By Johnattan Ontiveros
In mid-March, the NBA season halted amidst concerns of what was to be a global pandemic. After reorienting itself for a couple of months, however, the NBA took precautions and formed a ‘bubble’ of players and staff at Disneyworld to continue the season’s playoffs. This season, games are being played in home stadiums, but one major factor of what makes a home stadium a ‘home’ is missing: the fans.
How much has the traditional ‘home advantage’ that teams gain by playing in their own stadium changed during this 2020/2021 season? Now I know there are multiple factors that make up a home advantage such as not having to travel, but many of those factors are still present in this strange season except for the drastic reduction of fans at games.
To explore this change in home advantage, I used a normal linear model to fit upon the strength of each team and the home advantage of that season. The way this works is that each game of the season is written as:
PointsHome – PointsAway = StrengthHome – StrengthAway + HomeAdvantage
The model then uses the entire season’s data to find an estimate for the strength of each of the NBA’s 30 teams and of Home Advantage. Since the model uses the difference in scores, an estimate of (HomeAdvantage = 2) directly translates to an estimated 2 point increase in a home team’s score.
I used data from the 2017/18 – 2020/21 NBA seasons to get a home advantage estimate for each season. Seasons 2017/18 up to March of the 2019/20 serve to represent the home advantage in a standard season with fans. The second half of the 2019/20 season in the bubble represents an interesting case in and of itself since neither team has the home advantage at Disneyworld. Lastly, the current 2020/21 season represents the games with home advantage without fans. The estimates of home advantage and standard deviation are graphed and tabled below.
From these estimates, we get some interesting findings. In regular seasons, the estimated home advantage seems to float around 2.5 points per game. All of these estimates were statistically significant on the 0.05 level, meaning that these estimates are likely not due to a chance outcome of games. When in the bubble, the home advantage estimate was at 1.34, but the much higher amount of error in this estimate and statistical insignificance indicates that this estimate is not reliable. This means that our estimated home advantage for the second half of the 2019/20 season is not significantly different from the null case: no home advantage (HomeAdvantage = 0). This is expected due to the much smaller sample size of this mini-playoff.
In this current season (up to March 1st), the estimated home advantage is 0.88 points, with a smaller standard error than the Bubble games, but is still not significantly different from 0. The data on this season is much larger than in the bubble (502 games vs 168) and will grow as the season progresses. As of now, it seems that the lack of fans at games is tied to a non-significant home advantage at NBA games. The home advantage just doesn’t seem to have as strong and clear of an effect as it does in past seasons, but may change at the end of the season when we have a larger sample size.