Guy video, but Duke holds an extremely high defensive free throw rate from 2001-2011. They are second only to Boston College, whose opponents shoot a little over 3% points worse away from Conte Arena. Duke shoots free throws nearly the same (75.2%) in Cameron Indoor as they do on the road (72.21%).
Wake Forest has the largest margin between their home free throw percentage (71.32%) compared to their away free throw percentage (67.52%). This differential of 3.80% points is nearly 2% points greater than the next team (Virginia Tech at 1.68%). Most teams measured are consistent when shooting at home compared to away. All teams’ numbers are below.
So what does this all mean? As I said above, I’m no statistician but just looking at the data, we can see that no one team has dominated any of the differential statistics. If I were to draw any conclusions from this study for a scientific research paper, I would not be able to come to any significant realization based on the data collected. Just using these metrics and using the numbers that each team ranked in relation to their other league members, we can come to a number by adding up where all teams stand in these four categories. Obviously, the lower the number the better. Let’s start there and then hopefully my readers or statisticians out there who happen to be reading can input their thoughts below.
Win % Differential
- Virginia 45.17%
- Wake Forest 39.43%
- Georgia Tech 37.43%
- Clemson 37.15%
- FSU 37.04%
- Virginia Tech 36.66%
- NC State 32.97%
- Maryland 32.30%
- Miami 28.07%
- Duke 23.58%
- UNC 22.97%
- Boston College 21.64%
Points Per Game Differential
- UNC 8.47
- Virginia 8.3
- Maryland 8.13
- Wake Forest 7.08
- Duke 6.81
- NC State 6.67
- Virginia Tech 6.63
- Miami 6.61
- Florida State 5.16
- Clemson 5.12
- Georgia Tech 4.11
- Boston College 3.82
Defensive Points Per Game Differential
- Georgia Tech 10.14
- Florida State 8.32
- Maryland 7.37
- Wake Forest 7.1
- Virginia 6.66
- Clemson 6.39
- NC State 6.34
- Duke 5.48
- Virginia Tech 5.3
- Miami 4.17
- UNC 3.57
- Boston College 2.64
Free Throw % Differential
- Wake Forest 3.80%
- Virginia Tech 1.68%
- Florida State 1.12%
- NC State .81%
- Maryland .65%
- Boston College .45%
- Virginia .32%
- Duke .31%
- UNC -.45%
- Clemson -.67%
- Miami -1.03%
- Georgia Tech -1.74%
Defensive Free Throw Differential
- Boston College 3.61%
- Duke 3.3%
- Georgia Tech 2.87%
- UNC 2.7%
- Maryland 2.53%
- Virginia 2.41%
- Wake Forest 1.1%
- Florida State .71%
- Clemson .55%
- Miami .39%
- NC State -.2%
- Virginia Tech -.27%
Cumulative Rankings
- Wake Forest 18
- Virginia 21
- Maryland 24
- FSU 27
- Georgia Tech 30
- Duke 33
- NCSU 35
- VA Tech 36
- UNC 36
- Miami 37
- Clemson 39
- BC 43
So there you have it. According to these numbers, Wake Forest had the best home court advantage of all ACC teams from 2001-2011. Boston College had the worst. Obviously, these numbers are not definitive in any way, just one take on the subject.
I will have a follow up post soon with the numbers for each team over this period. Please let me know if you have any thoughts or questions in the comments section below.
A big thank you as well to Brian Cox for his fabulous work on the excel graphs and charts.
I like the idea but it seems like a stretch to say that Wake had the best home court advantage based on the data points chosen. It seem that Duke has the better home court advantage because they win more often on their home court, followed by UNC. Perhaps a better conclusion would be to state that second and third tier teams depend more on playing at home because of an inability to win on the road. Great work, though.
There is something called the odds ratio that is used specifically to prove your claim but I don’t really understand it completely and didn’t want to talk out of my a$$ more than I already was.
I had the same thought as you though. Duke’s consistency and excellence is really hurting them using differential and they shouldn’t be penalized for it, but they are here.
Seems to me, the problem with using differential is it “penalizes” a team who performs well on the road. Duke won 71% of their road games during your study. If they won every single home game (best they could do) their win% differential would be 9th, barely ahead of the 10th place team. Same with all the other categories. If I’m reading the numbers properly, Duke is best or 2nd best in every category you charted except away defensive ft%, and not coincidentally, defensive ft% is the only differential in which Duke scores well.
Put another way, since your higher number is capped at 100%, using a differential skews your results against good teams.
Kedsy, that was a revelation I came to as I got further into the research. I mentioned it earlier in the post. I wish there were a way to reward teams for consistency while still factoring in differential but I’m not enough of a statistician to do so. Hopefully, one of the readers of this post is.
Did you think about using standard deviation or mean difference? What it seems you need is some independent anchor to deal with the “consistency” issue that the above comments have pointed out. Perhaps establishing the average win percentage at home and then measuring how much higher or lower each team is on that level. Then do the same thing with the road win percentage. Then measure the deviation between those two differences. That might be a good anchor.
Also, I think rooting the points per game stat with an adjustment for pace of the game would still be more telling, because ppg can easily be effected by opponent’s style of place, particularly with an unbalanced conference schedule where not everyone has a home and home match up.
But the biggest problem may be sample size. With the change in talent level among the teams, it may be best to use 10 years worth of stats to really get an idea about this. I’m not sure what time period these stats are from (I probably missed it in the article) but the changes in talent level that can also affect alot of these stats, and not only the level of talent on the team being measured, but also their opponent’s level of talent.
But, yeomen’s work here. Statistical analysis is a big, complex problem.
I’m not sure it’s much of a “penalty” for teams who also win a good percentage of games on the road. A good team is going to win games regardless of where the game is played, meaning the location of the game has little impact on the team being analyzed. If a team struggles on the road but wins at home, isn’t that an indication that the location has a larger impact on the outcome of the game, which speaks to “home court advantage”?