Is Home Court Advantage Real?
Posted: Tue May 09, 2017 1:38 pm
Nathan Sliver here with sevenfortyseven to answer a question I've heard tossed around the league for a few seasons: how much does home court advantage matter? To answer this question, I compiled a new dataset containing every game played in sim league history, including the teams, score, home team, winning percentages of the teams at the time the game was played, attendance, and capacity of the home court. Hopefully the dataset can help answer the question of interest today and produce many articles in the future to pay the Blazers tax bill...
Does home court advantage matter?
Yes, at least in the regular season. First off, it is worth noting that I excluded games played in April because good teams may not be trying to win in much of April. I also exclude November because using winning percentage as a measure of team quality may be inaccurate at that point in the season. I also fail to account for some important things like how well the teams match up and injuries. Running a linear probability model with a binary variable for winning as the dependent variable indicates that playing at home improves a team's probability of winning that game by about 22 percentage points. The estimate is pretty precise with a confidence interval ranging from 20.8 percentage points to 23.3 percentage points. Using a logit model to account for the binary dependent variable produces similar results, though they are not as easily interpreted.
It is also worth noting the estimates for the effect of team strength on winning a game. It is kind of difficult to describe the effect in words so I just made a prediction table so you can get a feel for how much the strength of an opponent relative to the focus team matters.
Does attendance matter?
No. It just doesn't. I tried measuring attendance using both a raw number and a percentage of the arena's capacity interacted with the home indicator and in no case was it significant. It turns out putting butts in the seats does not affect outcomes.
But what about my team?
Nope. Not even you. I also tested to see if any certain team got an additional benefit from being at home or having more fans in attendance and no estimate was significant for any team. Sorry.
What about the playoffs?
Now this is where things get really interesting. Using the same simple model, it seems like it produces similar results (see top table below). Being at home improves the probability of winning by about 22 percentage points, although there is more asymmetry in the effect of focus team winning percentage and opponent winning percentage. When I add the capacity-home interaction things get crazy. Being at home now has an insignificant though suggestively negative result and capacity is positive and significant. Does this mean there is a home court disadvantage in the playoffs? Not really, looking at the magnitude of the estimates, as long as the home team fills over 50% of the house, there is still a positive effect. It does suggest that home court advantage in the playoffs has to be gained by filling the seats and, with a full house, is likely stronger than during the regular season by improving the probability of winning by about 24 percentage points.
Does home court advantage matter?
Yes, at least in the regular season. First off, it is worth noting that I excluded games played in April because good teams may not be trying to win in much of April. I also exclude November because using winning percentage as a measure of team quality may be inaccurate at that point in the season. I also fail to account for some important things like how well the teams match up and injuries. Running a linear probability model with a binary variable for winning as the dependent variable indicates that playing at home improves a team's probability of winning that game by about 22 percentage points. The estimate is pretty precise with a confidence interval ranging from 20.8 percentage points to 23.3 percentage points. Using a logit model to account for the binary dependent variable produces similar results, though they are not as easily interpreted.
- LPM_Home_Results.PNG (7.9 KiB) Viewed 1132 times
- LPM_Home_Predictions.PNG (28.92 KiB) Viewed 1132 times
No. It just doesn't. I tried measuring attendance using both a raw number and a percentage of the arena's capacity interacted with the home indicator and in no case was it significant. It turns out putting butts in the seats does not affect outcomes.
But what about my team?
Nope. Not even you. I also tested to see if any certain team got an additional benefit from being at home or having more fans in attendance and no estimate was significant for any team. Sorry.
What about the playoffs?
Now this is where things get really interesting. Using the same simple model, it seems like it produces similar results (see top table below). Being at home improves the probability of winning by about 22 percentage points, although there is more asymmetry in the effect of focus team winning percentage and opponent winning percentage. When I add the capacity-home interaction things get crazy. Being at home now has an insignificant though suggestively negative result and capacity is positive and significant. Does this mean there is a home court disadvantage in the playoffs? Not really, looking at the magnitude of the estimates, as long as the home team fills over 50% of the house, there is still a positive effect. It does suggest that home court advantage in the playoffs has to be gained by filling the seats and, with a full house, is likely stronger than during the regular season by improving the probability of winning by about 24 percentage points.
- LPM_Home_Playoffs.PNG (17.02 KiB) Viewed 1132 times